---
title: "M&A and Fintech Glossary"
url: "https://mahouse.world/glossary"
type: "index"
---

# M&A and Fintech Glossary

434 terms of M&A, fintech, payments, licensing and AI, each with key facts, how it works in practice, a worked example and common mistakes. Each entry below links to the term as Markdown.

- [3-D Secure](https://mahouse.world/glossary/3-d-secure.md): 3-D Secure is the card schemes' protocol for authenticating a cardholder during an online payment: the merchant's request passes to the issuer, which verifies the cardholder — once by a password on a pop-up, now by a banking-app prompt — and returns the result with the authorisation. Success shifts fraud liability to the issuer. The first version cost merchants a fifth of their conversions; version 2 rebuilt it.
- [3-D Secure 2](https://mahouse.world/glossary/3-d-secure-2.md): 3-D Secure 2 is EMVCo's rebuilt authentication protocol for card-not-present payments: the merchant sends the issuer rich transaction and device data, the issuer authenticates most transactions without a challenge, challenges the rest in the banking app inside the merchant's page, and the SCA exemptions ride in the messages. It replaced version 1 in 2022 and made strong customer authentication in Europe workable.
- [AI Accelerator](https://mahouse.world/glossary/ai-accelerator.md): An AI accelerator is a processor designed for the arithmetic neural networks are made of — vast numbers of multiply-and-add operations on matrices, run in parallel at low precision — which it performs tens or hundreds of times faster per watt than a general-purpose CPU. Nvidia's GPUs dominate; Google's TPUs, AMD and the hyperscalers' custom chips compete; access to them is the constraint on who can build large models.
- [AI Agent](https://mahouse.world/glossary/ai-agent.md): An AI agent is a program built on a foundation model that pursues a goal over several steps — reading, deciding, calling tools such as a database, an API or a payment system, checking the result and continuing — rather than answering one prompt with one reply. In fintech it is the form AI takes when it is allowed to do things, not only to say them.
- [AI Alignment](https://mahouse.world/glossary/ai-alignment.md): AI alignment is the work of making an AI system's objectives and behaviour match what its makers and users actually intend — not merely what its training objective rewarded — through reinforcement learning from human feedback, rule-based training, red-teaming and evaluation against the intended outcome. For a fintech it is whether the credit model, the collections agent or the support bot optimises what the business meant.
- [AI Bias](https://mahouse.world/glossary/ai-bias.md): AI bias is the tendency of a model to produce systematically different outcomes for different groups — by sex, race, age, postcode, nationality — not because the group matters to the task but because the training data, the features or the objective encoded the difference. In fintech it is a fair-lending and consumer-protection problem with law behind it: a credit model that declines women more, an identity check that fails darker faces more.
- [AI Tokens](https://mahouse.world/glossary/ai-tokens.md): AI tokens are the units into which a language model splits text — whole words, parts of words, punctuation or bytes — before processing it, and the unit in which model use is measured and billed. Providers price input and output tokens separately, usually per million, so token volume drives the running cost of any AI product.
- [AI Underwriting](https://mahouse.world/glossary/ai-underwriting.md): AI underwriting is the use of machine-learning models to decide whether to lend to, or insure, an applicant and on what terms, scoring risk from credit files, bank transactions and other data rather than from fixed rules alone. In consumer lending it remains a regulated credit decision: the lender must explain refusals and show the model does not discriminate.
- [AML](https://mahouse.world/glossary/aml.md): Anti-money laundering (AML) is the body of law and the set of controls that require financial institutions to prevent, detect and report the movement of criminal proceeds through their services: identifying customers and their beneficial owners, assessing their risk, monitoring transactions, screening against sanctions, reporting suspicious activity and keeping records. Its rules come from FATF's standards, transposed into EU, UK and US law; for a fintech they are a licensing condition and the most common subject of enforcement.
- [API](https://mahouse.world/glossary/api.md): An application programming interface (API) is the defined contract by which one piece of software requests data or actions from another — the endpoints, the requests and responses, the authentication, the errors — so that systems built by different teams and companies can work together without sharing their internals. Fintech runs on APIs: a payment is an API call, a bank account is read through one, and a fintech's own product is often an API its customers integrate.
- [API Banking](https://mahouse.world/glossary/api-banking.md): API banking is the exposure of banking services — account information, payment initiation, virtual accounts, cards, FX, lending — as programmable interfaces that authorised third parties and corporate customers call directly from their own systems, instead of through a branch, a file transfer or a portal; it includes the mandated open-banking APIs under PSD2, the commercial APIs banks sell to corporate treasuries and platforms, and banking-as-a-service APIs. It turned the bank into a service that software calls.
- [API Endpoints](https://mahouse.world/glossary/api-endpoints.md): An API endpoint is a specific address — a URL and a method — at which a piece of software accepts requests or delivers responses through an application programming interface: POST /payments creates a payment, GET /accounts/{id}/transactions lists transactions. An API is the set of its endpoints and the contract each follows; the design of the endpoints — what each does, what it requires, how it fails, whether it is safe to retry — decides how easily customers integrate.
- [Account-to-Account Payment](https://mahouse.world/glossary/account-to-account-payment.md): An account-to-account payment moves money directly from the payer's bank account to the payee's, without a card network in between — a transfer initiated by the payer, a direct-debit mandate, or, under open banking, a licensed third party at the checkout. It settles over domestic rails, instant where they exist, costs cents rather than a percentage, and is the alternative to cards the industry has built since PSD2.
- [Accretive Acquisition](https://mahouse.world/glossary/accretive-acquisition.md): An accretive acquisition raises the buyer's earnings per share: the target's earnings, plus synergies, less the cost of the financing, add more to the buyer's profit than the shares issued or the interest paid take away. It is the test listed acquirers announce first, because it is what analysts ask, and it is a test of financing arithmetic as much as of value — a deal can be accretive and still destroy it.
- [Acquihire](https://mahouse.world/glossary/acquihire.md): An acquihire is an acquisition made mainly to hire a company's people — usually an engineering or research team — rather than to own its product, revenue or customers. Most of the buyer's spend goes to the team as retention packages, the product is often shut down, and the start-up's investors may recover only part of what they put in.
- [Acquirer](https://mahouse.world/glossary/acquirer.md): An acquirer is the bank or licensed institution that lets a merchant accept card payments: it holds the merchant account, routes each transaction to the card network for the issuer's approval, settles the money to the merchant and is liable to the network for the merchant's chargebacks. In M&A the same word means the buyer of a company; in payments it means the merchant's bank — the most consolidated corner of fintech M&A.
- [Acquirer Reference Number](https://mahouse.world/glossary/acquirer-reference-number.md): An acquirer reference number is the 23-digit identifier the acquirer assigns to a card transaction when it submits it for clearing — the acquirer's identity, the date and a sequence — travelling with the transaction to the network, the issuer and the cardholder's bank, so a refund, a chargeback or a missing credit can be traced to one cleared item. It is what a cardholder's bank asks a merchant for.
- [Acquisition](https://mahouse.world/glossary/acquisition.md): An acquisition is the purchase of one company, or control of it, by another: the buyer takes the target's shares or its assets, the owners are paid in cash, shares or a mix, and the target joins the buyer's group. It is the commoner half of "mergers and acquisitions" — most deals called mergers are acquisitions with a politer name — and in fintech the usual exit and the usual way an incumbent buys a capability.
- [Address Verification Service](https://mahouse.world/glossary/address-verification-service.md): The address verification service is a check on card-not-present transactions in which the numeric parts of the billing address and postcode the customer enters are compared with the address the issuer holds, returning a match code — full, partial or none — that the merchant uses to accept, decline or review. It is old, cheap, US-centric and weak alone; it stays in every fraud stack because a mismatch is a signal.
- [Adjusted EBITDA](https://mahouse.world/glossary/adjusted-ebitda.md): Adjusted EBITDA is reported earnings before interest, tax, depreciation and amortisation with items added back or removed to show what the business earns in a normal year under the buyer's ownership — one-off costs out, an owner's above-market salary out, a discontinued product's losses out, and sometimes the run-rate of cost savings already made. It is the number sellers present, buyers test in a quality of earnings review, and multiples are applied to.
- [Agent Credential](https://mahouse.world/glossary/agent-credential.md): An agent credential is verifiable information that an AI agent presents to a merchant, bank or platform to prove what it is, which person or business it acts for and what it has been authorised to do. The recipient checks it, usually by verifying a cryptographic signature, before letting the agent browse, check out or pay.
- [Alpha](https://mahouse.world/glossary/alpha.md): Alpha is the part of an investment's return not explained by its market — what a manager added, or lost, by choosing differently from the index, after adjusting for the risk taken. A fund that returned 12% when its market rose 10% at a beta of 1.0 earned 2% of alpha; at a beta of 1.5 it earned negative alpha, because the market alone would have given 15%. It is what active management sells and mostly fails to deliver.
- [Alternative Payment Method](https://mahouse.world/glossary/alternative-payment-method.md): An alternative payment method is any way of paying that is not a card — digital wallets, account-to-account transfers, buy-now-pay-later, mobile money, cash vouchers, national schemes like iDEAL, BLIK, Pix or UPI — and in most markets outside the US and UK the "alternative" is the majority. For a merchant selling across borders, offering the right local methods decides conversion more than price does, and a PSP's method coverage is what it sells after approval rates.
- [Angel Investor](https://mahouse.world/glossary/angel-investor.md): An angel investor is an individual who invests their own money in a startup at its earliest stage — typically $10,000 to $250,000 per company, often before any fund will — in exchange for equity or a convertible instrument, and usually bringing experience, contacts and a willingness to lose the money. In fintech the useful angel is one who has run a regulated business and knows what the first licence application and the first sponsor bank will ask.
- [Annual Recurring Revenue](https://mahouse.world/glossary/annual-recurring-revenue.md): Annual recurring revenue (ARR) is the value of a company's active subscription contracts, normalised to one year: monthly recurring revenue multiplied by twelve, with one-off fees, usage charges and services left out. For a software or fintech business sold on subscriptions, it is the number a buyer prices first and audits hardest.
- [Anti-Dilution Protection](https://mahouse.world/glossary/anti-dilution-protection.md): Anti-dilution protection is a term of preferred shares that adjusts the investor's conversion price downward if the company later sells shares at a lower price, so that the investor ends up with more common shares than it originally paid for and its stake is protected against the down round. A full ratchet resets the price to the new round's; a weighted-average ratchet — the usual form — adjusts it in proportion to how much was raised at the lower price.
- [Antitrust Clearance](https://mahouse.world/glossary/antitrust-clearance.md): Antitrust clearance is the approval a competition authority gives to a merger before it may complete, after reviewing whether the combination would substantially lessen competition — the DOJ or FTC in the US, the Commission in the EU, the CMA in the UK, and every jurisdiction whose thresholds the deal crosses. It is a condition precedent in most deals above a size, a timetable of months, and why some signed deals never close.
- [Approval Rate](https://mahouse.world/glossary/approval-rate.md): The approval rate is the share of payment authorisation requests that the issuer approves — 100% less the decline rate — measured by count or by value, by card type, issuer, country and channel. It is the payments metric that moves revenue most directly: a merchant losing 8% of attempts loses 8% of sales it had already made, and a provider that approves two points more than a rival is worth more than one that charges two basis points less.
- [Artificial General Intelligence](https://mahouse.world/glossary/artificial-general-intelligence.md): Artificial general intelligence (AGI) is a proposed form of AI able to learn, reason and perform across the full range of cognitive tasks at a broadly human level — rather than one task at a time, as every deployed system does — and, in the labs' stated missions, the goal they are building toward. The term has no agreed definition; for business the practical meanings are the labs' timelines, their contracts and the capability trajectory of rented models.
- [Artificial Intelligence](https://mahouse.world/glossary/artificial-intelligence.md): Artificial intelligence (AI) is software that performs tasks which would otherwise need human judgement — recognising patterns, interpreting language and images, predicting outcomes, making or recommending decisions — mostly, since the 2010s, by learning from data rather than by following rules written by people. In fintech it runs fraud detection, credit decisions, identity verification, customer service and compliance; in M&A it is both a sector of its own and a capability every target claims and few can evidence.
- [Artificial Superintelligence](https://mahouse.world/glossary/artificial-superintelligence.md): Artificial superintelligence is a hypothetical AI system whose capabilities exceed the best human performance across essentially every intellectual domain — not merely matching people, which defines artificial general intelligence, but surpassing them by a margin that would make its judgement better than any person's or institution's. It does not exist; whether it could is contested; and the idea shapes the labs' missions, safety research and funding.
- [Asset Allocation](https://mahouse.world/glossary/asset-allocation.md): Asset allocation is the division of a portfolio among asset classes — equities, bonds, cash, property, alternatives — and within them by region and type, set to match the investor's goals, horizon and tolerance for loss. Studies since the 1980s attribute most of the variation in portfolio returns to the allocation rather than to the choice of securities within it, which is why wealthtech is built on allocation and low-cost funds rather than on stock-picking.
- [Asset Purchase](https://mahouse.world/glossary/asset-purchase.md): An asset purchase is the acquisition of a business by buying its individual assets — customer contracts, software, brand, equipment, staff — and chosen liabilities from the company that owns them, leaving the legal entity and everything not listed behind with the seller. It is the alternative to buying the company's shares.
- [Assets Under Management](https://mahouse.world/glossary/assets-under-management.md): Assets under management (AUM) is the total market value of the client money a firm manages or advises on: an asset manager's funds, a wealth platform's portfolios, a robo-adviser's accounts. Fees are charged as a percentage of it, so AUM is the revenue base, and the price of a wealth business is quoted as a share of it.
- [Attention Mechanism](https://mahouse.world/glossary/attention-mechanism.md): The attention mechanism is the component of a neural network that, for each element of the output being produced, computes a weighting over the elements of the input — which words in a sentence, which transactions in a history matter for this step — and combines them accordingly. Self-attention is the core of the transformer (2017); it is why language models can relate a word to one fifty pages earlier, and why cost grows with the square of context length.
- [Auction Process](https://mahouse.world/glossary/auction-process.md): An auction process is the structured way a company is sold to the best of several bidders: a teaser, an information memorandum under confidentiality, indicative offers, a shortlist into the data room, binding offers, and a final negotiation with one or two. The seller's adviser runs it to a timetable the bidders do not control, so that buyers compete on price and terms at the same moment.
- [Audit Trail](https://mahouse.world/glossary/audit-trail.md): An audit trail is the chronological, tamper-evident record of who did what, when, and on whose authority — every transaction, approval, configuration change and decision in a system. In a regulated financial business it is what turns "we have controls" into evidence, and in an acquisition it is the first thing a buyer's technical and compliance diligence asks to see.
- [Authorisation Hold](https://mahouse.world/glossary/authorisation-hold.md): An authorisation hold is the amount an issuer reserves against a cardholder's available balance or credit limit when it approves an authorisation — visible as "pending" — until the merchant captures, voids, or the hold expires under the issuer's and schemes' timing rules. It is why a hotel's deposit reduces the balance at check-in and why "the merchant hasn't charged me but the money is gone" is a support call.
- [Authorisation and Capture](https://mahouse.world/glossary/authorisation-and-capture.md): Authorisation and capture are the two steps of a card payment: authorisation, in which the issuer approves the transaction and holds the amount against the cardholder's balance, and capture, in which the merchant confirms the amount to take and submits it for clearing. They can happen together or days apart, which lets a merchant confirm stock before charging, adjust the amount at check-out, or take a deposit never captured.
- [Automated Clearing House](https://mahouse.world/glossary/automated-clearing-house.md): An automated clearing house is a network that moves money between bank accounts in batches — payroll, direct debits, bill payments, account-to-account transfers — by collecting instructions from originating banks, netting them and settling the results across central-bank accounts, once or a few times a day. The US ACH network is the largest; every country has an equivalent, and every fintech paying out or pulling funds by bank transfer rides one.
- [Autonomous Systems](https://mahouse.world/glossary/autonomous-systems.md): An autonomous system is a machine or piece of software that senses its surroundings, decides what to do and acts on that decision without a person directing each step — a driverless car, a warehouse robot, an inspection drone, an AI agent. Autonomy comes in degrees, and the degree largely decides who is responsible when the system gets it wrong.
- [Average Ticket Size](https://mahouse.world/glossary/average-ticket-size.md): Average ticket size is a merchant's card volume divided by its number of transactions — the typical value of one sale — and it shapes almost everything in payments economics: the fixed per-transaction fee matters on a €4 coffee and not on a €4,000 sofa; fraud and chargeback exposure scale with it; underwriters read it as the size of each thing that can go wrong; and a PSP's pricing model is built around where a merchant sits on the distribution.
- [BaaS](https://mahouse.world/glossary/baas.md): Banking as a service (BaaS) is the model in which a licensed institution — a bank, an e-money or payment institution — provides regulated products through APIs (accounts, payments, card issuing, sometimes lending) to companies that offer them under their own brand without holding a licence; a BaaS platform often sits between them. It supplied the licences behind most neobanks and embedded-finance products, and the regulatory actions against several BaaS banks in 2023–2024 showed where its oversight had failed.
- [Backpropagation](https://mahouse.world/glossary/backpropagation.md): Backpropagation is the algorithm that trains a neural network: after a forward pass produces a prediction and a loss function measures its error, backpropagation applies the chain rule backward through the layers to compute how much each parameter contributed to the error — the gradient — and gradient descent adjusts every parameter a small step to reduce it. Popularised in 1986 and made practical by GPUs, it is the engine of all deep learning.
- [Bank Charter](https://mahouse.world/glossary/bank-charter.md): A bank charter is the licence that lets a company take deposits, lend against them and hold them under deposit insurance, granted by a banking regulator after a review of its capital, owners, management and business plan. Fintechs get one by applying for it, by buying a bank that has one, or by renting one from a partner bank.
- [Bank Identification Number](https://mahouse.world/glossary/bank-identification-number.md): A bank identification number is the first six — now eight — digits of a card number, identifying the issuing bank, the network, the card type (debit, credit, prepaid, commercial), the product tier and the country. Every card is issued under a BIN the scheme assigned to a member; a fintech without its own membership issues under a BIN sponsor's number, and a merchant's fraud and routing engines read the BIN first.
- [Beta](https://mahouse.world/glossary/beta.md): Beta is the sensitivity of an investment's return to the market's: a beta of 1.0 moves with the market, 1.5 half again as much in both directions, 0.5 half as much. It measures the risk diversification cannot remove — the market's own — and in the capital asset pricing model it scales the equity premium, so a fintech at 1.4 is discounted at a higher rate than a utility at 0.6.
- [Bilateral Negotiation](https://mahouse.world/glossary/bilateral-negotiation.md): A bilateral negotiation is a sale of a company to a single buyer without a competitive process — the seller and one counterparty agree the price and the terms directly, usually because the buyer approached first, the fit is obvious, or the seller values confidentiality and speed over the price an auction might add. It is how most small and many strategic fintech deals happen, and its risk is the leverage the seller never had.
- [Binding Offer](https://mahouse.world/glossary/binding-offer.md): A binding offer is a bidder's firm proposal at the end of the second round of a sale process: a price, a marked-up sale agreement it is prepared to sign, evidence of committed funding, its approvals and conditions, and a timetable to signing. Unlike the indicative offer it is not a range and not a conversation-opener; it is what the bidder will sign, and the seller chooses between binding offers on the terms as much as the number.
- [Blended Pricing](https://mahouse.world/glossary/blended-pricing.md): Blended pricing charges a merchant one rate for every card transaction — 1.5%, or 2.9% plus 30 cents — whatever the card's actual interchange and scheme fees, with the provider absorbing the differences and keeping the spread. It is simple, predictable and, for a merchant whose customers use cheap cards, expensive; it is how small merchants pay and how payment service providers earn most of their margin.
- [Blockchain](https://mahouse.world/glossary/blockchain.md): A blockchain is a shared ledger maintained by many participants rather than one operator, in which transactions are grouped into blocks, each cryptographically linked to the last, so that the record cannot be altered without the alteration being evident, and agreed by a consensus mechanism rather than a central authority. Public blockchains are open to anyone; permissioned ledgers are run by known participants. In finance it underlies crypto-assets, stablecoins and tokenised securities.
- [Blockchain-Based Settlement](https://mahouse.world/glossary/blockchain-based-settlement.md): Blockchain-based settlement is the completion of a transaction — the final transfer of an asset, of money, or of both at once — by recording it on a shared distributed ledger instead of in the separate books of banks, depositories and clearing houses. Its promise is settlement in minutes, at any hour, with the asset and the cash moving together.
- [Bolt-On Acquisition](https://mahouse.world/glossary/bolt-on-acquisition.md): A bolt-on acquisition is a small company bought by a larger one already in the same market and folded into it — to add customers, a product, a geography or a licence — rather than run as a separate business. In private equity it is the "build" in buy-and-build; in a corporate it is how a platform grows faster than it could by hiring.
- [Book Building](https://mahouse.world/glossary/book-building.md): Book-building is the process by which underwriters set the price of a share issue: they announce a price range, collect indications from institutional investors of how many shares they would buy at what price, and from that book of orders choose the final price and allocate the shares. It replaced fixed-price offers because it prices from demand rather than from the banks' guess, and it is where an IPO's "money left on the table" is decided.
- [Break Fee](https://mahouse.world/glossary/break-fee.md): A break fee is a sum one party to a signed acquisition agreement pays the other if the deal fails for a reason the agreement assigns to it — the target accepting a higher offer, or the buyer failing to obtain financing or regulatory approval. A reverse break fee is the same mechanism pointed at the buyer.
- [Bridge Financing](https://mahouse.world/glossary/bridge-financing.md): Bridge financing is money raised to carry a company to its next round, a sale or profitability when the current runway will not — usually a convertible note or SAFE from existing investors, sized for six to twelve months, converting at a discount into the round it bridges to. A bridge from insiders who believe in the plan is ordinary; a bridge because nobody else will invest is the round before the down round.
- [Burn Rate](https://mahouse.world/glossary/burn-rate.md): Burn rate is the cash a company loses each month: gross burn is total spending, net burn is spending less revenue — the figure that, divided into the cash balance, gives the months of runway. It is the number a startup's board watches most, the one that decides when the next round must close, and in a fintech it is complicated by capital the licence requires the company to hold and not spend.
- [Buy Now Pay Later](https://mahouse.world/glossary/buy-now-pay-later.md): Buy now, pay later is short-term consumer credit offered at the checkout — typically the purchase split into three or four interest-free instalments over six weeks to three months — with the merchant paying the provider a fee of 2–8% for higher conversion and a larger basket, and the provider bearing the credit risk. Klarna, Afterpay, Affirm and PayPal built the category; losses, funding costs and regulation reshaped it.
- [Buy-Side Advisory](https://mahouse.world/glossary/buy-side-advisory.md): Buy-side advisory is the work an investment bank or M&A boutique does for a buyer: finding and screening targets, making the approach, valuing the business, structuring the offer, running diligence and negotiating the terms to signing. The seller's adviser runs the process; the buyer's adviser gets the buyer through it at the right price — or tells it to stop.
- [Buy-and-Build](https://mahouse.world/glossary/buy-and-build.md): Buy-and-build is a private equity strategy in which an investor acquires one company as a platform and then buys several smaller ones in the same market to add to it, integrating them into a larger business that sells for a higher multiple than the pieces were bought for. The value comes from scale, shared costs and the gap between small-company and large-company multiples.
- [CBDC](https://mahouse.world/glossary/cbdc.md): A central bank digital currency (CBDC) is money issued by a central bank in digital form as a direct liability of the central bank — like cash, unlike a bank deposit or an e-money balance, which are claims on private institutions. A retail CBDC would be held and spent by the public through wallets provided by banks and payment firms; a wholesale CBDC would settle between financial institutions on programmable infrastructure. The digital euro is in preparation.
- [CET1 Ratio](https://mahouse.world/glossary/cet1-ratio.md): The CET1 ratio is a bank's common equity tier 1 capital — ordinary shares, share premium, retained earnings and reserves, less regulatory deductions such as goodwill — divided by its risk-weighted assets. It is the main measure of a bank's loss-absorbing strength: Basel III sets a 4.5% minimum, and buffers and supervisory add-ons lift the real requirement well above it.
- [CLIP](https://mahouse.world/glossary/clip.md): CLIP — Contrastive Language-Image Pre-training — is a model released by OpenAI in 2021 that learned to place images and their text descriptions into a shared embedding space, trained on 400 million image-caption pairs by pulling matching pairs together and pushing mismatched ones apart. It made zero-shot image classification possible, became the text encoder guiding diffusion image generators, and established the recipe every multimodal model since has used to connect vision and language.
- [CLOUD Act](https://mahouse.world/glossary/cloud-act.md): The CLOUD Act (Clarifying Lawful Overseas Use of Data Act) is a 2018 US federal law requiring providers of communication and cloud services to disclose customer data in their possession, custody or control in response to valid US legal process, wherever that data is stored. It also lets the US sign data-access agreements with foreign governments that meet its privacy standards.
- [Cap Table](https://mahouse.world/glossary/cap-table.md): A cap table is the record of who owns what in a company — every class of share, every holder, every option, warrant and convertible instrument, with the price paid and the rights attached — and what each would receive at a given exit value once preferences and conversions are applied. It is the first document an investor, an acquirer or a regulator asks for, and the one most startups keep badly.
- [Capital Call](https://mahouse.world/glossary/capital-call.md): A capital call is a fund's demand that its investors pay in part of the money they committed — a limited partner who committed €50m pays nothing at signing, then receives call notices over the investment period as the fund makes investments and pays fees. It is why committed and invested capital differ, why dry powder is a number, and why an LP that cannot meet a call is in default.
- [Capital Structure](https://mahouse.world/glossary/capital-structure.md): Capital structure is the mix of debt and equity that funds a company's assets — senior loans, junior and mezzanine debt, preferred and ordinary shares — and the order in which those claims are paid. It sets the cost of capital, the risk the equity bears and the room to grow; a buy-out rewrites it in a day, and for a bank or a licensed fintech the regulator has already written the lower bound.
- [Capture](https://mahouse.world/glossary/capture.md): Capture is the merchant's confirmation that an authorised payment should be taken — the instruction to the acquirer that turns the issuer's hold into a charge by submitting the transaction for clearing. It may be for the full amount, for less with the balance released, or in several parts as an order ships; it must happen within the authorisation's window, and until it does the merchant has approval but no money.
- [Card Network](https://mahouse.world/glossary/card-network.md): A card network is the company — Visa, Mastercard, American Express, UnionPay and the domestic schemes — that sets the rules and runs the infrastructure connecting the bank that issued a card with the bank or acquirer serving the merchant, so that a payment authorised in one place is settled in another. Banks and licensed fintechs join as members; everyone else pays them for access.
- [Card-Not-Present Transaction](https://mahouse.world/glossary/card-not-present-transaction.md): A card-not-present transaction is any card payment in which the card is not physically read by a terminal — online, in an app, by phone, by mail, or a stored-card charge — so the merchant cannot verify the chip, the PIN or the cardholder's presence. It carries higher interchange, higher fraud and a liability that rests with the merchant unless authentication shifts it; its controls are most of e-commerce payments engineering.
- [Card-Present Transaction](https://mahouse.world/glossary/card-present-transaction.md): A card-present transaction is a card payment in which the card — or the phone standing in for it — is physically read by a terminal: inserted for a chip transaction, tapped for contactless, or swiped on a stripe where that still exists. The chip's cryptogram and the PIN or biometric prove the card and the cardholder, so fraud is rare, interchange lower, and the issuer bears the fraud loss if it approves.
- [Carried Interest](https://mahouse.world/glossary/carried-interest.md): Carried interest is the share of a fund's profits — typically 20% — that its general partner receives once the limited partners have got their capital back and, usually, a preferred return; it is the manager's reward for performance, distinct from the management fee that pays the firm's costs. It is where the partners of private equity and venture firms make their money, why the fund structure exists, and a tax question in every jurisdiction that has one.
- [Cart Abandonment](https://mahouse.world/glossary/cart-abandonment.md): Cart abandonment is the share of shoppers who add items to a basket and leave without completing the purchase — around 70% across e-commerce by most studies — of which a meaningful part happens at the payment step: a forced account, a missing payment method, a clumsy authentication, a decline, a form that fails on a phone. Payments teams own that part, measure it as checkout abandonment, and recover it with guest checkout, wallets, local methods, frictionless authentication and retries.
- [Carve-Out](https://mahouse.world/glossary/carve-out.md): A carve-out is the sale of part of a company — a division, a product line, a subsidiary — that has to be separated from the rest before or as it is sold, because it shares systems, staff, contracts or licences with what the seller keeps. The separation work, not the negotiation, is what makes carve-outs slow and expensive.
- [Cascading](https://mahouse.world/glossary/cascading.md): Cascading is the automatic resubmission of a payment that failed or was soft-declined on one acquirer or route to a second, and if necessary a third, within the same checkout — so an issuer's decline that depended on the acquiring path, or an acquirer's outage, is not the customer's failure. It is a feature of orchestration platforms, bounded by the schemes' reattempt rules and by the fact that a genuine decline cascades into another.
- [Cash-Free Debt-Free](https://mahouse.world/glossary/cash-free-debt-free.md): Cash-free debt-free is the basis on which most private company prices are quoted: the buyer pays for the business as if it had no cash and no debt, and the price is adjusted at completion — up for cash left in, down for debt inherited and debt-like items, and for the gap between actual and normal working capital. The headline is enterprise value; what the seller receives is equity value.
- [Central Bank Money](https://mahouse.world/glossary/central-bank-money.md): Central bank money is a direct liability of the central bank: banknotes held by the public, and the reserve or settlement balances that banks and other eligible institutions hold at the central bank. Carrying no private credit risk, it is the asset in which interbank payments finally settle and the anchor that keeps bank deposits worth par.
- [Chain of Thought](https://mahouse.world/glossary/chain-of-thought.md): Chain of thought is the technique of having a language model set out intermediate reasoning steps before its final answer — "let's think step by step" — which measurably improves accuracy on arithmetic, logic and multi-step problems the model gets wrong when asked directly. Described in a 2022 Google paper, it became standard prompting practice and then, in the reasoning models of 2024–2025, a built-in behaviour. The trace reads like an explanation but is not always the model's real computation.
- [Change of Control Clause](https://mahouse.world/glossary/change-of-control-clause.md): A change-of-control clause is a term in a contract — a customer agreement, a licence, a lease, a loan, a scheme membership, a sponsor-bank agreement — that gives the other party rights when the company's ownership changes: to be notified, to consent, to renegotiate, or to terminate. In a fintech acquisition the clauses in the target's contracts decide which relationships survive the sale, and finding them is half of legal due diligence.
- [Chargeback](https://mahouse.world/glossary/chargeback.md): A chargeback is the reversal of a card payment by the cardholder's issuing bank at the cardholder's request — for fraud, goods not received, or a charge not recognised — under rules set by the card network, with the money taken back from the merchant through its acquirer. The merchant may contest it by representment, and a high chargeback ratio can cost a merchant its ability to accept cards.
- [Chargeback Ratio](https://mahouse.world/glossary/chargeback-ratio.md): The chargeback ratio is the number of chargebacks a merchant receives in a month divided by its number of transactions — Visa counts the same month's transactions, Mastercard the previous month's — and it is the measure the card schemes use to decide whether a merchant is a problem. Above about 0.9–1% (with a minimum count), the merchant enters a monitoring programme with fines that escalate monthly; above 1.8–2%, the acquirer faces pressure to terminate it.
- [Chargeback Reason Code](https://mahouse.world/glossary/chargeback-reason-code.md): A chargeback reason code is the category the card scheme assigns to a dispute — fraud, authorisation, processing error, or consumer dispute (not received, not as described, cancelled recurring, credit not processed) — which tells the merchant why the payment was reversed, what evidence would win a representment, and how long everyone has. Visa's are numbers like 10.4 and 13.1, Mastercard's like 4837 and 4853.
- [Clearing](https://mahouse.world/glossary/clearing.md): Clearing is the process by which payment institutions exchange the records of transactions between them and agree what each owes the other — the acquirer submits the day's captured card transactions to the network, the network validates and sorts them by issuer, computes the fees, and produces the obligations that settlement will pay. It is the bookkeeping between authorisation (the promise) and settlement (the money): after clearing, the amounts are fixed and final; before settlement, no money has moved.
- [Coding Agents](https://mahouse.world/glossary/coding-agents.md): Coding agents are AI systems, built on large language models with tools, that plan and carry out software-development tasks with a degree of autonomy — reading a codebase, changing code across files, running tests, fixing what fails, and preparing a pull request for a person to review — rather than only suggesting the next line. They became the fastest-adopted category of enterprise AI and the subject of the largest AI application deals of 2025.
- [Commercial Due Diligence](https://mahouse.world/glossary/commercial-due-diligence.md): Commercial due diligence is the buyer's investigation of a target's market and its position in it: the size and growth of the market, the competitors and their pricing, what customers actually think of the product and whether they would leave, and whether the forecast the price rests on is one the market will allow. Financial diligence tests the numbers the business reported; commercial diligence tests the numbers it promises.
- [Commoditization](https://mahouse.world/glossary/commoditization.md): Commoditization is the point at which a product that once won on capability wins only on price, because every supplier's version is good enough and the buyer can no longer tell them apart. In fintech it has happened to card processing, cloud infrastructure and, increasingly, to the basic bank account — and it is the most common reason two competitors merge.
- [Community Reinvestment Act](https://mahouse.world/glossary/community-reinvestment-act.md): The Community Reinvestment Act (CRA) is a 1977 US federal law requiring the federal banking agencies to encourage the banks they supervise to meet the credit needs of the communities where they operate, including low- and moderate-income neighbourhoods, consistent with safe and sound operation. Examiners rate each bank's record, which weighs on its applications for mergers, acquisitions and branches.
- [Comparable Company Analysis](https://mahouse.world/glossary/comparable-company-analysis.md): Comparable company analysis values a business by the multiples at which similar listed companies trade — enterprise value to revenue, to EBITDA, or price to earnings or to book — applied to the target's own figures. It answers "what would the market pay for this today", is the first page of every valuation, and for a fintech it depends on the honest choice of which listed companies are actually comparable.
- [Compelling Evidence](https://mahouse.world/glossary/compelling-evidence.md): Compelling evidence is the documentation a merchant submits to reverse a chargeback — proof that the cardholder made the purchase and received what was bought — and, since 2023, the name of Visa's rule (CE 3.0) under which a merchant defeats a card-absent fraud chargeback by showing two earlier undisputed transactions from the same cardholder, matched on device, IP, login or shipping address.
- [Completion Accounts](https://mahouse.world/glossary/completion-accounts.md): Completion accounts are a balance sheet drawn up as at the day a deal completes, used to adjust the price paid on that day for the actual cash, debt and working capital of the company. The buyer pays an estimated price at completion and the difference is settled once the accounts are agreed, usually within three months.
- [Compute](https://mahouse.world/glossary/compute.md): Compute, in AI, is the processing capacity consumed by training and running models — measured in floating-point operations (a frontier training run in 2024–2025 used on the order of 10^25 to 10^26 FLOPs), in accelerator-hours and in dollars — and it is the resource the industry is constrained by. Capability has scaled with it for a decade, and the EU AI Act uses a compute threshold to define the models it regulates most.
- [Computer Vision](https://mahouse.world/glossary/computer-vision.md): Computer vision is the field of AI that interprets images and video — recognising what is in them, reading text, locating objects, verifying faces, detecting manipulation — from raw pixels rather than from structured input; deep learning made it work (2012 onward) and it now runs in every phone. For a fintech it is document capture and reading, identity verification, card scanning, fraud evidence review, and the defence against generated and manipulated images; most of it is rented from vendors.
- [Conditions Precedent](https://mahouse.world/glossary/conditions-precedent.md): Conditions precedent are the things that must happen between signing a sale agreement and completing it — regulatory approvals, competition clearance, third-party consents, shareholder votes, the absence of a material adverse change — before either side is obliged to close. They are why a fintech deal is signed and completed months apart, and where the risk of a deal that never closes is allocated.
- [Conglomerate Merger](https://mahouse.world/glossary/conglomerate-merger.md): A conglomerate merger combines two companies in unrelated businesses — neither competitors nor supplier and customer — so that the combined group is more diversified but no more efficient in any one market. It is the form regulators worry about least and shareholders discount most, and in fintech it appears when a retailer, a telecom or an industrial group buys a payments or lending business it does not know how to run.
- [Contactless Payment](https://mahouse.world/glossary/contactless-payment.md): A contactless payment is a card-present transaction made by holding a card, phone or wearable near a terminal, which exchanges the EMV chip data over near-field communication in under a second without inserting the card or, below a limit, entering a PIN. It made small payments faster than cash, took most of cash's share in the markets that adopted it, and moved the card into the phone — where the wallet's biometric replaced the PIN and the limit disappeared.
- [Control Premium](https://mahouse.world/glossary/control-premium.md): A control premium is the amount above a company's market price that a buyer pays to own it outright — to appoint the board, set the strategy, take the cash flows and capture the synergies — expressed as a percentage of the share price before the bid was known. It is why acquisition multiples exceed trading multiples, and why a listed company's price is not what it would sell for.
- [Convertible Note](https://mahouse.world/glossary/convertible-note.md): A convertible note is a loan to a company that converts into shares at its next priced funding round, usually at a discount to that round's price or at a valuation cap, rather than being repaid in cash. It lets a startup take money now without agreeing a valuation, gives the investor debt's protection until conversion and equity's upside after it, and is the older, more creditor-like cousin of the SAFE.
- [Convolutional Neural Network](https://mahouse.world/glossary/convolutional-neural-network.md): A convolutional neural network is a neural network built for grid-shaped data — images above all — that slides small learned filters across the input to detect local patterns (edges, textures, then shapes, then objects) and stacks these layers so that detail builds into recognition. It started the deep-learning era (AlexNet, 2012), dominated image recognition for a decade, and still runs wherever speed on small hardware matters, including the document and card scanners in fintech apps.
- [Cost of Equity](https://mahouse.world/glossary/cost-of-equity.md): The cost of equity is the return a company's shareholders require to hold its shares given the risk — the rate their expected cash flows are discounted at, and the hurdle any use of their capital must clear. It is estimated, not observed: the risk-free rate plus an equity premium scaled by beta, plus premiums for size and illiquidity; for a private fintech it lands between 12% and 20%.
- [Cost-Aware Routing](https://mahouse.world/glossary/cost-aware-routing.md): Cost-aware routing is the selection, transaction by transaction, of the cheapest available path — which acquirer, which network, and for debit cards which of the networks the card supports — within the constraints of approval rates and the rules. Its best-known form is least-cost routing of debit in the US and Australia; its general form is what payment orchestration sells to multi-acquirer merchants.
- [Cost-to-Income Ratio](https://mahouse.world/glossary/cost-to-income-ratio.md): The cost-to-income ratio is a bank's operating expenses divided by its operating income, expressed as a percentage: the share of each unit of revenue consumed by running the business before loan losses and tax. Lower is better. It is the standard efficiency measure for banks and lenders, and the number most cost-cutting plans and bank mergers are built around.
- [Counter-Terrorism Financing](https://mahouse.world/glossary/counter-terrorism-financing.md): Counter-terrorism financing is the set of duties and controls that stop money reaching terrorist organisations — screening against designated lists, monitoring for the patterns that fund terrorism, reporting suspicions — enforced alongside anti-money-laundering rules under the same FATF standards. It differs in what it looks for: laundering hides large criminal sums; terrorist financing moves small, often clean sums to a prohibited end.
- [Counterparty Risk](https://mahouse.world/glossary/counterparty-risk.md): Counterparty risk is the risk that the other side of a transaction fails to perform — does not pay, does not deliver, or becomes insolvent between the moment you have performed and the moment they should. In payments it is the acquirer's exposure to a merchant that takes money for goods it will never ship, and the whole reason reserves, escrows and settlement rules exist.
- [Covenant](https://mahouse.world/glossary/covenant.md): A covenant is a promise in a loan agreement about what the borrower will and will not do: keep leverage below a ratio and interest cover above one (financial covenants, tested quarterly), deliver accounts, not borrow more, sell assets or pay dividends without consent. It is the lender's early warning and its lever — a breach lets the lender demand repayment, and in practice renegotiate the price of not doing so.
- [Credit](https://mahouse.world/glossary/credit.md): Credit is an arrangement in which a lender provides money, goods or services now in exchange for the borrower's promise to repay later, usually with interest or fees. It covers loans, credit cards, overdrafts, trade credit between businesses and buy-now-pay-later, and its price reflects the lender's view of how likely the borrower is to repay.
- [Credit Derivatives](https://mahouse.world/glossary/credit-derivatives.md): Credit derivatives are contracts that transfer the risk of a borrower's default from one party to another without transferring the loan itself. The buyer of protection pays a periodic premium; the seller pays out if a defined credit event — default, bankruptcy, restructuring — occurs. They are the instrument behind synthetic securitisation, and the way a bank keeps a customer relationship while selling its risk.
- [Credit Rating](https://mahouse.world/glossary/credit-rating.md): A credit rating is an independent agency's assessment of how likely a borrower or a debt instrument is to default — on a scale from AAA through BBB (the lowest investment grade) to CCC and D — that sets what the borrower pays to borrow and which investors may lend to it. Moody's, S&P and Fitch assign most; a fintech meets them when it issues bonds or securitises its loans.
- [Cross-Border Fee](https://mahouse.world/glossary/cross-border-fee.md): A cross-border fee is the additional charge a card scheme, and often the issuer, applies when a card was issued in a different country from the merchant's acquirer — a scheme assessment of roughly 0.4–1% on top of the domestic fee, and interchange at the higher inter-regional rate — passed to the merchant through its acquirer. It is why a foreign card costs a merchant two to four times a domestic one.
- [Cross-Border Payments](https://mahouse.world/glossary/cross-border-payments.md): Cross-border payments are transfers in which the payer and the payee are in different countries, usually in different currencies, so the money crosses at least one border, one currency and one banking system on its way. They are slower, dearer and less transparent than domestic payments, which is why they are where fintechs have taken the most share from banks and where the largest payments acquisitions have been made.
- [Cultural Due Diligence](https://mahouse.world/glossary/cultural-due-diligence.md): Cultural due diligence is the buyer's assessment of how a target actually works — how decisions are made, how people are paid and promoted, how risk and compliance are treated, how the founders lead — against the buyer's own ways, so the integration plan is built on what the two organisations are rather than on their org charts. It is the diligence most often skipped and the one whose absence explains most failed integrations.
- [Customer Due Diligence](https://mahouse.world/glossary/customer-due-diligence.md): Customer due diligence is the process a regulated firm must complete before and during a business relationship: identify and verify the customer, identify any beneficial owner, understand the purpose and expected activity, assess the risk, and keep it current. It is what KYC means in law, and it comes in three strengths: simplified for low-risk, standard for most, enhanced for the high-risk and the politically exposed.
- [DORA](https://mahouse.world/glossary/dora.md): DORA — the Digital Operational Resilience Act, Regulation (EU) 2022/2554 — is the EU regulation, applicable since 17 January 2025, that sets common rules for how banks, payment and e-money institutions, insurers, investment firms, crypto-asset service providers and other financial entities manage technology risk, report ICT incidents, test their resilience and control their dependence on outside technology providers.
- [Data Augmentation](https://mahouse.world/glossary/data-augmentation.md): Data augmentation is the practice of enlarging a training set by producing modified versions of its examples — rotated, cropped and re-lit images; paraphrased sentences; resampled or perturbed transactions — so that a model sees more variation than the original data contains and learns the pattern rather than the particulars. It is a standard defence against overfitting when labelled data is scarce, a way to rebalance rare classes, and, with generative models producing the examples, a powerful and risky one.
- [Data Room](https://mahouse.world/glossary/data-room.md): A data room is the secure online repository in which a seller places the documents a buyer needs for due diligence — contracts, accounts, licences, correspondence, code — organised by topic and opened to named users under a non-disclosure agreement. Its index becomes the record of what the buyer was shown, which matters for warranties later.
- [Data Sovereignty](https://mahouse.world/glossary/data-sovereignty.md): Data sovereignty is the principle that data is subject to the laws of the jurisdiction where it is collected, stored or processed — and, in practice, to the laws of every state that can compel the company holding it. For a fintech or AI business it decides where data may sit, who may handle it and who can demand it.
- [Data as a Service](https://mahouse.world/glossary/data-as-a-service.md): Data as a service is a business model in which a provider sells data itself — collected, cleaned, kept up to date and delivered on demand through an API, a feed or a cloud data share — usually for a subscription or usage fee. The customer buys the finished data rather than software to gather and process it themselves.
- [DeFi](https://mahouse.world/glossary/defi.md): Decentralised finance (DeFi) is the set of financial services — lending and borrowing, trading, derivatives, asset management, stablecoin issuance — operated by smart contracts on public blockchains rather than by licensed intermediaries: users interact with code that holds collateral, matches trades and enforces terms, governed by token holders and open to anyone with a wallet. It reached tens of billions of dollars of deposited value, suffered hacks and collapses that cost billions, and is partly absorbed into regulated finance.
- [Decline Rate](https://mahouse.world/glossary/decline-rate.md): The decline rate is the share of card payment attempts that are refused at authorisation — by the issuer, the network or the merchant's own fraud rules — measured against attempts. For an online merchant every declined legitimate transaction is lost revenue, and for a payments company the approval rate it achieves for its merchants is a product in itself.
- [Decline Recovery](https://mahouse.world/glossary/decline-recovery.md): Decline recovery is the set of practices that turn a declined payment into a completed one — retrying soft declines at the right moments, refreshing expired or replaced cards through account updaters and network tokens, routing the retry through another acquirer, and asking the customer for a new card through dunning. For a subscription business it is the difference between involuntary churn and revenue.
- [Deep Learning](https://mahouse.world/glossary/deep-learning.md): Deep learning is machine learning with neural networks of many layers, which learn useful representations of raw data — pixels, audio, text, sequences of events — layer by layer, instead of relying on features designed by hand; it is the technology behind image recognition, speech, the transformers and large language models, and generative AI. It needs large data and compute and wins wherever the input is unstructured; on the tabular data of most financial decisions, gradient-boosted trees still usually win.
- [Depreciation](https://mahouse.world/glossary/depreciation.md): Depreciation is the accounting allocation of the cost of a tangible asset — equipment, vehicles, buildings, hardware — over its useful life, so that each year's profit bears a share of the investment that produced it rather than the whole cost in the year of purchase; amortisation is the same for intangible assets — software, licences, and the intangibles a buyer records after an acquisition. It is a non-cash expense, which is why EBITDA adds it back.
- [Diffusion Model](https://mahouse.world/glossary/diffusion-model.md): A diffusion model is a generative model that learns to produce data — images above all, also audio, video and molecules — by learning to reverse a process that gradually turns data into noise: trained to remove a little noise at each step, it can start from pure noise and denoise its way to an image matching a text prompt. Stable Diffusion, DALL·E and Midjourney are diffusion models; they make the synthetic documents and faces fintech identity teams detect.
- [Digital Payments](https://mahouse.world/glossary/digital-payments.md): Digital payments are payments in which money moves electronically and is initiated through a digital channel — an online checkout, a card tap, a mobile wallet, a banking app or a QR code — rather than with cash or cheques. They run on card networks, bank transfer systems or e-money balances, each with its own costs, speed and risks.
- [Digital Wallet](https://mahouse.world/glossary/digital-wallet.md): A digital wallet is an application that holds a customer's payment credentials or a stored balance and pays with them — by tapping a phone, by a button online, by a transfer to a friend. Pass-through wallets (Apple Pay, Google Pay) tokenise a card and pass the payment to it; stored-value wallets (PayPal, Alipay, M-Pesa) hold the customer's money under a licence. The first is a layer on the card system; the second is a payment system.
- [Dilutive Acquisition](https://mahouse.world/glossary/dilutive-acquisition.md): A dilutive acquisition lowers the buyer's earnings per share: the target's earnings, net of synergies and the cost of financing, add less to profit than the new shares or interest take away — usually because the target was bought at a higher earnings multiple than the buyer's own, or because it does not yet earn much. It is what a bank buying a fast-growing fintech usually announces, and the announcement is a statement about growth, not a confession.
- [Direct Debit](https://mahouse.world/glossary/direct-debit.md): A direct debit is a payment pulled from the payer's bank account by the payee under a mandate the payer signed once — the subscription, the utility bill, the loan repayment — through a scheme (SEPA Direct Debit, Bacs, ACH debit) that gives the payer a right to refund. It costs cents, needs no card that can expire, and carries a return risk: a SEPA Core debit can be reclaimed for eight weeks with no reason given.
- [Direct Listing](https://mahouse.world/glossary/direct-listing.md): A direct listing is a way of going public without an IPO: existing shares are admitted to trading and their holders may sell, but no new shares are sold, no underwriters guarantee a price and no book is built — an opening auction sets the price. It suits companies that need liquidity rather than capital and want to avoid the fee, the dilution and the discount; it lacks the IPO's proceeds, price support and curated shareholder base.
- [Disclosure Letter](https://mahouse.world/glossary/disclosure-letter.md): A disclosure letter is the seller's list of everything it knows that would make a warranty in the sale agreement untrue — the litigation, the customer that has given notice, the regulator's open enquiry — delivered at signing so that the buyer cannot later claim for what it was told. It is the seller's protection against the warranties, and the document a buyer's lawyers read most carefully in the last week before signing.
- [Discounted Cash Flow](https://mahouse.world/glossary/discounted-cash-flow.md): Discounted cash flow is a valuation method that estimates what a business is worth today by forecasting the cash it will generate in future years and reducing each year's figure by a rate that reflects the time value of money and the risk of not receiving it. The sum of those discounted cash flows, plus a terminal value for the years beyond the forecast, is the enterprise value.
- [Distressed M&A](https://mahouse.world/glossary/distressed-m-a.md): Distressed M&A is the purchase of a company or its assets when it is insolvent or about to be — through an administrator or bankruptcy court, a pre-packaged sale, or a rescue signed days before the money runs out. Prices are low, timetables short and warranties absent. In fintech the asset is usually a customer book and a platform without the licence, on a clock the regulator sets.
- [Diversification](https://mahouse.world/glossary/diversification.md): Diversification is the reduction of risk by holding many investments whose fortunes do not move together, so the losses of one are offset by the gains of others and only the risk common to all — the market's — remains: the same expected return at lower volatility. It fails when holdings move together after all, which is what a crisis is; and it applies to a loan book and a customer base as much as to a portfolio.
- [Divestiture](https://mahouse.world/glossary/divestiture.md): A divestiture is the sale or separation of a business unit, subsidiary or asset by the company that owns it — to a buyer, to its own shareholders as a spin-off, or into a joint venture — usually because the unit no longer fits the strategy, needs capital the parent will not provide, or must be sold to satisfy a regulator.
- [Dividend Yield](https://mahouse.world/glossary/dividend-yield.md): Dividend yield is a company's annual dividend per share divided by its share price — the cash return a shareholder receives from dividends alone, before any change in the price. A 4% yield on a €50 share is €2 a year. It is high for mature companies with more cash than uses, near zero for growth companies that reinvest everything, and, for a bank, a number the regulator has a view on, since every dividend is capital leaving.
- [Double Descent](https://mahouse.world/glossary/double-descent.md): Double descent is the observed pattern, named in 2019, that as a model's size or training time grows, its error on unseen data first falls, then rises as the model approaches the point where it can fit the training data exactly (the interpolation threshold), and then falls again as the model grows further — contradicting the classical picture in which bigger models simply overfit more. It explains why enormous networks generalise well; the mid-sized model may be the worst.
- [Down Round](https://mahouse.world/glossary/down-round.md): A down round is a funding round in which a company raises money at a lower valuation than its previous round, so that existing shareholders are diluted more than they expected and earlier investors' anti-dilution rights may be triggered. It is the usual outcome for a company that raised at the top of a cycle and grew more slowly than the price assumed.
- [Drag-Along Right](https://mahouse.world/glossary/drag-along-right.md): A drag-along right lets shareholders holding a defined majority force the remaining shareholders to sell their shares to the same buyer on the same terms, so that a buyer who wants the whole company can get it without every small holder's consent. It is standard in venture and private equity documents, and it is the clause that decides whether a fintech with forty angel investors and an option pool can be sold at all.
- [Dry Powder](https://mahouse.world/glossary/dry-powder.md): Dry powder is the capital that investors have committed to private equity, venture and credit funds but that the funds have not yet called or invested — money waiting for deals. It is measured in the trillions across the industry, rises when fundraising outpaces investing, and matters to sellers because it must be spent: a fund that does not deploy its capital within its investment period returns it, and no manager wants to.
- [Due Diligence](https://mahouse.world/glossary/due-diligence.md): Due diligence is the investigation a buyer or investor conducts before committing to a transaction, to verify what the seller has presented and to find what the seller has not: the financial position and quality of earnings, the legal and contractual state, the commercial prospects, the technology, the people, the tax history and, for a regulated fintech, the licence, the compliance controls and the regulator's file. Its findings shape the price, the structure and the warranties.
- [Duration](https://mahouse.world/glossary/duration.md): Duration measures how much a bond's price — or a portfolio's, or a bank's balance sheet's — changes when interest rates move: a duration of five means a one-point rise in rates cuts the value by about 5%. It is the weighted average time until the cash flows arrive, and it explains why long bonds fell 30% in 2022 and why Silicon Valley Bank failed.
- [Dynamic Currency Conversion](https://mahouse.world/glossary/dynamic-currency-conversion.md): Dynamic currency conversion is the offer, at a terminal, ATM or checkout, to charge a foreign cardholder in their card's home currency rather than the local one — converted on the spot at a rate the merchant's acquirer or DCC provider sets, with a margin of 3–7% shared between provider and merchant. The cardholder sees a familiar amount and pays more; the schemes and EU rules require the choice and the mark-up to be shown.
- [E-Money](https://mahouse.world/glossary/e-money.md): E-money is electronically stored monetary value, issued on receipt of funds and accepted as payment by parties other than the issuer — the balance in a prepaid card, a wallet or a business account at a non-bank. In the EU and UK it may be issued only by banks and authorised e-money institutions, and the funds behind it must be safeguarded, not lent.
- [EBITDA](https://mahouse.world/glossary/ebitda.md): EBITDA — earnings before interest, tax, depreciation and amortisation — is a company's operating profit with the financing decisions, the tax regime and the accounting allocations for past investment added back, so that businesses with different capital structures, tax positions and asset bases can be compared on operating performance; it is the number most private-company acquisitions are priced on, as a multiple. It is not cash flow: it ignores capital expenditure, working capital and debt service.
- [EMV](https://mahouse.world/glossary/emv.md): EMV is the global standard for chip cards and the terminals that read them — named for Europay, Mastercard and Visa and maintained by EMVCo — under which a card's chip generates a unique cryptogram for every transaction, so copied data cannot be replayed as a magnetic stripe could. Its liability-shift rules drove Europe's migration in the 2000s and the US's from 2015; contactless, tokenisation and 3DS 2 are its specifications too.
- [ESG Investing](https://mahouse.world/glossary/esg-investing.md): ESG investing is the practice of weighing environmental, social and governance factors alongside financial ones — screening out some businesses, favouring others, pressing companies on their practices — because these factors affect risk and return, and because investors' beneficiaries and regulators require it. In private markets it is a diligence workstream and a reporting duty; in fintech the S and G — conduct, data, governance — matter most.
- [EV/EBITDA Multiple](https://mahouse.world/glossary/ev-ebitda-multiple.md): The EV/EBITDA multiple is a company's enterprise value divided by its earnings before interest, tax, depreciation and amortisation — the number of years of operating cash earnings a buyer is paying for, before financing and accounting choices. It is the most used yardstick in private M&A because it compares businesses regardless of how they are funded or how they depreciate, and it is the multiple a fintech seller hears first.
- [Earn-Out](https://mahouse.world/glossary/earn-out.md): An earn-out is the part of an acquisition price that is paid later, and only if the business hits targets agreed at signing — usually revenue or EBITDA over one to three years. It bridges the gap between what the seller believes the business will do and what the buyer is prepared to pay for today.
- [Embedded Finance](https://mahouse.world/glossary/embedded-finance.md): Embedded finance is the offering of financial products — payments, accounts, lending, insurance, cards — inside a non-financial product at the point where the customer needs them: the loan in a software vendor's invoicing screen, the insurance sold with the ticket, the wallet inside the marketplace. The non-financial company distributes; a licensed provider supplies the regulated product through APIs; and the revenue is shared. It moved the distribution of finance into the software customers already use.
- [Emergent Behaviour](https://mahouse.world/glossary/emergent-behaviour.md): Emergent behaviour is a capability a model displays without having been explicitly trained for it, appearing — or appearing to appear — only above a certain scale: arithmetic, translation, code, multi-step reasoning showing up in large language models where smaller ones trained the same way showed none. It matters because a capability nobody designed is one nobody evaluated; and it is contested, because some apparent emergence is an artefact of measurement.
- [End-to-End Learning](https://mahouse.world/glossary/end-to-end-learning.md): End-to-end learning is training a single model to map raw input directly to the desired output — pixels to a document's fields, audio to a transcript, a transaction record to a fraud score — rather than building a pipeline of separately engineered stages, each with its own logic and errors. It trades the pipeline's interpretability and modularity for a model that optimises the whole task at once and often does better, provided there is enough labelled data at the ends.
- [Enhanced Due Diligence](https://mahouse.world/glossary/enhanced-due-diligence.md): Enhanced due diligence is the deeper level of customer checks the AML rules require where the risk is higher: politically exposed persons, customers from or transacting with high-risk countries, complex ownership structures, unusual business models, correspondent relationships — anything the firm's own risk assessment flags. It adds source-of-wealth and source-of-funds enquiries, senior-management approval, closer monitoring and more frequent review to the standard process, and it is where a fintech's compliance function earns its cost or fails its inspection.
- [Enterprise Management Incentives](https://mahouse.world/glossary/emi-regime.md): Enterprise Management Incentives are a UK tax-advantaged share option scheme for smaller, independent trading companies. Employees granted options at market value pay no income tax or National Insurance when they exercise, and capital gains tax when they sell, often at the Business Asset Disposal Relief rate. The scheme's company limits were raised substantially from 6 April 2026.
- [Enterprise Value](https://mahouse.world/glossary/enterprise-value.md): Enterprise value is the value of a company's operating business, independent of how it is financed: what a buyer would pay for the whole of it before settling its debt and taking its cash. It is the number a deal is priced on, and the numerator of the multiples — EV/EBITDA, EV/revenue — that acquisitions are compared by.
- [Equity Value](https://mahouse.world/glossary/equity-value.md): Equity value is the value of a company's shares — what its shareholders receive when it is sold, or what the market says they hold, in the case of a listed company. It is enterprise value plus cash and minus debt, and in a deal it is the price actually paid at completion after the bridge from the headline offer has been agreed.
- [Escrow](https://mahouse.world/glossary/escrow.md): Escrow is money from the purchase price held by an independent third party — a bank or a law firm — for an agreed period after completion, to be paid to the seller if no claim arises and to the buyer if one does. It gives a buyer's warranty and indemnity claims a fund to be paid from without suing the seller for it.
- [Exchange-Traded Fund](https://mahouse.world/glossary/exchange-traded-fund.md): An exchange-traded fund is an investment fund whose shares trade on a stock exchange throughout the day, usually tracking an index — a market, a sector, bonds, gold, since 2024 bitcoin — for fees of hundredths of a per cent. Creation and redemption by authorised participants keep its price near the value of its holdings. It is the building block of robo-advisers and the product that took the fund industry's fees away.
- [Exclusivity Period](https://mahouse.world/glossary/exclusivity-period.md): An exclusivity period is the time during which a seller agrees not to negotiate with, solicit or provide information to any other buyer, granted to one bidder so that it can spend on due diligence and drafting with some assurance the deal will not be sold from under it. It is the one binding clause in an otherwise non-binding letter of intent, and its length is a negotiation in itself.
- [Exit Strategy](https://mahouse.world/glossary/exit-strategy.md): An exit strategy is the plan for how a company's investors and founders will turn their shares into cash — a sale to a strategic buyer or a fund, a listing, a secondary sale of a stake, or a wind-down — and when. Every venture and private equity investment is made with an exit in view, because the fund must return money within its life; companies that plan the exit early are worth more when it comes.
- [Experience Economy](https://mahouse.world/glossary/experience-economy.md): The experience economy is the idea, from Pine and Gilmore's 1998 book, that customers choose and pay for the quality of the experience of a product, not only its features and price — and in financial services, where the products are largely commodities, the observation that the interface competes as hard as the rate: onboarding in minutes, a payment that needs no thought, a dispute resolved in the app. It is the basis of the neobanks' growth.
- [Expert System](https://mahouse.world/glossary/expert-system.md): An expert system is a program that encodes a human expert's knowledge as explicit rules — thousands of if-then statements over a knowledge base, run by an inference engine — to reach conclusions in a narrow domain. The dominant commercial AI of the 1980s, it fell from favour when the rules proved brittle and costly to maintain, and it lives on as the rules engines that run transaction monitoring, underwriting and compliance at every fintech.
- [Explainable AI](https://mahouse.world/glossary/explainable-ai.md): Explainable AI is the set of methods that make a model's outputs understandable to a person — inherently interpretable models (scorecards, decision trees), post-hoc explanations of complex ones (feature attributions such as SHAP, counterfactuals: "you would have been approved with €300 more income"), and the documentation around them. For a lender it is the law: a declined applicant must be told the principal reasons.
- [FX Margin](https://mahouse.world/glossary/fx-margin.md): The FX margin is the difference between the exchange rate a bank, card, remittance company or payment provider gives its customer and the wholesale (interbank or mid-market) rate at which it obtains the currency — expressed as a percentage, and usually larger than the fee the customer is shown. It is the main revenue of remittance and travel-money businesses, a meaningful line for neobanks and acquirers, and the reason "no fee" international payments are not free.
- [Factuality Evaluation](https://mahouse.world/glossary/factuality-evaluation.md): Factuality evaluation is the measurement of how often a language model's statements are true and supported, checked against known answers, source documents or trusted references by test sets, automated grading and human review. It turns "the model sometimes makes things up" into a rate that can be tracked, compared between models and used as a release threshold.
- [Failover](https://mahouse.world/glossary/failover.md): Failover is the automatic switching of payment traffic from a provider that is failing — an acquirer, a gateway, a processor, a 3-D Secure server — to a second one, so an outage on one side does not stop the checkout. It needs two integrations, tokens usable on both, health checks that decide when to switch, and a plan for the switch back; it is why large merchants run two PSPs.
- [Fairness Opinion](https://mahouse.world/glossary/fairness-opinion.md): A fairness opinion is a letter from an investment bank or valuation firm to a company's board stating that the price in a proposed transaction is fair, from a financial point of view, to the shareholders — backed by valuation work, but not stating a number. Boards obtain one before recommending a takeover or a sale to an insider, because it is what they hold up if the price is later challenged.
- [False Positives](https://mahouse.world/glossary/false-positives.md): A false positive is an alert or a decision that flags legitimate activity as suspicious: a transaction-monitoring alert on an ordinary payment, a sanctions hit on a namesake, a fraud rule that declines a real customer. In financial crime compliance most alerts are false positives, and the cost of working them is one of the largest operating costs a fintech carries.
- [Family Office](https://mahouse.world/glossary/family-office.md): A family office is a private company that manages the wealth of one wealthy family (single-family) or several (multi-family) — investments, tax, estate planning, often philanthropy — with no outside investors and no fund life. In fintech it is a direct investor in rounds and secondaries, a limited partner in funds, and sometimes the buyer of a whole company, with patience funds cannot offer.
- [Federated Learning](https://mahouse.world/glossary/federated-learning.md): Federated learning is a method of training a shared model across several parties — banks, hospitals, devices — without pooling their data: each party trains on its own records locally and sends only model updates to a coordinator, which combines them into a global model that is sent back; the records never leave their owner. In finance it is the candidate for cross-institution fraud and money-laundering models where sharing data is impossible; its limits are leakage and consortium governance.
- [Financial Due Diligence](https://mahouse.world/glossary/financial-due-diligence.md): Financial due diligence is the buyer's investigation of a target's financial position and performance before it commits: the quality of its earnings, the reliability of its accounts, its working capital and cash conversion, its net debt and off-balance-sheet obligations, and the credibility of its forecasts. It produces the numbers the price, the adjustments and the warranties are built on, and in a fintech it extends into the regulatory balance sheet the accounts do not show.
- [Fine-Tuning](https://mahouse.world/glossary/fine-tuning.md): Fine-tuning is continuing to train a pre-trained model on a smaller, task-specific dataset so that it adapts to a domain, a format, a style or a task — a language model on a company's ticket categories, a vision model on its document types — while keeping the general capability pre-training gave it. Parameter-efficient methods (LoRA) train a few million added weights instead of billions; the choice against prompting or retrieval turns on consistency, volume and how narrow the task is.
- [Fintech](https://mahouse.world/glossary/fintech.md): Fintech is financial services delivered through technology by firms that compete with, supply or sit inside the incumbent banks, insurers and asset managers — payments, lending, banking, wealth, insurance, infrastructure and compliance software — and the sector those firms make up, whose acquisitions run to tens of billions a year. Most fintechs are licensed, or operate on a partner's licence, and are regulated as the financial institutions they are; the technology is the delivery, the product is money.
- [Follow-On Investment](https://mahouse.world/glossary/follow-on-investment.md): A follow-on investment is money an existing investor puts into a company's later round — exercising its pro-rata right to keep its stake, or investing more to increase it — from reserves the fund set aside when it first invested. Follow-ons are how venture funds concentrate capital in their winners, how founders read which investors believe in them, and what a new lead investor watches before it commits.
- [Forward Propagation](https://mahouse.world/glossary/forward-propagation.md): Forward propagation is the computation that passes an input through a neural network's layers in order — each layer multiplying by its weights, adding biases and applying an activation function — to produce the output; it is the operation performed on every prediction and every generated token, and, during training, the first half of each step before backpropagation adjusts the weights. Its cost, multiplied by the number of requests, is the inference bill a fintech pays on every transaction.
- [Foundation Model](https://mahouse.world/glossary/foundation-model.md): A foundation model is a large machine-learning model trained on broad data at scale — text, code, images — that can be adapted to many tasks it was not specifically trained for, by prompting or fine-tuning. Large language models are the best-known kind, and the companies that build them are the most expensive acquisition targets, and acquirers, in technology.
- [Free Cash Flow](https://mahouse.world/glossary/free-cash-flow.md): Free cash flow is the cash a business generates after paying for its operations, its taxes and the investment it needs to keep running — the cash left over for its providers of capital. Unlevered free cash flow is measured before interest and is what a discounted cash flow valuation discounts; levered free cash flow is what remains for shareholders after debt service.
- [Frictionless Flow](https://mahouse.world/glossary/frictionless-flow.md): The frictionless flow is the path in 3-D Secure 2 where the issuer authenticates a transaction on the data the merchant sent — device, history, addresses, amount — without challenging the cardholder: the customer sees nothing, the merchant gets an authentication value and, if the issuer authenticated rather than merely accepted an exemption, the liability shift. It is what the protocol was rebuilt to allow, and a merchant's frictionless rate is the measure of how well it has done so.
- [Friendly Fraud](https://mahouse.world/glossary/friendly-fraud.md): Friendly fraud — the schemes now say first-party misuse — is a chargeback raised by the genuine cardholder on a purchase they made: they did not recognise the descriptor, a family member used the card, or a refund by chargeback was easier than by the merchant. It is the largest category of card disputes by most estimates, and the merchant loses by default unless it can show the cardholder was there.
- [Fund of Funds](https://mahouse.world/glossary/fund-of-funds.md): A fund of funds is an investment fund that commits its capital to other funds — a portfolio of ten to thirty private equity or venture funds across managers, strategies and vintages — rather than to companies directly. It gives smaller investors diversification and access to managers they could not reach alone, at the cost of a second layer of fees; it is how many pensions first invested in venture.
- [GENIUS Act](https://mahouse.world/glossary/genius-act.md): The GENIUS Act — the Guiding and Establishing National Innovation for US Stablecoins Act, signed into law in July 2025 — is the first US federal framework for payment stablecoins: it restricts issuance to licensed issuers, requires reserves of at least one dollar in cash, insured deposits or short-term Treasuries for every token, mandates monthly reserve disclosure, prohibits paying interest to holders, and applies anti-money-laundering obligations to issuers. It gave the dollar stablecoin the status MiCA gave the euro's.
- [GNN](https://mahouse.world/glossary/gnn.md): A graph neural network (GNN) is a model that learns over data structured as a graph — entities as nodes, relationships as edges — by passing information along the edges so that each node's representation reflects its neighbours and their neighbours, and using that structure as a signal; a customer who shares a device with three accounts linked to a known fraudster is suspicious in a way no feature about the customer alone reveals. GNNs find fraud rings.
- [General Partner](https://mahouse.world/glossary/general-partner.md): A general partner is the manager of a private equity or venture fund organised as a limited partnership: the entity that raises the fund, makes and manages the investments, is liable for the partnership's obligations, and is paid a management fee and a share of the profits. The limited partners supply the money and the general partner supplies the judgement; the partnership agreement between them is the constitution of the fund.
- [Generative AI](https://mahouse.world/glossary/generative-ai.md): Generative AI is artificial intelligence that produces new content — text, images, audio, video, code, structured data — rather than only classifying, scoring or predicting from what it is given: large language models generate text and code, diffusion models generate images and video. It became a mass technology with ChatGPT in late 2022, reached fintech as assistants, document processing and code generation — and, from the other side, as the synthetic documents, faces and voices fraud teams now defend against.
- [Generative Adversarial Network](https://mahouse.world/glossary/generative-adversarial-network.md): A generative adversarial network is a pair of neural networks trained in competition: a generator that produces synthetic data from random input and a discriminator that tries to tell synthetic from real, each improving against the other until the generator's output is hard to distinguish from real data. Introduced in 2014, GANs produced the first photorealistic synthetic faces and the first deepfakes, led image generation until diffusion models overtook them around 2022, and remain in use for synthetic data.
- [Golden Parachute](https://mahouse.world/glossary/golden-parachute.md): A golden parachute is a contractual payment to a senior executive if the company is acquired and the executive loses their job or their role changes — typically a multiple of salary and bonus, accelerated vesting of equity, and continued benefits. It is written in advance so that executives negotiating a sale are not negotiating against their own livelihood, and a buyer reads it as a cost of the deal and a signal of who will not be staying.
- [Goodwill](https://mahouse.world/glossary/goodwill.md): Goodwill is the part of an acquisition price that cannot be attributed to any identifiable asset — the premium a buyer pays for the assembled business, its future customers and the synergies it expects — recorded as an asset on the buyer's balance sheet after the purchase price allocation and tested each year for impairment.
- [Gradient Descent](https://mahouse.world/glossary/gradient-descent.md): Gradient descent is the optimisation method by which nearly every machine-learning model is trained: compute the gradient of the loss function — the direction in which error increases fastest — and move the parameters a small step the other way; repeat until the loss stops falling. Stochastic gradient descent estimates the gradient from a small batch rather than the whole dataset, which makes training on billions of examples possible. The step size, the learning rate, matters most.
- [Gross Payment Volume](https://mahouse.world/glossary/gross-payment-volume.md): Gross payment volume is the total value of transactions a payments company processes in a period — every card, wallet and transfer that passes through its systems, before fees — and it is the number the industry leads with: Stripe's trillion, Adyen's billions, a facilitator's "$50bn GPV". It is a measure of scale and reach, not of revenue: the company keeps a take rate on it, and a buyer values the take rate's durability, not the volume's size.
- [Growth Equity](https://mahouse.world/glossary/growth-equity.md): Growth equity is investment in established, growing companies that need capital to expand faster than their cash flows allow — a minority stake, usually without leverage, in a business that has proved its model but is not ready for a buy-out or a listing. It sits between venture and private equity, and in fintech it is the money behind Series C-to-E rounds scaling into new markets.
- [Guardrails](https://mahouse.world/glossary/guardrails.md): Guardrails are the controls placed around an AI system — rather than inside the model — that constrain what it is permitted to do or say: filters on inputs, checks on outputs (grounding, format, policy, figures verified against a source), limits on actions (which tools the system may call, under whose approval), and the monitoring and kill switches around all of it. They exist because a model's training cannot be relied on for a company's rules.
- [Guest Checkout](https://mahouse.world/glossary/guest-checkout.md): Guest checkout lets a shopper complete a purchase without creating an account — entering delivery and payment details for this order only, with the option to register afterwards. Forced registration is among the most cited reasons for abandonment; guest checkout removes it at the cost of the account data and saved card an account would hold — a trade wallets and network tokens have narrowed.
- [Hallucination](https://mahouse.world/glossary/hallucination.md): A hallucination is an output from a language model that is false, invented or unsupported — a citation that does not exist, a fee the product does not charge, a regulation misquoted — delivered in the same confident register as a correct answer. It is a property of how the models work, reduced by grounding in retrieved documents and constraining what the model may assert; for a fintech it is why a customer-facing model needs a source and a human.
- [Hard Decline](https://mahouse.world/glossary/hard-decline.md): A hard decline is an issuer's refusal of a payment for a reason that will not change if the merchant tries again — the card is reported lost or stolen, the account is closed, the number is invalid, the transaction type is not permitted — as opposed to a soft decline, where the reason is temporary. The schemes prohibit retrying hard declines; the right response is to stop and, for a stored card, ask for another.
- [Hardware Security Module](https://mahouse.world/glossary/hardware-security-module.md): A hardware security module is a tamper-resistant physical device that generates, stores and uses cryptographic keys without ever exposing them — the keys that verify PINs, validate EMV cryptograms, encrypt card data in transit, sign tokens and protect the certificates a payment system runs on. Every issuer, acquirer, processor and scheme runs them; PCI standards require them for PIN and key management; and cloud HSM services have made them a subscription rather than a rack.
- [Heads of Terms](https://mahouse.world/glossary/heads-of-terms.md): Heads of terms are a short document recording the main points a buyer and a seller have agreed before the lawyers draft the full sale agreement — price, structure, conditions, timetable, exclusivity — mostly non-binding, with the exclusivity and confidentiality clauses binding. It is the UK and European name for what a letter of intent is in the US, and the point at which a negotiation becomes a transaction.
- [Hedge Fund](https://mahouse.world/glossary/hedge-fund.md): A hedge fund is a privately offered fund that trades liquid markets — equities, credit, currencies, derivatives — with fewer constraints than a mutual fund: it can sell short, use leverage, concentrate, and charge a performance fee. It is not private equity, though the two overlap — crossover funds in late-stage fintech rounds, arbitrage funds in takeovers, credit funds in a fintech's debt.
- [Hidden Layer](https://mahouse.world/glossary/hidden-layer.md): A hidden layer is any layer of a neural network between the input layer, which receives the data, and the output layer, which produces the prediction; it is where the network forms the intermediate representations it computes with — edges and shapes in a vision model, syntax in a language model, risk patterns in a fraud model. "Deep" learning means many hidden layers. No one specifies what they represent; training decides, the source of their power and their opacity.
- [Horizontal Merger](https://mahouse.world/glossary/horizontal-merger.md): A horizontal merger combines two companies that compete in the same market at the same level — two payment processors, two neobanks, two lenders to the same customers — so that the combined firm has more share, lower unit costs and, if the regulator allows, more pricing power. It is the merger with the largest cost synergies and the most competition scrutiny, and it is the shape of consolidation in every commoditising fintech segment.
- [Hosted Payment Page](https://mahouse.world/glossary/hosted-payment-page.md): A hosted payment page is a checkout page — or, in its modern form, embedded fields or an iframe within the merchant's page — served by the payment provider rather than the merchant, so the card number is entered into the provider's domain and never touches the merchant's systems. It takes the merchant almost entirely out of PCI DSS scope, once at the cost of a redirect, which hosted fields removed.
- [Hostile Takeover](https://mahouse.world/glossary/hostile-takeover.md): A hostile takeover is an attempt to acquire a listed company against the wishes of its board, by going directly to the shareholders with a tender offer or a proxy fight rather than negotiating an agreed deal. It is possible because a listed company's shareholders can sell to whom they like; it is rare in fintech because a licensed company's value sits in people, relationships and regulatory approvals that a hostile buyer cannot inspect and may not keep.
- [Hurdle Rate](https://mahouse.world/glossary/hurdle-rate.md): A hurdle rate is the minimum return a fund must deliver to its investors — typically 8% a year — before its manager receives any carried interest; the same term names the return a company requires before approving an investment or an acquisition. In the fund it protects investors from paying for returns they could have earned elsewhere; in the company it is the cost of capital with a margin, and the number an acquisition case must clear.
- [Hybrid and Multi-Cloud](https://mahouse.world/glossary/hybrid-and-multi-cloud.md): Hybrid cloud combines an organisation's own infrastructure or private cloud with one or more public clouds, run as a connected whole; multi-cloud means using public cloud services from two or more providers. Financial firms adopt one or both to limit dependence on a single provider, meet data-location rules and keep a credible exit route open.
- [Hyperparameter Tuning](https://mahouse.world/glossary/hyperparameter-tuning.md): Hyperparameter tuning is the search for the settings that govern how a model is trained or shaped but are not learned from data — learning rate, batch size, layers, regularisation strength, tree depth — by training candidates with different settings and comparing them on validation data. Grid, random and Bayesian search are the methods; the validation set is the judge; and the common failure is tuning against the data then used to report the result, which flatters the model.
- [IPO Window](https://mahouse.world/glossary/ipo-window.md): An IPO window is a period in which market conditions let companies sell shares to the public at a price their owners will accept: stable or rising equity markets, low volatility, recent listings trading above their offer price, and fund managers with appetite for new issues. Windows open and shut with markets, not with any one company's readiness.
- [Idempotency Key](https://mahouse.world/glossary/idempotency-key.md): An idempotency key is a unique value a client attaches to a payment request so that if the request is sent again — because the network dropped the response, the client timed out, a retry fired — the server recognises it as the same operation and returns the original result instead of performing it twice. It is the mechanism that keeps a slow response from becoming a double charge, and every payments API worth integrating supports it.
- [Identity Verification](https://mahouse.world/glossary/identity-verification.md): Identity verification is the process of establishing that a person is who they claim to be — checking a document, matching a face to it, testing that the face is live, and confirming the details against independent records. It is the first step of know-your-customer, the control that decides who gets an account, and a market whose vendors have been bought by the payments and credit-data groups at nine-figure prices.
- [Impact Investing](https://mahouse.world/glossary/impact-investing.md): Impact investing is investment made to produce a measurable social or environmental benefit alongside a financial return — investors back companies for the outcome they deliver, measure it and report it, at returns from concessional to market-rate. ESG screens ordinary investments for risk; an impact investor chooses the investment for the impact. In fintech the thesis is financial inclusion.
- [Index Fund](https://mahouse.world/glossary/index-fund.md): An index fund holds the securities of an index — the S&P 500, a global benchmark, a bond universe — in the index's proportions, earning the market's return less a fee of a few hundredths of a per cent, instead of paying a manager to try to beat the market and usually fail. By 2024 index funds and ETFs held more US equity fund assets than active managers, and their fee compression is the ground wealthtech stands on.
- [Indicative Offer](https://mahouse.world/glossary/indicative-offer.md): An indicative offer is a buyer's non-binding statement of what it would pay and on what terms, made after reading the information memorandum and before due diligence — usually a valuation range, the form of consideration, the funding, the conditions and the timetable. In an auction it is the first round: the seller uses the indicative offers to decide which bidders see the data room.
- [Inference](https://mahouse.world/glossary/inference.md): Inference is the use of a trained model to produce an output for a new input — a fraud score for a transaction, a reply to a customer's message, a document's summary — as opposed to training, which produced the model. It is where AI meets production: served behind an API, measured in latency and cost per query, and at scale the larger part of the compute bill, since a model is trained once and queried millions of times.
- [Information Memorandum](https://mahouse.world/glossary/information-memorandum.md): An information memorandum is the document a seller's adviser writes to describe a business to potential buyers — its market, products, customers, team, financials and the reasons to buy it — sent under a non-disclosure agreement to invite indicative offers. It is a sales document, and a buyer reads it as one.
- [Initial Public Offering](https://mahouse.world/glossary/initial-public-offering.md): An initial public offering (IPO) is the first sale of a company's shares to public investors, after which the shares trade on an exchange: the company (and often existing holders) sells shares through underwriters who market them to institutions, build a book of demand, set the price and allocate; the company becomes subject to listing rules and continuous disclosure, and its founders and investors gain a market for their shares after a lock-up.
- [Instant Payout](https://mahouse.world/glossary/instant-payout.md): An instant payout is the transfer of funds to a merchant, marketplace seller, gig worker or claimant within minutes of the request rather than on the standard cycle — over instant bank rails (Faster Payments, SEPA Instant, RTP, Pix, UPI) or by pushing funds to a debit card through Visa Direct or Mastercard Send. Platforms sell it as a feature for a fee, and its economics rest on what the recipient will pay for speed.
- [Instruction Tuning](https://mahouse.world/glossary/instruction-tuning.md): Instruction tuning is the supervised fine-tuning of a pre-trained language model on a dataset of instructions paired with good responses — "summarise this contract" with a summary, "classify this ticket" with the class — so that the model, which pre-training taught only to continue text, learns to follow directions and adopt a format. It is the first stage of post-training, usually followed by preference training, and it separates a base model from a chat model.
- [InsurTech](https://mahouse.world/glossary/insurtech.md): InsurTech is technology applied to insurance — distribution, underwriting, pricing, policy administration and claims — by new entrants and by incumbents rebuilding those functions: digital brokers and comparison platforms, full-stack digital insurers with their own licences, managing general agents writing on a carrier's capital, embedded insurance sold inside other products, and software and data vendors serving all of them. The 2021 IPO cohort's collapse taught the sector that software does not change loss ratios by itself.
- [Interchange Fee](https://mahouse.world/glossary/interchange-fee.md): An interchange fee is the fee the merchant's acquirer pays the cardholder's issuing bank on every card transaction, set by the card network as a percentage of the amount plus a fixed sum, to compensate the issuer for the credit risk, fraud and cost of running the card. It is the largest component of what a merchant pays to accept cards, and the main revenue of an issuing fintech.
- [Interchange-Plus Pricing](https://mahouse.world/glossary/interchange-plus-pricing.md): Interchange-plus pricing charges a merchant the actual interchange fee and scheme fees on each transaction, passed through at cost, plus the acquirer's own margin stated separately — so that the merchant can see what goes to the issuer, what goes to the network and what the acquirer keeps. It is the transparent alternative to a blended rate, the norm for larger merchants, and the pricing model that made acquiring margins visible and therefore competitive.
- [Internal Rate of Return](https://mahouse.world/glossary/internal-rate-of-return.md): The internal rate of return is the annual rate at which a series of cash flows — money out, money back — discounts to zero: the compound return the investment actually earned, allowing for when each euro left and returned. It is the standard measure of private equity and venture performance, and the one most easily flattered, because a euro returned early counts for more than a euro returned late whatever the total.
- [Investment Thesis](https://mahouse.world/glossary/investment-thesis.md): An investment thesis is the argument for an investment — why this company, at this price, will be worth more at exit: what the market is doing, what the company has that others lack, the plan, what could go wrong, and how the money comes back. It is written as a memo for an investment committee, tested in diligence, and read again at exit to see whether the return came from its reasons or from luck.
- [Issuer](https://mahouse.world/glossary/issuer.md): An issuer is the bank or licensed institution that provides a card to a cardholder under a scheme's rules — opening the account, deciding each authorisation, extending credit or holding the deposit, taking the fraud and credit risk, and earning interchange on every purchase. It is the other side of every transaction from the acquirer, and the side a neobank sits on: a fintech issuing cards is an issuer with an issuer's obligations.
- [J-Curve](https://mahouse.world/glossary/j-curve.md): The J-curve is the shape of a private equity or venture fund's cumulative return over its life: negative in the early years, as fees are paid and capital is called before any company is sold, then rising through zero as exits arrive and, if the fund works, well above it. It is why a fund's reported return in year three means little, why investors commit across vintages to smooth it, and why subscription credit lines were invented to flatten it.
- [KYC](https://mahouse.world/glossary/kyc.md): Know your customer (KYC) is the process by which a financial institution establishes and verifies who a customer is — their identity, and for a business its ownership and controllers — before providing a service, and keeps that knowledge current throughout the relationship; it is the first component of the anti-money laundering regime and the gate every fraud attempt must pass. In practice it is documents, biometrics, database checks and screening at onboarding, risk-rated by customer and product.
- [Know Your Agent (KYA)](https://mahouse.world/glossary/know-your-agent-kya.md): Know your agent (KYA) is the set of checks that establishes, before an AI agent may buy or move money, which agent it is, who operates it, which person or business it acts for, and what that principal has authorised it to do. It applies know-your-customer logic to software acting on someone's behalf, using cryptographic credentials rather than documents.
- [Know Your Business](https://mahouse.world/glossary/know-your-business.md): Know your business is customer due diligence applied to a company: verifying that the business exists and is what it claims, identifying its directors, tracing its ownership to the natural persons who control it, understanding what it does, and screening all of them. It is harder than verifying a person — companies own companies across borders — and it is the gate every B2B fintech and payment facilitator onboards through.
- [Large Language Model](https://mahouse.world/glossary/large-language-model.md): A large language model (LLM) is a neural network — a transformer with billions to trillions of parameters — trained on vast quantities of text to predict the next token, which turns out to be enough for it to write, summarise, translate, answer questions, reason and produce code; post-training turns the predictor into an assistant. GPT-4, Claude, Gemini and Llama are LLMs; for a fintech they power assistants and document processing, with hallucination and governance as the constraints.
- [Latent Space](https://mahouse.world/glossary/latent-space.md): Latent space is the compressed, learned representation inside a model in which each input — a word, an image, a transaction history, a customer — becomes a vector of numbers positioned so that similar inputs sit near each other and directions carry meaning. Embeddings are points in it; search by meaning is nearest-neighbour search in it; generative models decode outputs from points in it; and arithmetic on positions works because the model organised the space by its training data's regularities.
- [Legal Due Diligence](https://mahouse.world/glossary/legal-due-diligence.md): Legal due diligence is the buyer's lawyers' review of the target's legal position: its corporate records and share ownership, its material contracts and their change-of-control clauses, its licences and regulatory history, its employees and their terms, its intellectual property and its data protection, its litigation and disputes. It tells the buyer what it is legally acquiring, which consents it needs, and what the warranties and indemnities must cover.
- [Letter of Intent](https://mahouse.world/glossary/letter-of-intent.md): A letter of intent is a short document in which a buyer sets out the price, structure and main conditions on which it proposes to acquire a company, signed before due diligence begins. It is not binding on the deal itself; only its confidentiality, exclusivity and cost clauses usually bind the parties.
- [Leverage Ratio](https://mahouse.world/glossary/leverage-ratio.md): A leverage ratio measures how much debt a company carries against its capacity to service it — net debt divided by EBITDA for an operating company; capital divided by total assets for a bank, the regulator's simplest test. It is the number that sizes a buy-out's debt, sets the covenants, and tells a buyer how much of a target's enterprise value the lenders already own.
- [Leveraged Buy-Out](https://mahouse.world/glossary/leveraged-buy-out.md): A leveraged buy-out is the acquisition of a company mostly with borrowed money, secured on the company's own assets and repaid from its own cash flows, so that the buyer — usually a private equity fund — puts in a fraction of the price as equity and earns a multiple on it if the debt is paid down and the business grows. It is the private equity model, and it suits the steady cash flows of payments and software fintechs.
- [Limited Partner](https://mahouse.world/glossary/limited-partner.md): A limited partner is an investor in a private fund organised as a limited partnership — a pension fund, insurer, endowment, sovereign fund, family office or fund of funds — who commits capital, funds it when called, receives distributions, and bears no liability beyond its commitment and takes no part in management. The general partner runs the fund; the limited partners own it, and their willingness to commit to the next one is the discipline that the industry runs on.
- [Liquidation Preference](https://mahouse.world/glossary/liquidation-preference.md): A liquidation preference is the right of an investor's preferred shares to be paid a set amount — usually the money invested, sometimes a multiple of it — before ordinary shareholders receive anything when the company is sold or wound up. It decides who gets what at an exit that is smaller than everyone hoped, which is the exit most fintechs have.
- [Liquidity Discount](https://mahouse.world/glossary/liquidity-discount.md): A liquidity discount is the reduction applied to the value of shares that cannot be readily sold — a private company's, a minority stake's, shares under a lock-up — against what identical shares would fetch in a liquid market. It is why a private fintech is valued below its listed comparables, why secondary sales price below the last round, and why a minority holder owns less than their percentage suggests.
- [Lock-Up Period](https://mahouse.world/glossary/lock-up-period.md): A lock-up period is the time after an IPO — usually 180 days — during which insiders and pre-IPO shareholders agree not to sell, so the market absorbs the offering before it absorbs them. The same term covers shares received as consideration in an acquisition, locked for a year or two to keep the seller aligned. At expiry the locked shares can be sold, and the price often says so.
- [Locked Box](https://mahouse.world/glossary/locked-box.md): A locked box is a pricing mechanism in which the equity price of a company is fixed at signing by reference to a balance sheet dated before the deal, with the seller promising that no value has left the company since that date except as agreed. The alternative, completion accounts, sets the price after completion instead.
- [Loss Function](https://mahouse.world/glossary/loss-function.md): A loss function is the measure of how wrong a model's predictions are on its training examples — mean squared error for a number, cross-entropy for a probability, and many task-specific variants — and the quantity training exists to minimise: gradient descent moves the parameters to reduce it, and the resulting model is the one that scores lowest on it. The loss decides what "better" means to training; one that ignores the business cost of errors trains the wrong model.
- [M&A](https://mahouse.world/glossary/m-a.md): Mergers and acquisitions (M&A) are the transactions in which companies combine — a merger — or one buys control of another — an acquisition — through the purchase of shares or assets, for cash, shares or both, negotiated between the parties or, for public companies, through offers to shareholders. Companies do them to grow faster than they can organically, to buy capabilities, customers or licences, to consolidate an industry, or to sell what they no longer want.
- [MLOps](https://mahouse.world/glossary/mlops.md): MLOps — machine learning operations — is the engineering practice of taking models from development into production and keeping them working there: reproducible training pipelines, model registries and versioning, deployment and serving, monitoring of inputs, outputs and performance, drift detection, retraining, rollback, and the documentation that ties each version to its data, code and evaluation. In a regulated fintech it is the operational half of model-risk management; LLMOps is the same discipline for prompts, retrieval and rented models.
- [Machine Learning](https://mahouse.world/glossary/machine-learning.md): Machine learning (ML) is the discipline of building systems that learn to perform a task from data — finding the patterns that connect inputs to outcomes and using them to predict, classify, rank or detect — rather than being programmed with explicit rules for the task. Supervised learning learns from labelled examples, unsupervised learning finds structure without labels, reinforcement learning learns from rewards. It is the working core of fintech's fraud, credit and identity systems.
- [Management Buy-Out](https://mahouse.world/glossary/management-buy-out.md): A management buy-out is the purchase of a company by the team that runs it, usually with a private equity fund providing most of the equity and banks providing debt, so that managers who owned little or nothing become meaningful shareholders. It is how founders exit without selling to a competitor, how corporates dispose of units nobody else understands, and — in fintech — how a licensed business changes hands without changing the people the regulator approved.
- [Market Power](https://mahouse.world/glossary/market-power.md): Market power is a firm's ability to raise prices, lower quality or exclude rivals without losing enough customers to make it unprofitable — the condition competition law exists to police. In a merger review it is the question the regulator asks of the combined firm, and in fintech it is why the Visa–Plaid acquisition was abandoned and why every large payments deal now waits on Washington, Brussels and London.
- [Material Adverse Change](https://mahouse.world/glossary/material-adverse-change.md): A material adverse change clause lets a buyer withdraw from a signed acquisition if, between signing and completion, something happens that seriously and durably damages the target's business — with the definition of "material", and the list of things that do not count, negotiated line by line. In practice the clause is invoked far more often than it is upheld.
- [Merchant Account](https://mahouse.world/glossary/merchant-account.md): A merchant account is the account an acquirer opens for a business to accept card payments: the agreement under which the acquirer processes the business's transactions, settles the funds to its bank account, charges its fees, holds reserves, and is liable to the schemes for its chargebacks. It is not a bank account but a contract with underwriting behind it; a business either has its own, or accepts cards as a sub-merchant under a payment facilitator's.
- [Merchant Discount Rate](https://mahouse.world/glossary/merchant-discount-rate.md): The merchant discount rate is the total fee a merchant pays its acquirer or PSP on a card transaction, as a percentage of the amount — the interchange that goes to the issuer, the scheme fees that go to the network, and the acquirer's own margin. "Discount" because the merchant is settled the sale less this amount; "merchant service charge" in UK usage. It is the number the industry's margins hide inside.
- [Merchant Identification Number](https://mahouse.world/glossary/merchant-identification-number.md): A merchant identification number is the unique code an acquirer assigns to a merchant account, sent with every transaction so that the schemes, the issuer and the acquirer know which merchant it belongs to — for routing, settlement, fees, disputes and the monitoring programmes that count chargebacks per MID. A business may hold several — one per location, channel, currency or brand — and a payment facilitator holds one master MID with sub-merchant identifiers beneath it.
- [Merchant Underwriting](https://mahouse.world/glossary/merchant-underwriting.md): Merchant underwriting is the acquirer's or payment facilitator's assessment of a merchant before it may accept cards: what it sells, to whom, how far ahead of delivery it takes money, its finances, its owners, its history — to decide whether to accept it, at what price, and with what reserve. The acquirer is extending a credit line equal to the merchant's undelivered sales, and underwriting sizes that exposure.
- [Merger](https://mahouse.world/glossary/merger.md): A merger is the combination of two companies into one, either by one absorbing the other or by both transferring into a new entity, with the shareholders of each receiving shares in the combined business. In law most "mergers" are acquisitions in which the price is paid in shares; a true merger of equals, with neither side in control, is rare.
- [Mezzanine Debt](https://mahouse.world/glossary/mezzanine-debt.md): Mezzanine debt is the layer between senior debt and equity: a subordinated loan that ranks behind the banks, is repaid later, charges a high rate — much of it rolled up — and often carries warrants over the equity. It fills the gap when senior lenders will not lend enough and the sponsor will not add equity, at a price that reflects being second in line.
- [MiCA](https://mahouse.world/glossary/mica.md): MiCA — the Markets in Crypto-Assets Regulation, Regulation (EU) 2023/1114 — is the EU's single rulebook for crypto-assets not already covered by financial-services law. It licenses crypto-asset service providers with a passport across the EU, sets authorisation, reserve and redemption rules for stablecoin issuers, and requires white papers and market-abuse controls for other tokens.
- [Minting](https://mahouse.world/glossary/minting.md): Minting is the creation of new tokens on a blockchain by calling the function of a smart contract that issues them — a stablecoin issuer minting a million tokens against a million dollars received, a fund tokenising shares, a protocol issuing its governance token. Burning is the reverse: tokens destroyed on redemption. For a regulated issuer the mint-and-burn cycle is how supply matches reserves, and the controls over who may mint, on what evidence, are what an auditor examines.
- [Misalignment](https://mahouse.world/glossary/misalignment.md): Misalignment is the gap between what an AI system was built to do and what it actually optimises for — a model or agent pursuing the goal it was given in a way its makers did not intend. In a business that deploys AI on customers or money it is an operational risk with a regulatory cost, and in an acquisition of an AI company it is a diligence question.
- [Mixture of Experts](https://mahouse.world/glossary/mixture-of-experts.md): A mixture of experts is a neural-network architecture in which certain layers are replaced by several parallel sub-networks — the experts — and a small router chooses, for each input token, which few experts process it; the total parameter count grows with the number of experts while the computation per token stays that of the few that ran. Mixtral, DeepSeek-V3 and Gemini use it; a model can have hundreds of billions of parameters and run like a small one.
- [Mobile POS](https://mahouse.world/glossary/mobile-pos.md): Mobile POS is card acceptance on a phone or tablet — originally a small reader plugged in or paired by Bluetooth, now increasingly the phone's own NFC — combined with a payment-facilitator model that onboards the merchant in minutes at a flat rate. Square's 2010 reader made it a category; iZettle, SumUp and their peers spread it across Europe; and it is how tens of millions of small merchants first accepted cards.
- [Model Context Protocol](https://mahouse.world/glossary/model-context-protocol.md): The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in November 2024, that defines how AI applications connect to external tools and data. A system exposes its functions and content once, as an MCP server, and any compatible AI client can discover and use them, replacing a custom connector between every model and every system.
- [Model Parameters](https://mahouse.world/glossary/model-parameters.md): Model parameters are the learned numbers inside a neural network — the weights that multiply inputs and the biases added to them — set by training and fixed at inference; their count is the customary, if crude, measure of a model's size, from millions in a fraud model to hundreds of billions in a frontier language model. More parameters mean more capacity, more data and compute to train, more memory to run, and, past what the data supports, overfitting.
- [Multimodal Model](https://mahouse.world/glossary/multimodal-model.md): A multimodal model is an AI model that works across more than one type of data — reading images and text together, hearing audio, watching video, and producing text, images or speech in return — rather than one modality alone. GPT-4V, Gemini and the open-weight vision-language models joined a vision encoder to a language model; for a fintech that means a model that reads a bank statement, a receipt, an ID document or a customer's screenshot, and answers in words.
- [Multiple on Invested Capital](https://mahouse.world/glossary/multiple-on-invested-capital.md): The multiple on invested capital is the total value returned or held divided by the capital invested — a 2.5× means every euro in became two and a half. It ignores time, which is its limit and its honesty: a multiple cannot be flattered by a subscription line or an early exit, so investors read it beside the IRR. Its variants split the multiple into cash already back (DPI), value still held (RVPI) and the two together (TVPI).
- [NLP](https://mahouse.world/glossary/nlp.md): Natural language processing (NLP) is the field of getting software to read, interpret and produce human language — classifying documents, extracting entities and figures, answering questions, translating, summarising, detecting sentiment, and conversing. It moved from hand-written rules through statistical methods to transformer models that learned language from text at scale, and the large language models absorbed most of its tasks into one model. For a fintech it is document processing, support automation and compliance screening.
- [Near Field Communication](https://mahouse.world/glossary/near-field-communication.md): Near-field communication is the short-range wireless standard — a few centimetres, a fraction of a second — through which a contactless card or phone exchanges payment data with a terminal, and through which a phone can act as a terminal itself. It is the physical layer under every tap; EMV is the logic above it. Who controls a phone's NFC chip controls who can build a wallet on it, which is why Apple's was opened in 2024.
- [Neobank](https://mahouse.world/glossary/neobank.md): A neobank is a bank built for digital delivery from the start — an app, no branches, onboarding in minutes — operating either on its own banking or e-money licence or on a partner institution's, offering accounts, cards, payments and, as it matures, lending, savings and investment. The 2010s produced hundreds; by the mid-2020s a handful had reached scale and profit, many had been acquired or merged, and many had closed; the survivors' economics rest on lending and deposits.
- [Net Asset Value](https://mahouse.world/glossary/net-asset-value.md): Net asset value is a fund's assets less its liabilities — for a mutual fund or ETF, computed daily from market prices and divided by the units in issue to give the dealing price; for a private equity or venture fund, computed quarterly from the manager's valuation of unlisted companies. The public NAV is a fact; the private NAV is an estimate, and the secondary market's discount to it is the market's opinion of the estimate.
- [Net Settlement](https://mahouse.world/glossary/net-settlement.md): Net settlement is the settlement of many payments between institutions by adding up what each owes and is owed over a period and transferring only the net balance — once or a few times a day, across accounts at a settlement bank or the central bank — instead of moving every payment. Card schemes and clearing houses settle net; it needs far less liquidity than gross and carries the risk that a participant fails in the gap.
- [Network Token](https://mahouse.world/glossary/network-token.md): A network token is a substitute for a card number issued by the card scheme itself — Visa Token Service, Mastercard's MDES — bound to one merchant, wallet or device, so a stored credential is never the real number, is useless if stolen elsewhere, is updated automatically when the card is replaced, and is approved by issuers at higher rates. It is the technology behind Apple Pay and, since 2020, the standard for cards on file.
- [Non-Disclosure Agreement](https://mahouse.world/glossary/non-disclosure-agreement.md): A non-disclosure agreement is a contract under which a party receiving confidential information agrees to use it only for a stated purpose — evaluating a possible transaction — and not to disclose it. In a sale process it is signed before the information memorandum is sent, and it is the first binding document of the deal.
- [OCC-Regulated](https://mahouse.world/glossary/occ-regulated.md): OCC-regulated describes a US financial institution chartered and supervised by the Office of the Comptroller of the Currency, the Treasury bureau responsible for national banks, federal savings associations and the federal branches of foreign banks. For a fintech it usually means holding, or depending on, a federal charter that allows nationwide operation under one federal supervisor instead of state-by-state licensing.
- [Objective Function](https://mahouse.world/glossary/objective-function.md): An objective function is the quantity a model is trained to maximise or minimise — the loss in supervised learning, the reward in reinforcement learning, the preference score in RLHF — and therefore the only definition of "better" the training process has: whatever the objective rewards, the model becomes good at, including things nobody intended. The gap between the objective written down and the outcome wanted is the source of reward hacking and of the alignment problem at every scale.
- [Observability](https://mahouse.world/glossary/observability.md): Observability is the degree to which the internal state of a running system can be worked out from the data it emits — in software, its metrics, logs and traces. For AI it adds prompts, responses, token counts, cost, tool calls and quality scores, showing not only that a system is up but whether its answers are right.
- [One-Click Checkout](https://mahouse.world/glossary/one-click-checkout.md): One-click checkout lets a returning shopper complete a purchase with a single confirmation, using payment and delivery details stored from a previous order — a network token, a saved address, a tap or a fingerprint. Amazon patented the pattern in 1999 and it expired in 2017; wallets, tokens and checkout networks then made it available to every merchant, and Europe's SCA rules made it a question of exemptions.
- [Open Finance](https://mahouse.world/glossary/open-finance.md): Open finance extends open banking beyond payment accounts to the rest of a customer's financial life — savings, investments, pensions, insurance, mortgages, loans — so that, with the customer's consent, a third party can read the data and initiate actions across all of it through standard APIs. Where open banking made a bank account portable, open finance makes the whole balance sheet portable.
- [Originate-to-Distribute (OTD)](https://mahouse.world/glossary/originate-to-distribute-otd.md): Originate-to-distribute is a lending model in which the lender makes loans in order to sell them — through securitisation, loan syndication or direct sales to investors — rather than holding them to maturity. The originator earns fees and servicing income instead of interest margin and needs far less capital; the credit risk passes to whoever buys the loans.
- [Outcome-Based Pricing](https://mahouse.world/glossary/outcome-based-pricing.md): Outcome-based pricing charges a customer for a result the software produced — a support ticket resolved, a debt collected, a document processed, a sale closed — rather than for a seat, a licence or a unit of usage. It has spread with AI agents, whose vendors sell work done rather than tools, and it changes what a buyer of such a company is buying: not recurring revenue, but a share of the customer's outcomes.
- [Overfitting](https://mahouse.world/glossary/overfitting.md): Overfitting is a model learning its training data so closely — including its noise, its coincidences and its particular quirks — that it performs well on the examples it saw and worse on anything new; the model has memorised rather than generalised. It is detected by the gap between training and validation performance, prevented by more data, simpler models, regularisation and early stopping, and it is the most common reason a model that looked excellent in development disappoints in production.
- [PCI Compliance Level](https://mahouse.world/glossary/pci-compliance-level.md): A PCI compliance level is the tier the card schemes assign to a merchant or service provider by its annual card volume — Level 1 above six million transactions, down to Level 4 — which decides how it must prove PCI DSS compliance: an annual audit by a qualified assessor at Level 1, a self-assessment and quarterly scans below. The level sets the cost, and products are designed to keep card data out of scope.
- [PCI DSS](https://mahouse.world/glossary/pci-dss.md): PCI DSS — the Payment Card Industry Data Security Standard — is the set of security requirements the card schemes impose on every organisation that stores, processes or transmits cardholder data: twelve requirements from firewalls and encryption to access control, logging and testing, enforced through acquirers' contracts. It is not law, but failing it brings fines, the loss of card acceptance and, after a breach, the fraud bill.
- [PSD2](https://mahouse.world/glossary/psd2.md): PSD2 — the second Payment Services Directive (EU 2015/2366, in force from January 2018) — is the EU law that regulates payment services and payment institutions and, in its best-known provisions, required banks to give licensed third parties access to customers' account data and payment initiation with the customer's consent (open banking) and required strong customer authentication for electronic payments. Its successor, PSD3 with the Payment Services Regulation, will replace it around 2027.
- [Passporting](https://mahouse.world/glossary/passporting.md): Passporting is the right of a financial firm authorised in one European Economic Area country to serve customers in the others under its home licence, by notifying its home regulator rather than seeking local licences. It exists licence by licence — banking, investment services, payments, e-money, funds, insurance — and UK firms lost it at the end of 2020.
- [Payment Aggregator](https://mahouse.world/glossary/payment-aggregator.md): A payment aggregator is a provider that accepts payments on behalf of many merchants under its own merchant account or licence and settles the money on to them, so that each merchant need not have an account of its own — the model of early PayPal, of marketplaces collecting for sellers, and of the emerging-market aggregators that give small merchants cards, wallets and transfers through one integration. The payment facilitator is its regulated, registered form.
- [Payment Descriptor](https://mahouse.world/glossary/payment-descriptor.md): A payment descriptor is the text that appears on a cardholder's bank statement to identify a charge — up to 22 characters plus a phone number or city, set by the merchant through its acquirer — and it is the first thing a cardholder reads before deciding whether to call the bank. A descriptor that says "SQ *COFFEE HOUSE" is remembered; one that says "TXN 4471 LTD" is disputed. Fixing it is the cheapest chargeback reduction there is.
- [Payment Facilitator](https://mahouse.world/glossary/payment-facilitator.md): A payment facilitator is a company that holds one master merchant account with an acquirer and lets other businesses accept card payments as its sub-merchants, onboarding them itself in minutes rather than sending each through the acquirer's underwriting. Stripe and Square built the model; software platforms adopted it to embed payments in their products; and it is now the standard way a fintech becomes a payments company without becoming an acquirer.
- [Payment Friction](https://mahouse.world/glossary/payment-friction.md): Payment friction is any step in a payment that costs the customer effort, time or attention — entering card details, creating an account, a redirect to a bank page, a one-time code, a failed attempt and a retry — and each step measurably reduces completion. Removing friction raises conversion by points that compound into revenue; adding it where risk requires is the control regulation and fraud demand. The design skill is putting friction where it prevents loss and nowhere else.
- [Payment Gateway](https://mahouse.world/glossary/payment-gateway.md): A payment gateway is the software layer that takes a payment from a merchant's checkout or terminal, encrypts and tokenises the card data, sends the authorisation request to the acquirer or processor, returns the response, and later submits the transaction for settlement. It is the plumbing between merchant and acquiring side — a product on its own or part of a payment service provider — and it neither moves money nor takes risk.
- [Payment Orchestration](https://mahouse.world/glossary/payment-orchestration.md): Payment orchestration is a software layer between a merchant and its payment providers that connects to several acquirers, PSPs, fraud tools and payment methods through one integration and decides, per transaction, where to send it — by cost, approval rate, availability, geography or risk — with the tokens, reconciliation and reporting unified across them. It is the merchant's answer to PSP lock-in and single points of failure, sold by independent platforms and, increasingly, by the PSPs themselves.
- [Payment Service Provider](https://mahouse.world/glossary/payment-service-provider.md): A payment service provider gives a merchant everything it needs to accept payments under one contract — gateway, access to card acquiring, local payment methods, fraud tools, settlement and reporting — for a blended fee per transaction. In European law the term means any licensed payment institution; in the market it means the Stripes, Adyens and Worldpays a merchant signs with.
- [Payout Cycle](https://mahouse.world/glossary/payout-cycle.md): A payout cycle is the schedule on which a payment provider transfers a merchant's settled funds to its bank account — daily on a delay (T+1, T+2), weekly, or on request — set by the provider's own settlement from the schemes, its view of the merchant's risk, and its appetite for the float between. For a merchant it is working capital; for the provider a risk buffer and a small income on the float.
- [Platform Risk](https://mahouse.world/glossary/platform-risk.md): Platform risk is the risk that a business built on another company's platform — an app store, marketplace, API, cloud service, payment network or sponsor bank — is harmed when that platform changes its fees, rules, algorithms or access, starts competing with it, or fails. The dependent business carries the exposure but controls none of the decisions that create it.
- [Point of Sale](https://mahouse.world/glossary/point-of-sale.md): Point of sale is where a sale is completed — the till, the counter, the checkout — and, in payments usage, the system that does it: the terminal that accepts the card and the software that records the sale, manages inventory, prices and staff. The terminal and the till were separate for decades; the integrated POS system is now the product, and the payments are the revenue that funds the software.
- [Point-to-Point Encryption](https://mahouse.world/glossary/point-to-point-encryption.md): Point-to-point encryption is the encryption of card data at the moment it is read — in the terminal's secure hardware — so that it travels through the merchant's systems and networks unreadable and is decrypted only inside the processor's hardware security module. A PCI-validated P2PE solution takes the merchant's point-of-sale environment almost entirely out of PCI DSS scope; a merchant that cannot see card numbers cannot leak them, and the questionnaire that proves it is a few pages long.
- [Poison Pill](https://mahouse.world/glossary/poison-pill.md): A poison pill is a defence a listed company's board adopts against an unwanted takeover: a shareholder rights plan that lets every shareholder except the bidder buy new shares at a deep discount once the bidder's stake crosses a threshold, diluting the bidder so heavily that acquiring control without the board's consent becomes uneconomic.
- [Politically Exposed Person](https://mahouse.world/glossary/politically-exposed-person.md): A politically exposed person is someone entrusted with a prominent public function — a head of state, minister, member of parliament, senior judge, military commander, central banker, board member of a state enterprise, senior official of an international organisation — and, by extension, their family members and close associates. The AML rules treat them as higher-risk for bribery and corruption and require enhanced due diligence: source-of-wealth checks, senior-management approval and closer monitoring, whether or not anyone suspects anything.
- [Pontes](https://mahouse.world/glossary/pontes.md): Pontes is the Eurosystem's service for settling wholesale transactions in tokenised assets in central bank money. It links market distributed-ledger (DLT) platforms with TARGET Services, so that the cash leg of a trade recorded on a DLT platform settles in euro at the central bank, with finality in T2. It launched on 21 September 2026.
- [Portfolio Company](https://mahouse.world/glossary/portfolio-company.md): A portfolio company is a business a private equity or venture fund has invested in and holds among its other investments — a platform if the fund controls it, a minority position if not. The label changes how it is run: to the fund's plan, on the fund's timetable, toward the exit the fund's investors await. A buyer of one is buying from a seller whose fund life sets the clock.
- [Post-Merger Integration](https://mahouse.world/glossary/post-merger-integration.md): Post-merger integration is the work after completion of making two companies one — combining the people, the platforms, the products, the contracts and the compliance programmes — so that the synergies the price assumed actually arrive. It is where most acquisitions succeed or fail, and in fintech it turns on one decision: which platform survives, and how the customers are moved to it without leaving.
- [Post-Money Valuation](https://mahouse.world/glossary/post-money-valuation.md): Post-money valuation is the value of a company immediately after a funding round: the pre-money valuation plus the money raised, or, equivalently, the round's price per share multiplied by all shares outstanding after it. It is the number the press reports, the number "unicorn" refers to, and the number a company's next round is measured against — and it is the price of the last shares sold, not what the whole company would fetch.
- [Pre-Arbitration](https://mahouse.world/glossary/pre-arbitration.md): Pre-arbitration is the stage of a card dispute after a merchant has represented a chargeback and the issuer disagrees: the issuer sends the case back with reasons, the merchant decides whether to accept the loss or continue, and if neither yields the case goes to arbitration, where the scheme rules and the loser pays a fee of several hundred dollars. It is where dispute economics turn against small amounts.
- [Pre-Money Valuation](https://mahouse.world/glossary/pre-money-valuation.md): Pre-money valuation is the value placed on a company immediately before a funding round — the price the new investors agree the existing business is worth, before their money is added. Together with the amount raised it fixes the price per share and how much of the company the round buys: an investment of €5m at a €20m pre-money buys 20% of a company worth €25m post-money.
- [Pre-Training](https://mahouse.world/glossary/pre-training.md): Pre-training is the first and by far the most expensive phase of building a modern AI model: training it on a very large, broadly unlabelled corpus — trillions of tokens of text, billions of images — with a self-supervised objective such as predicting the next token, so that it learns the general structure of language, images or code before any specific task. The result is a base or foundation model; fine-tuning and preference training adapt it. Few can afford it.
- [Precedent Transaction Analysis](https://mahouse.world/glossary/precedent-transaction-analysis.md): Precedent transaction analysis values a business by the multiples paid in past acquisitions of similar companies — what buyers actually paid for control, including the premium for it and the synergies they expected — applied to the target. Where trading comparables say what the market would pay for a share, precedents say what an acquirer paid for the whole company, and the difference is the control premium.
- [Prepaid Card](https://mahouse.world/glossary/prepaid-card.md): A prepaid card is a scheme-branded card that spends a balance loaded in advance rather than drawing on a bank account or a credit line — issued under an e-money licence or by a bank, run by a programme manager, and used for gift cards, payroll, travel money, expense management and the first product of most neobanks. Its economics are interchange and fees on a safeguarded float; its regulation is the e-money and AML rules.
- [Price-to-Earnings Ratio](https://mahouse.world/glossary/price-to-earnings-ratio.md): The price-to-earnings ratio is a company's share price divided by its earnings per share — its market value divided by net profit: how many years of after-tax earnings the market pays for. Banks and lenders are valued on it because interest is part of their business; for fast-growing or loss-making fintechs it is huge or meaningless, and other multiples take over.
- [Primary Account Number](https://mahouse.world/glossary/primary-account-number.md): The primary account number is the card number itself — usually sixteen digits: the issuer's BIN at the front, the account identifier in the middle, a Luhn check digit at the end — the value that identifies the account on every transaction and that, with an expiry and a CVV, is enough to buy online. It is the data PCI DSS exists to protect, and the data tokens and encryption exist to keep out of everyone's systems.
- [Private Equity](https://mahouse.world/glossary/private-equity.md): Private equity is the business of buying control of companies with a mix of a fund's capital and borrowed money, improving them over four to six years, and selling them for a multiple of the equity invested. The funds are raised from institutions, run by general partners paid a fee and a share of profits, and the model — leverage, operational change, a timed exit — has owned much of the payments industry since 2015.
- [Private Placement](https://mahouse.world/glossary/private-placement.md): A private placement is a sale of securities — shares, convertibles, bonds — to a selected group of investors rather than to the public, without a full prospectus, under exemptions for sophisticated buyers. A venture round is one; so is a listed company's overnight placing to institutions, and the PIPE that funds a SPAC merger. It is faster and cheaper than a public offer, reaches fewer investors, and is priced at a discount.
- [Pro Rata Right](https://mahouse.world/glossary/pro-rata-right.md): A pro-rata right is an investor's contractual right to buy enough of a company's future rounds to keep its percentage ownership from being diluted — an investor with 10% may take 10% of each new round. It is the mechanism through which early investors follow their winners, the term angels and seed funds fight hardest for, and the one new lead investors most often cut back, because the round they are pricing is only so large.
- [Processor](https://mahouse.world/glossary/processor.md): A processor is the company whose systems run the transaction machinery for an acquirer or an issuer — receiving authorisations, applying rules, routing to the networks, clearing, computing fees and settlement — as a service bought rather than built. Acquirer processors serve the merchant side; issuer processors serve the card-issuing side, including the modern platforms fintechs issue on. The word is also used loosely for any PSP, so contracts define it.
- [Programmable Finance](https://mahouse.world/glossary/programmable-finance.md): Programmable finance is money or a financial instrument that carries its own logic — conditions on when it can be spent, restrictions on who can receive it, automatic behaviour such as splitting, escrow, distribution or settlement against another asset — enforced by the asset's own code rather than by systems around it. Smart contracts made it possible; tokenised deposits, stablecoins and tokenised securities are where it is being built.
- [Prompt](https://mahouse.world/glossary/prompt.md): A prompt is the text given to a language model that sets its task — instructions, context, examples and the question — and for a large language model it does the work that configuration, code and training data do elsewhere. A system prompt, written by the developer, sets the model's role and rules; the user's message follows. Prompt engineering is the craft of writing prompts that work; prompt injection is the attack of smuggling instructions into content the model reads.
- [Prompt Engineering](https://mahouse.world/glossary/prompt-engineering.md): Prompt engineering is the practice of designing what a language model is given — the instructions, the role, the context, the examples, the output format and the constraints — so that it produces reliable, correct and correctly formatted output for a task; it moved from tricks in 2022 to a discipline by 2024, with versioned prompts, evaluation sets and review, and it is the first tool to reach for before fine-tuning. A prompt cannot guarantee safety or add knowledge.
- [Prompt Injection Attack](https://mahouse.world/glossary/prompt-injection-attack.md): A prompt injection attack hides instructions in content an AI system reads — a user message, an email, a web page, a document — so that the model treats the attacker's text as instructions and ignores its own rules, leaks data, or takes actions on the attacker's behalf. Direct injection comes from the user; indirect injection comes through content the system retrieves, the dangerous form for agents that act. The defence is limiting what a fooled model can do.
- [Purchase Price Allocation](https://mahouse.world/glossary/purchase-price-allocation.md): Purchase price allocation is the accounting exercise, required under IFRS 3 and ASC 805, in which a buyer divides the price it paid for a company among the identifiable assets and liabilities it acquired at their fair values — customer relationships, technology, brand, contracts, deferred tax — with whatever is left recorded as goodwill.
- [QR Code Payment](https://mahouse.world/glossary/qr-code-payment.md): A QR code payment is initiated by scanning a two-dimensional code — printed at a stall, shown on a screen, generated in an app — carrying the payee's identifier and often the amount, so a wallet or banking app completes the payment without a card or a terminal. It made acceptance cost a printed sign, and became the dominant retail payment in China, India, Brazil and much of Southeast Asia.
- [Quality of Earnings](https://mahouse.world/glossary/quality-of-earnings.md): A quality of earnings report is the buyer's financial due diligence on what a business really earns: an accountant's analysis of the seller's EBITDA that strips out one-offs, tests the adjustments management has added back, checks revenue against contracts and cash, and sets the working capital and net debt figures the price will be adjusted on. The number it produces — normalised, run-rate EBITDA — is the one the multiple is applied to.
- [RAG](https://mahouse.world/glossary/rag.md): Retrieval-augmented generation (RAG) is the design in which a system first retrieves the documents relevant to a question — from a knowledge base, a customer's records, a policy library — and then gives them to a language model as context, so that the answer is grounded in specific, current, citable sources rather than in what the model remembers from training. It is the standard architecture for enterprise assistants and the main defence against hallucination.
- [RLHF](https://mahouse.world/glossary/rlhf.md): RLHF — reinforcement learning from human feedback — is the post-training method that shapes a language model's behaviour by human preference: people compare pairs of the model's responses and pick the better one; a reward model is trained to predict those preferences; and the language model is optimised to produce responses the reward model scores highly, constrained to stay close to its starting point. Introduced at scale with InstructGPT (2022), it made chat models helpful; its side effects include sycophancy.
- [Real-Time Gross Settlement (RTGS)](https://mahouse.world/glossary/real-time-gross-settlement-rtgs.md): Real-time gross settlement is the settlement of payments one by one, in central bank money, the moment each is processed and with finality — as opposed to net settlement, where obligations are accumulated and settled as a balance at fixed times. Fedwire, CHAPS and the euro area's T2 are RTGS systems; they move the largest values in each economy and are the accounts everything else settles into.
- [Real-World Assets](https://mahouse.world/glossary/real-world-assets.md): Real-world assets (RWAs) are assets that exist off a blockchain — government bonds, fund shares, loans, gold, property — represented by tokens on one. A token is only as good as the legal structure behind it: the custodian, fund or special-purpose vehicle that holds the asset and treats the token as a record of ownership or as a claim.
- [Reconciliation](https://mahouse.world/glossary/reconciliation.md): Reconciliation is the matching of a company's own record of payments — orders, payouts, refunds, fees — against what its providers report and what actually arrived in its bank accounts, transaction by transaction, so every euro is accounted for and every difference explained. In a payments company it is also the daily proof that client money equals customer balances; everywhere it is where lost money surfaces.
- [Refund](https://mahouse.world/glossary/refund.md): A refund is the merchant's return of money to a cardholder for a transaction that was captured and settled — submitted through the acquirer as a credit, cleared back through the network, and posted to the cardholder's account days later. It differs from a void, which cancels an authorisation before capture and costs nothing, and from a chargeback, which the cardholder's bank forces; a merchant that refunds quickly and visibly prevents the chargeback that a slow refund invites.
- [RegTech](https://mahouse.world/glossary/regtech.md): RegTech — regulatory technology — is software and data services that help regulated firms meet their obligations more accurately and at lower cost: identity verification and KYC, sanctions and PEP screening, transaction monitoring, regulatory reporting, trade surveillance, record-keeping, policy management, and the horizon-scanning that tracks rule changes. It is the compliance function's supply chain, a sector of its own, and one of the most active corners of fintech M&A, because compliance spend is non-discretionary and the products are sticky.
- [Regularisation](https://mahouse.world/glossary/regularisation.md): Regularisation is any addition to training that discourages a model from fitting its training data too closely — a penalty on large weights (L2, weight decay), a penalty that drives weights to zero (L1), random disabling of units during training (dropout), limits on tree depth, early stopping, and data augmentation — so that the model captures the pattern rather than the noise. It is the standard defence against overfitting; its strength is set on validation data; too much underfits.
- [Regulatory Capital](https://mahouse.world/glossary/regulatory-capital.md): Regulatory capital is the loss-absorbing funding a financial firm must hold by law — for a bank, common equity and certain subordinated instruments measured against risk-weighted assets; for a payment or e-money institution, own funds measured against its volume or float. It caps how fast a regulated business can grow and shapes what an acquirer pays, because goodwill from a purchase is deducted from it.
- [Reinforcement Learning](https://mahouse.world/glossary/reinforcement-learning.md): Reinforcement learning (RL) is training a system by rewarding the outcomes of its actions rather than by showing it correct answers: an agent acts in an environment, receives a reward, and adjusts its policy — its mapping from situations to actions — to earn more reward over time. It produced the game-playing systems (AlphaGo, 2016), it is the RL in RLHF that shaped chat models, and it trained the reasoning models of 2024–2025; in fintech it optimises pricing and collections.
- [Reinforcement Learning Environments](https://mahouse.world/glossary/reinforcement-learning-environments.md): A reinforcement learning environment is the world an AI agent is trained in: the tasks it faces, the tools it can use, and the grader that turns what it does into a reward. For language-model agents it is usually a sandboxed copy of real software — a code repository, a browser, a ledger — that checks each task automatically.
- [Remittance](https://mahouse.world/glossary/remittance.md): A remittance is a cross-border transfer by a migrant worker to family in another country — small, regular, and in aggregate enormous: about $650 billion a year to low- and middle-income countries, more than foreign direct investment or aid. It moves through banks, money-transfer operators, fintechs and mobile-money networks; it costs an average of over 6%, which the G20 wants at 3%; and it is where AML rules meet the customers least able to bear their cost.
- [Representations and Warranties](https://mahouse.world/glossary/representations-and-warranties.md): Representations and warranties are the seller's statements of fact about the business in the sale agreement — the accounts are true, the licences valid, there is no litigation, the contracts are as listed — on which the buyer relies and for which it can claim if they prove false. They are the buyer's protection against what diligence did not find; the disclosure letter is where the seller says what it knows to be untrue.
- [Representment](https://mahouse.world/glossary/representment.md): Representment is the merchant's formal response to a chargeback: the transaction is presented to the issuer a second time with evidence that the dispute is invalid — proof of delivery, of authorisation, of a cancellation that never came — and if the issuer accepts, the funds return. It is the only way to recover a charged-back sale; it wins a fifth to a half of cases; and it does not lower the chargeback ratio.
- [Reserve Account](https://mahouse.world/glossary/reserve-account.md): A reserve account is money an acquirer or payment provider withholds from a merchant's settlements and holds against future chargebacks, refunds and fines — typically a rolling percentage of volume held for a period, or a fixed sum — released when the risk has passed. It is the acquirer's collateral for the credit it extends to every merchant that takes payment before it delivers.
- [Retrieval Reference Number](https://mahouse.world/glossary/retrieval-reference-number.md): A retrieval reference number is the 12-digit reference assigned to a card transaction at authorisation — by the acquirer, processor or terminal, under ISO 8583 — and carried in subsequent messages so all parties can refer to the same transaction: on a receipt, in a reversal, in an issuer's retrieval request. It exists from authorisation, unlike the acquirer reference number, which is assigned at clearing.
- [Retry Logic](https://mahouse.world/glossary/retry-logic.md): Retry logic is the set of rules, or the model, that decides whether and when to resubmit a declined payment — which decline codes to retry, how many times, at what intervals and times of day, through which acquirer — within the card schemes' limits on reattempts. It is the largest lever in decline recovery for subscription businesses, and the difference between a naive engine and a good one.
- [Return on Tangible Equity](https://mahouse.world/glossary/return-on-tangible-equity.md): Return on tangible equity is net profit attributable to ordinary shareholders divided by average tangible equity — shareholders' equity less goodwill, other intangible assets and preferred equity. It is the profitability measure banks set targets on and investors compare them by, because it strips out the goodwill that past acquisitions leave on the balance sheet.
- [Revenue-Sharing Model](https://mahouse.world/glossary/revenue-sharing-model.md): A revenue-sharing model is a commercial arrangement in which partners split the revenue a product generates — a percentage of interchange, processing margin, interest, commission or subscription — rather than one paying the other a fixed fee, so that both are paid in proportion to the product's success. It is the standard structure of embedded finance, banking as a service and referral arrangements; the terms that matter are what "revenue" means, who bears the costs, and what happens at termination.
- [Reversal](https://mahouse.world/glossary/reversal.md): A reversal is the message a merchant or acquirer sends to cancel an authorisation, in full or in part, so the issuer releases the hold on the cardholder's balance — because the sale did not go through, the terminal timed out, the customer cancelled, or the final amount was less than the estimate. It is the mechanism behind a void and a partial release; the word is also used loosely for refunds and chargebacks, which differ.
- [Reverse Merger](https://mahouse.world/glossary/reverse-merger.md): A reverse merger is the way a private company becomes listed by merging into a company that already is — a shell or a fading listed business — so that the private company's shareholders end up owning most of the combined entity and its business becomes the listed one, without an initial public offering. The SPAC boom of 2020–2021 was reverse mergers at scale, and it took a dozen fintechs public before the market turned.
- [Rights Issue](https://mahouse.world/glossary/rights-issue.md): A rights issue is an offer of new shares to existing shareholders in proportion to what they hold — one new share for every four — at a discount to the market price, with the right to buy tradeable so that holders who do not want more shares can sell the entitlement. It respects pre-emption, raises large sums, and is how banks and listed fintechs rebuild capital when a placing would be too small or unfair.
- [Risk-Based Monitoring](https://mahouse.world/glossary/risk-based-monitoring.md): Risk-based monitoring is the practice of scrutinising customers and transactions in proportion to the risk each presents — tighter thresholds, more frequent review and deeper questions for the higher-risk, lighter and automated controls for the rest — rather than applying one rule to everyone. It is what the FATF standards and every anti-money-laundering regime require, and the difference between a compliance function that finds cases and one that drowns in alerts.
- [Risk-Based Routing](https://mahouse.world/glossary/risk-based-routing.md): Risk-based routing chooses a transaction's path by its assessed risk — a low-risk payment goes frictionless with an SCA exemption on the acquirer with the best approval rate; a medium-risk one through a 3-D Secure challenge; a high-risk one to review, to a stricter acquirer, or to a decline — so friction and cost are spent where the risk is. It is where the fraud engine and the routing engine meet.
- [Risk-Free Rate](https://mahouse.world/glossary/risk-free-rate.md): The risk-free rate is the return on an investment with no default risk — in practice the yield on a government bond in the currency and at the horizon of the cash flows being valued — and the base every required return is built on. When it rose from near zero to 4% in 2022–2023, every valuation built on it fell, and fintech's fell most.
- [Robo-Adviser](https://mahouse.world/glossary/robo-adviser.md): A robo-adviser is a digital service that builds and manages an investment portfolio automatically — a questionnaire sets the customer's risk level, an algorithm allocates the money across low-cost index funds, and software rebalances it — for a fee of a fraction of a per cent of assets. It brought portfolio management to accounts too small for a human adviser and forced the fees of the whole wealth industry down.
- [Roll-Up](https://mahouse.world/glossary/roll-up.md): A roll-up is a strategy of acquiring many small companies in a fragmented market and combining them into one, on the arithmetic that small firms sell for low multiples and large ones for high — so the combined business is worth more than the prices paid before any synergy. It is how private equity built the payments consolidators, and how a strategy becomes a spreadsheet when integration does not keep up.
- [Rolling Reserve](https://mahouse.world/glossary/rolling-reserve.md): A rolling reserve is a percentage of each settlement — typically 5–20% — that an acquirer or payment facilitator withholds from a merchant and releases after a set period, usually 90 to 180 days, so a fund is always on hand for the merchant's chargebacks, refunds and fraud if the merchant cannot pay. It rolls as each day's holdback is released when an older one matures; for the merchant it is working capital it cannot use.
- [Runway](https://mahouse.world/glossary/runway.md): Runway is the number of months a company can operate before its cash runs out at its current net burn — available cash divided by monthly burn. It is the clock every venture-backed company lives on: a round is raised with a year of it left, a bridge sought at six months, cuts or a sale at three. For a fintech, available cash excludes the licence's capital and the customers' money.
- [SAFE](https://mahouse.world/glossary/safe.md): A SAFE — a simple agreement for future equity — is an instrument under which an investor pays a company now and receives shares at the next priced round, at a discount or a valuation cap, with no interest, no maturity and no debt. Introduced by Y Combinator in 2013 and revised to the post-money form in 2018, it is the standard seed instrument in the US because it is five pages long and takes a day to sign.
- [SCA Exemption](https://mahouse.world/glossary/sca-exemption.md): An SCA exemption is one of the cases in which a European electronic payment may proceed without strong customer authentication — the two-factor check PSD2 requires — because the transaction is low-value, low-risk under the acquirer's or issuer's fraud rates, recurring, to a merchant the customer has whitelisted, or between two accounts of the same person. Exemptions are the difference between a checkout that converts and one that does not, and applying them well is a payments company's competitive edge.
- [SOC 2 Compliance](https://mahouse.world/glossary/soc-2-compliance.md): SOC 2 is an attestation, issued by an independent accountant under the AICPA's standards, that a service company's controls over security — and optionally availability, processing integrity, confidentiality and privacy — are designed properly (Type I) and operated effectively over a period (Type II). It is the report a bank's procurement team asks a fintech vendor for before any contract, and one of the first documents in a technology company's data room.
- [SPAC](https://mahouse.world/glossary/spac.md): A SPAC — a special purpose acquisition company — is a listed shell that raises money in its own IPO with no business, holds it in trust, and has about two years to merge with a private company, which thereby becomes listed. The sponsor keeps a fifth of the shell's shares for almost nothing; IPO investors can redeem their cash at the merger vote. It took dozens of fintechs public in 2020–2021, most of them below $10 within a year.
- [Safeguarding](https://mahouse.world/glossary/safeguarding.md): Safeguarding is the set of rules requiring payment and e-money institutions to keep customers' money apart from their own and protect it from their creditors, either by holding it in segregated accounts or secure, liquid assets, or by covering it with an insurance policy or a bank guarantee. It is what stands between customers and a failed fintech.
- [Sanctions Screening](https://mahouse.world/glossary/sanctions-screening.md): Sanctions screening is the checking of customers, counterparties and payments against the lists of persons, entities, vessels and countries that governments and the UN have prohibited dealing with — OFAC's, the EU's, the UK's — so a match is stopped, frozen and reported before money moves. It is strict liability in most regimes: a payment to a listed person is a breach whether or not the firm knew.
- [Sandbox](https://mahouse.world/glossary/sandbox.md): A regulatory sandbox is a supervised environment in which a firm tests a regulated financial product with real customers, for a limited time and scale, under requirements the regulator has relaxed or tailored and with the regulator watching closely — a way for a new product to meet the market before its full licence, and for the regulator to learn how the rules fit it. The UK's FCA launched the first in 2016; the EU AI Act mandates AI sandboxes.
- [Scheme Fee](https://mahouse.world/glossary/scheme-fee.md): Scheme fees are the charges the card networks — Visa, Mastercard and their peers — levy on the acquirers and issuers that use them: assessments as a percentage of volume, per-transaction fees for authorisation and clearing, cross-border and currency assessments, and charges for services and programmes. They are the network's revenue, distinct from interchange, and they have risen as the regulated part of card costs was capped.
- [Seamless Payments](https://mahouse.world/glossary/seamless-payments.md): Seamless payments are payment experiences with no visible payment step at the moment of purchase: the customer authorised the relationship once — a stored credential, a mandate, a wallet linked to the account — and each transaction thereafter completes in the background, as in a ride-hailing app, a walk-out store or a subscription. They depend on stored credentials, merchant-initiated transaction rules, authentication done once and exempted thereafter, and a risk model that decides when to interrupt.
- [Secondary Market](https://mahouse.world/glossary/secondary-market.md): The secondary market for private companies is where existing shares change hands — an employee or an early investor selling to a fund or a new investor — without the company issuing new shares or raising money. It gives shareholders liquidity years before an exit, and it gives buyers a way into companies that are not raising.
- [Seed Investment](https://mahouse.world/glossary/seed-investment.md): Seed investment is the first institutional money into a startup: capital of roughly $500,000 to $5m from angels and seed funds, raised on a product that exists but a business that does not yet, to reach the revenue or usage that will justify a Series A. It is priced on a story and a team, structured on a SAFE or a priced round, and it sets the cap table every later round inherits.
- [Sell-Side Advisory](https://mahouse.world/glossary/sell-side-advisory.md): Sell-side advisory is the work an investment bank or M&A boutique does for the owners of a company that is being sold: preparing the business and the materials, building the buyer list, running the auction or the negotiation, managing diligence, and negotiating the price and the terms to signing. The adviser's job is to find the buyer who will pay the most and close, and its fee is a percentage of what that buyer pays.
- [Senior Debt](https://mahouse.world/glossary/senior-debt.md): Senior debt is the layer of a company's borrowing that ranks first: secured on the company's assets and shares, repaid before every other creditor and the shareholders, and therefore the cheapest money in the structure. Banks and, increasingly, private credit funds provide it; it is sized at a multiple of EBITDA, protected by covenants, and in a buy-out it is the largest single source of the price.
- [Series A](https://mahouse.world/glossary/series-a.md): A Series A is a startup's first large priced round of venture capital, raised once a product has found paying customers and the question has changed from whether it works to how fast it can grow. It is led by a venture fund that takes a board seat and preferred shares with a liquidation preference, and it is the round that turns a founding team into a company with governance.
- [Series B](https://mahouse.world/glossary/series-b.md): A Series B is the venture round raised to scale a business that has already proved itself at Series A — to build the sales organisation, enter new markets and, in fintech, to fund the capital and compliance that a licence at scale demands. It is priced on metrics rather than promise, led by larger funds, and it is the round at which the rights granted at Series A are tested.
- [Series D](https://mahouse.world/glossary/series-d.md): A Series D is a late-stage venture round — the fourth priced round after seed — raised by a company that is large enough to be preparing for an exit but not yet ready for one, or that needs more capital than the plan at its Series C assumed. It is led by growth funds, crossover investors and sometimes strategic buyers, priced close to public-market comparables, and it is the round where a down round is most visible.
- [Series E](https://mahouse.world/glossary/series-e.md): A Series E is the fifth lettered round of venture financing — after Series A to D — in which a late-stage private company sells a new class of preferred shares, usually to growth, crossover or sovereign investors. It is raised by companies not yet ready or willing to list or sell, or whose plans needed more money than expected.
- [Settlement](https://mahouse.world/glossary/settlement.md): Settlement is the transfer of funds between the institutions on either side of a payment that discharges the obligation clearing established — through the scheme's settlement bank or the central bank — and, in ordinary usage, the later transfer from the acquirer or PSP to the merchant. It is the step at which a payment becomes money rather than a promise; it is final when the system's rules say so.
- [Settlement Networks](https://mahouse.world/glossary/settlement-networks.md): Settlement networks are the systems through which the money behind a payment actually moves between institutions — card schemes, automated clearing houses, instant-payment schemes, SWIFT and the central-bank RTGS systems they all finally settle across. A fintech that "moves money" usually moves messages; the settlement network is where the balances change, and access to it is what a licence buys.
- [Share Buyback](https://mahouse.world/glossary/share-buyback.md): A share buyback is a company's purchase of its own shares — on the market over time, by a tender offer, or from a specific holder — which are cancelled or held in treasury, so the remaining shares represent a larger slice of the same business. It returns cash without a dividend, raises earnings per share by shrinking the denominator, and is what a company does with capital it cannot invest above its cost.
- [Share Purchase](https://mahouse.world/glossary/share-purchase.md): A share purchase is the acquisition of a company by buying its shares from the shareholders, so that the buyer takes over the legal entity with everything in it — contracts, licences, staff, assets and liabilities, known and unknown. The share purchase agreement is the contract that sets the price, the warranties and the conditions of that transfer.
- [Sharpe Ratio](https://mahouse.world/glossary/sharpe-ratio.md): The Sharpe ratio is an investment's return in excess of the risk-free rate divided by the volatility of that return — how much reward an investor earned per unit of risk taken, on the assumption that risk is variability. A ratio of 1 means the excess return equalled its standard deviation; above 1 is good, above 2 rare. Robo-advisers and fund selectors use it to compare portfolios and managers; it flatters strategies whose risk arrives all at once.
- [Signing and Closing](https://mahouse.world/glossary/signing-and-closing.md): Signing is the day the buyer and seller become bound by the sale agreement; closing — completion, in UK usage — is the day the shares and the money change hands. When nothing has to happen in between, the two are simultaneous; when approvals or consents are needed, they are split by a gap of weeks or months, during which the seller still runs the business and the buyer still has to pay for it.
- [Single Euro Payments Area](https://mahouse.world/glossary/single-euro-payments-area.md): The Single Euro Payments Area is the arrangement under which euro payments between bank accounts in 36 European countries work like domestic ones — one format (IBAN, ISO 20022), one set of scheme rules, one price rule — through the SEPA credit transfer, direct debit and instant schemes run by the European Payments Council. Every European account-to-account product rides it, and since 2025 instant euro payments are mandatory for banks.
- [Smart Contract](https://mahouse.world/glossary/smart-contract.md): A smart contract is a program deployed on a blockchain that executes automatically when its conditions are met — holding, transferring or issuing tokens according to its code — without a party able to stop, alter or reverse it once it runs; the term describes code, not a contract in the legal sense. Smart contracts issue every stablecoin and token and run every DeFi protocol; their flaws have cost billions, because code that holds money is money a bug moves.
- [Smart Routing](https://mahouse.world/glossary/smart-routing.md): Smart routing is the selection, for each transaction, of the path most likely to produce an approval at an acceptable cost — which acquirer, which network, which message format, which authentication route — using observed approval rates by issuer, card type, country, amount and time, plus cost and risk. It is the core function of a payment orchestration platform and a feature of the larger PSPs, and its value is measured in points of approval rate on the same transactions.
- [Soft Decline](https://mahouse.world/glossary/soft-decline.md): A soft decline is an issuer's refusal of a payment for a reason that may not apply next time — insufficient funds, the issuer's system unavailable, a velocity limit, a fraud rule, a request for authentication — as distinct from a hard decline, which is final. Soft declines are most e-commerce declines and the whole of decline recovery: retried, re-routed or resubmitted with a challenge, half become approvals.
- [Sovereign Wealth Fund](https://mahouse.world/glossary/sovereign-wealth-fund.md): A sovereign wealth fund is a state-owned investment fund, built from oil revenues, trade surpluses or pension reserves, that invests globally across public markets, private equity, real estate and, increasingly, direct stakes in growth companies. The largest — Norway's, Abu Dhabi's, Saudi Arabia's, Singapore's GIC and Temasek, Qatar's — manage hundreds of billions each, and their direct fintech investments since 2015 have been among the largest cheques in the sector's late-stage rounds.
- [Spin-Off](https://mahouse.world/glossary/spin-off.md): A spin-off is the separation of a division into an independent company whose shares are distributed to the parent's own shareholders, so that they end up owning two companies instead of one and no buyer is involved. It raises no cash for the parent, but it lets each business be valued and run on its own — which is why PayPal was spun off from eBay in 2015 and why conglomerates that own a fintech eventually let it go.
- [Split Payment](https://mahouse.world/glossary/split-payment.md): A split payment is a single payment from a customer that is divided at settlement among several recipients — a marketplace keeping its commission and paying the seller the rest, a delivery app paying the restaurant and the courier — by the payment provider, so the platform never holds the sellers' money itself. It is the primitive under every marketplace, and who holds the funds in between is a licensing question.
- [Sponsor-Bank Fees](https://mahouse.world/glossary/sponsor-bank-fees.md): Sponsor-bank fees are what a fintech pays the licensed bank whose charter it operates on: charges per account, per transaction and per card, a share of the interchange the fintech's cards earn, a programme or platform fee, and the cost of the compliance the bank performs on the fintech's customers. They are the price of not having a licence, and for many neobanks and card programmes the largest cost of revenue.
- [Stablecoin](https://mahouse.world/glossary/stablecoin.md): A stablecoin is a digital token issued on a blockchain and designed to hold a steady value against a reference — almost always the US dollar, increasingly the euro — by the issuer holding reserves of cash and short-term government securities against every token and redeeming at par on demand; algorithmic designs have failed. By 2025 stablecoins exceeded $200 billion in circulation and were regulated as a distinct category under MiCA and the US GENIUS Act.
- [Stablecoin Settlement](https://mahouse.world/glossary/stablecoin-settlement.md): Stablecoin settlement is the discharge of a payment obligation between businesses, such as a card network and an acquirer, by transferring a fiat-backed stablecoin on a blockchain instead of moving money between bank accounts. It runs around the clock and completes in minutes, but the asset is a claim on a private issuer, not central bank money.
- [Stalking Horse Bid](https://mahouse.world/glossary/stalking-horse-bid.md): A stalking horse bid is the first, binding offer for the assets of a company in insolvency, agreed with the debtor before an auction and used as the floor that other bidders must beat. In return for setting the floor and doing the diligence everyone else relies on, the stalking horse gets bid protections — a break-up fee and expense reimbursement if it is outbid — and the court-supervised auction decides who buys.
- [Stand-In Processing](https://mahouse.world/glossary/stand-in-processing.md): Stand-in processing is the card network's approval or decline of transactions on an issuer's behalf when the issuer's authorisation system is unreachable — applying limits and rules the issuer set in advance, so that cardholders can keep paying during an outage. The issuer carries the risk of what the network approved; the network carries the availability. For a neobank on a modern processor it is also what happens when the fintech's own real-time authorisation endpoint fails to answer.
- [Standstill Agreement](https://mahouse.world/glossary/standstill-agreement.md): A standstill agreement is a promise by a potential acquirer or a large shareholder not to buy the company's shares, launch a bid, solicit proxies or otherwise act to gain control for a stated period, typically given in exchange for confidential information or as part of a settlement with the board. It lets a listed company open its books to a bidder without arming it, and lets a board negotiate with an activist without a proxy fight running underneath.
- [Startup M&A](https://mahouse.world/glossary/startup-m-a.md): Startup M&A is the acquisition of, or merger with, a young company, usually backed by venture capital, by an incumbent, a private-equity platform or another startup. It differs from mature-company deals in the split: liquidation preferences, employee options, retention packages and holdbacks decide who receives the headline price, and the buyer often pays mainly for a team and its code.
- [Strong Customer Authentication](https://mahouse.world/glossary/strong-customer-authentication.md): Strong customer authentication is the PSD2 requirement that a customer initiating an electronic payment or accessing a payment account online be verified by two of three independent factors — something they know (a password), something they have (a phone, a card), something they are (a fingerprint, a face) — unless an exemption applies. It made European card-not-present fraud fall and checkout conversion fall with it, and building around it is most of what a European payments company's engineering does.
- [Supervised Learning](https://mahouse.world/glossary/supervised-learning.md): Supervised learning is the form of machine learning in which a model is trained on examples that carry the correct answer — transactions labelled fraud or not, applications labelled defaulted or repaid, documents labelled by type — so that it learns the mapping from inputs to labels and can predict the label for new inputs. Classification and regression are its two tasks; the labels are its cost and its limit, since the model learns exactly what they say.
- [Switching Costs](https://mahouse.world/glossary/switching-costs.md): Switching costs are what a customer gives up to change supplier — the money, time, risk and disruption of moving a bank account, a payments integration, a core banking system or a data feed to a rival — and therefore the price a supplier can charge above the competition without losing the customer. In fintech they are the moat most acquisition cases are built on, and the thing an integration most often destroys.
- [Symbolic AI](https://mahouse.world/glossary/symbolic-ai.md): Symbolic AI is the approach that represents knowledge as explicit symbols — facts, rules, logical relations, ontologies — and reasons by manipulating them: search, inference and planning over a model of the world written by people. It was "artificial intelligence" from the 1950s to the 1990s, produced expert systems and theorem provers, and lost to statistical machine learning on perception and language. It survives where explicit rules are the point — policy engines, compliance logic, knowledge graphs.
- [Syndicate](https://mahouse.world/glossary/syndicate.md): A syndicate is a group of investors that funds one round or one loan together under a lead who negotiates the terms, does the diligence and allocates the rest — a venture round with a lead and three followers, an angel syndicate pooling small cheques behind one experienced angel, a bank loan too large for one lender. It spreads risk, brings more names to the cap table, and puts the decisions in fewer hands than the money.
- [Synergies](https://mahouse.world/glossary/synergies.md): Synergies are the gains a buyer expects from combining two businesses that neither could achieve alone — cost synergies from removing duplicated functions, platforms and offices, and revenue synergies from selling each company's products to the other's customers. They are the justification for paying a premium over a company's standalone value, and the part of the price most often not recovered.
- [Synthetic Identity](https://mahouse.world/glossary/synthetic-identity.md): A synthetic identity is a fabricated person assembled from real and invented data — a genuine national identity number, often a child's or a recent immigrant's, attached to a made-up name, date of birth and address — that is used to open accounts, build a credit history and then borrow and disappear. It passes identity checks because its parts are real, and it is the fastest-growing form of credit fraud in the US.
- [Synthetic Securitisation](https://mahouse.world/glossary/synthetic-securitisation.md): A synthetic securitisation is a transaction in which a bank keeps a portfolio of loans on its balance sheet but transfers the risk of their losses to investors through a credit derivative or guarantee — typically selling the first-loss tranche to a specialist fund. The loans stay, the customers notice nothing, and the bank's capital requirement on the portfolio falls. It is how European and, increasingly, US banks free capital for new lending.
- [Tag-Along Right](https://mahouse.world/glossary/tag-along-right.md): A tag-along right lets minority shareholders join a sale by the majority on the same terms, so that a founder or a lead investor cannot sell their stake — and control — to a buyer while leaving everyone else behind with a new, unknown owner. It is the mirror of the drag-along: the drag lets the majority force a sale, the tag lets the minority share in one.
- [Take Rate](https://mahouse.world/glossary/take-rate.md): Take rate is the share of the volume flowing through a payments company or a marketplace that it keeps as revenue — gross take rate before the interchange, scheme fees and method costs it passes through, net take rate after them. It is the industry's unit economics in one number: multiply it by gross payment volume and the revenue appears; watch it over time and the pricing pressure, the merchant mix and the product attach are all in the trend.
- [Tangible Book Value](https://mahouse.world/glossary/tangible-book-value.md): Tangible book value is a company's shareholders' equity less goodwill and other intangible assets — the equity that would remain if everything that cannot be sold on its own were written off. For banks and lenders it is the measure that matters: acquisitions are priced as a multiple of it, regulators count capital from it, and the dilution a deal causes to it is the number a bank's board asks about first.
- [Tap to Pay](https://mahouse.world/glossary/tap-to-pay.md): Tap to pay, in its acceptance sense, is the use of an ordinary smartphone as a card terminal — the customer taps a contactless card or phone on the merchant's phone, whose NFC chip reads it and whose certified software does the terminal's EMV and security work, with no separate reader. Apple's Tap to Pay on iPhone and the Android SoftPOS products before it made the mPOS reader optional.
- [Teaser](https://mahouse.world/glossary/teaser.md): A teaser is the one- or two-page anonymous summary of a company for sale that the seller's adviser sends to potential buyers before any confidentiality agreement — the sector, the size, the growth, the headline financials and the reason for the sale, without the name. It exists to get a buyer interested enough to sign the NDA that unlocks the information memorandum, and to do so without telling the market who is selling.
- [Technological Singularity](https://mahouse.world/glossary/technological-singularity.md): The technological singularity is a hypothesised point at which artificial intelligence becomes able to improve itself, each generation designing a more capable successor faster than the last, so that progress accelerates beyond human ability to follow or control. The term is from Vernor Vinge (1993), the idea from I. J. Good's "intelligence explosion" (1965), the popular version from Ray Kurzweil; it is an argument about trend lines, not a dated forecast.
- [Tender Offer](https://mahouse.world/glossary/tender-offer.md): A tender offer is a public invitation by a bidder to a company's shareholders to sell their shares at a stated price within a set period — the way a listed company is bought when its shares are held by thousands of people who cannot all sign a share purchase agreement. In private companies the same word describes an organised window in which employees may sell shares to an approved buyer.
- [Tensor Processing Unit](https://mahouse.world/glossary/tensor-processing-unit.md): A tensor processing unit is Google's application-specific chip for neural-network arithmetic — designed from 2015 for the matrix multiplications that training and inference consist of, deployed in Google's data centres in pods of thousands, and rented through Google Cloud. It is the best-known of the custom accelerators the hyperscalers built to reduce dependence on Nvidia's GPUs, and the chip Google's own models are trained and served on.
- [Term Sheet](https://mahouse.world/glossary/term-sheet.md): A term sheet is a one- to three-page summary of the commercial terms of a proposed deal — price, structure, conditions, governance — set out as a list rather than as a contract, and signed as a basis for drafting the binding documents. In venture rounds it carries the investor's protective terms; in M&A it plays the role of the letter of intent.
- [Terminal Value](https://mahouse.world/glossary/terminal-value.md): Terminal value is the part of a discounted cash flow valuation that stands for everything the business earns after the explicit forecast ends — the value at the end of year five or ten of all the years beyond, estimated either as a growing perpetuity of the final year's cash flow or as an exit multiple of its earnings. It is usually the largest part of the valuation, and the part that rests on the fewest facts.
- [Tiered Pricing](https://mahouse.world/glossary/tiered-pricing.md): Tiered pricing charges a merchant one of several rates per transaction — "qualified", "mid-qualified" and "non-qualified" — depending on how the provider classifies the card and the way it was processed, with the provider deciding which tier each transaction falls into. It sits between a blend and interchange-plus in appearance and below both in transparency: the merchant sees three rates and cannot predict which applies, and the provider's margin lives in the downgrades.
- [Tokenised Deposits](https://mahouse.world/glossary/tokenised-deposits.md): A tokenised deposit is an ordinary bank deposit represented as a token on a distributed ledger, so that it can be transferred, settled and programmed with the speed and finality of a blockchain while remaining a claim on the bank — with deposit insurance, the bank's balance sheet behind it and the bank's compliance around it. It is the banks' answer to the stablecoin: the same rails, but the money stays money.
- [Tokenised Securities](https://mahouse.world/glossary/tokenised-securities.md): Tokenised securities are shares, bonds, fund units and other financial instruments issued or recorded as tokens on a distributed ledger, so that ownership passes by a transfer on the ledger rather than by an entry at a central securities depository. Tokenising changes the format of the instrument, not its legal nature: it is still a security.
- [Tokenization](https://mahouse.world/glossary/tokenization.md): Tokenization has two meanings that share a word and a principle. In payments it is replacing sensitive data — a card number above all — with a substitute value, the token, that is useless outside the system that issued it, so merchants and wallets can store and use the token while the real number stays in a vault. In capital markets it is representing an asset — a fund share, a bond — as a token on a blockchain.
- [Tokenized Public Equities](https://mahouse.world/glossary/tokenized-public-equities.md): Tokenized public equities are blockchain tokens that track a listed company's shares and are sold as a way to hold and trade them on-chain, around the clock and across borders. In most current forms the token is not the share but a claim on an intermediary that holds the stock, and what the holder gets in dividends, votes and an insolvency depends on that intermediary's terms.
- [Total Value Locked](https://mahouse.world/glossary/total-value-locked.md): Total value locked (TVL) is the market value, usually in US dollars, of the crypto assets users have deposited in a decentralised-finance protocol's smart contracts — lending collateral, exchange liquidity, vault deposits. It is DeFi's nearest equivalent to assets under management: a measure of deposits and trust, not of revenue or profit.
- [Training Data](https://mahouse.world/glossary/training-data.md): Training data is the set of examples a model learns from — the labelled transactions behind a fraud model, the trillions of tokens behind a language model, the annotated images behind a document reader — and it determines what the model can do and where it fails: the data's coverage becomes the model's coverage, its biases the model's biases, its errors the model's beliefs. Its provenance and rights are the first questions of any review.
- [Tranches](https://mahouse.world/glossary/tranches.md): Tranches are the slices into which a pool of loans or a financing is divided, each with its own priority of payment and its own risk: the senior tranche is paid first and loses last, the equity or first-loss tranche is paid last and loses first, and the mezzanine sits between. Tranching is how one portfolio of risk is sold to investors with different appetites — and, in a venture deal, how an investment is released in stages against milestones.
- [Transaction Lifecycle](https://mahouse.world/glossary/transaction-lifecycle.md): The transaction lifecycle is the sequence of states a card payment passes through from the moment a customer pays until the money is final: authorisation (approval and hold), capture (the merchant takes it), clearing (the record is exchanged), settlement (the money moves between banks), payout (the merchant is paid), and the after-life of refunds, reversals and chargebacks. Each state has an owner, a reference and a date.
- [Transaction Monitoring](https://mahouse.world/glossary/transaction-monitoring.md): Transaction monitoring is the automated surveillance of customers' payments and account activity for patterns that may indicate money laundering, terrorist financing or fraud — structuring below reporting thresholds, rapid movement of funds in and out, activity inconsistent with the customer's profile, links to high-risk jurisdictions — producing alerts that analysts investigate and, where suspicion remains, report to the financial intelligence unit. It is the largest operational cost in a fintech's compliance function and the control most often named in enforcement.
- [Transfer Learning](https://mahouse.world/glossary/transfer-learning.md): Transfer learning is reusing a model trained on one problem — usually a large, general one — as the starting point for another, related problem, so that the knowledge in its weights (edges and shapes in an image model, grammar and facts in a language model) carries over and the new task needs far less data and compute than training from scratch. Fine-tuning a pre-trained model is its usual form; almost no team trains from nothing.
- [Transformer](https://mahouse.world/glossary/transformer.md): The transformer is the neural-network architecture, introduced by Google researchers in 2017, that underlies every modern language model and most vision, speech and multimodal models: it processes a sequence of tokens in parallel through stacked layers of self-attention — in which every token weighs its relation to every other — and feedforward networks. It replaced the recurrent networks that processed sequences step by step, trained far faster on GPUs, and scaled to hundreds of billions of parameters.
- [Transitional Services Agreement](https://mahouse.world/glossary/transitional-services-agreement.md): A transitional services agreement is a contract under which the seller keeps providing services to a business it has sold — IT, payroll, finance, compliance, access to a licence or a platform — for a period after completion, at an agreed price, until the buyer can run them itself. It is what makes a carve-out possible: the unit can be sold before it is fully separable, and the TSA covers the gap.
- [Treasury Management](https://mahouse.world/glossary/treasury-management.md): Treasury management is the function that manages a company's cash and financial risk: where the money is, that there is enough where it is needed, how it is funded, how currency and rate exposure are hedged, and which banks hold it. In a payments company treasury also holds client money it must safeguard and settle, so it is a regulated operation, and a buyer reads its reconciliations before its forecasts.
- [Ultimate Beneficial Owners](https://mahouse.world/glossary/ultimate-beneficial-owners.md): Ultimate beneficial owners are the natural persons who ultimately own or control a company or other legal entity, directly or through a chain of companies, trusts or nominees. Anti-money-laundering rules require banks, payment firms and other regulated businesses to identify and verify them, usually starting from a 25% ownership or voting test, before serving the entity.
- [Unattended Terminal](https://mahouse.world/glossary/unattended-terminal.md): An unattended terminal is a card acceptance device operated by the customer alone — a fuel pump, a vending machine, a parking meter, a ticket machine, an EV charger, a self-checkout — with no staff present to check anything. The schemes give it its own categories and rules: contactless and PIN limits, estimate authorisations with completion advice, stricter physical security, because that is where skimmers and stolen cards go first.
- [Underfitting](https://mahouse.world/glossary/underfitting.md): Underfitting is a model too simple, too constrained or too briefly trained to capture the structure in its data, so that it performs poorly on the training set and on unseen data alike — the opposite failure to overfitting, where training performance is high and generalisation poor. Its signs are a training error that stays high and a small training–validation gap; its causes are too few parameters, too much regularisation, too few steps, or features that miss the signal.
- [Underlying Profit Before Tax](https://mahouse.world/glossary/underlying-profit-before-tax.md): Underlying profit before tax is statutory profit before tax adjusted to remove items management regards as unrepresentative of ongoing performance, such as restructuring costs, conduct charges, acquisition-related amortisation and one-off gains. It is an alternative performance measure: the company chooses the definition, and its usefulness depends on what is excluded and why.
- [Underserved Population](https://mahouse.world/glossary/underserved-population.md): An underserved population is a group of people or small businesses with little or no access to mainstream financial services: the unbanked, who have no account, and the underbanked, who have one but still rely on cheque cashers, payday lenders or informal credit because banks' products, prices or checks do not fit them. Financial-inclusion policy and many fintech models target it.
- [Underwriter](https://mahouse.world/glossary/underwriter.md): An underwriter is the investment bank that guarantees a company's share issue: it agrees to buy the shares at an agreed price and resells them to investors, so the company gets its money whatever the market does. In an IPO the underwriters also run the book-building that sets the price, allocate the shares and support the price in the first weeks — for a few per cent of the proceeds and the risk of being left holding the shares.
- [Unicorn](https://mahouse.world/glossary/unicorn.md): A unicorn is a privately held, venture-backed company valued at $1 billion or more at its most recent funding round. The term was coined in 2013, when such companies were rare; there are now well over a thousand, fintech is the largest sector among them, and the valuation the word refers to is the price of the last round's preferred shares, not what the whole company would fetch.
- [Unit Economics](https://mahouse.world/glossary/unit-economics.md): Unit economics is the revenue, variable cost and acquisition cost attached to a single unit of a business — a customer, a transaction, an account or an order — and the profit or loss each unit produces over its life. It shows whether growth creates value or multiplies losses, before fixed costs and scale blur the answer.
- [Unsupervised Learning](https://mahouse.world/glossary/unsupervised-learning.md): Unsupervised learning is machine learning on data that carries no labels: the model finds structure on its own — clusters of similar customers, transactions unlike the rest, a compressed representation that captures what varies — rather than a mapping to answers someone provided. Clustering, anomaly detection, dimensionality reduction and the self-supervised pre-training of foundation models are its forms; it needs no labelling, and it produces structure that has to be interpreted rather than answers that can be scored.
- [Validation Data](https://mahouse.world/glossary/validation-data.md): Validation data is the slice of data held out from training and used during development to make decisions — which hyperparameters, which features, when to stop training, which candidate model — so that those decisions rest on performance on examples the model did not learn from; it is kept separate from the test set, which is touched once at the end to report the result, because data used to choose a model no longer measures it honestly.
- [Valuation Cap](https://mahouse.world/glossary/valuation-cap.md): A valuation cap is the maximum company valuation at which a SAFE or convertible note converts into shares at the next priced round, so an early investor is not diluted into a small stake by a strong round. If the round prices above the cap, the note converts as if the company were worth the cap; if below, at the round price or a discount to it, whichever is better for the investor.
- [Velocity Check](https://mahouse.world/glossary/velocity-check.md): A velocity check is a fraud rule that counts how often, or how much, something has transacted in a window — a card used five times in ten minutes, a device that opened twelve accounts today, €5,000 through one account in an hour — and blocks, challenges or flags the transaction that crosses the limit. It is the oldest rule in the fraud engine and still among the most effective, because fraud is fast and repetitive.
- [Vendor Loan Note](https://mahouse.world/glossary/vendor-loan-note.md): A vendor loan note is part of the purchase price that the seller agrees to be paid later, as a loan from the seller to the buyer, with interest, ranking behind the buyer's bank debt. It bridges a gap between what the buyer can fund and what the seller wants, keeps the seller invested in the business's ability to pay, and is the usual way a founder's exit is financed when the bank will not lend the whole price.
- [Venture Capital](https://mahouse.world/glossary/venture-capital.md): Venture capital is investment in early-stage companies that could grow many times over — minority stakes bought in rounds from seed to late stage, from funds raised from institutions, expecting most companies to fail and one or two to return the fund. It has financed fintech since the 2000s, and its model — power-law returns, staged rounds, preferred shares, a ten-year fund life — shapes every term sheet.
- [Vertical Merger](https://mahouse.world/glossary/vertical-merger.md): A vertical merger combines two companies at different stages of the same supply chain — a payments company buying its processor, a lender buying the data vendor it underwrites on, a software platform buying the acquirer that settles its payments — so that the combined firm controls a step it previously bought. It removes a supplier's margin and a dependency, and the regulator's question is whether it also removes a competitor's access to that step.
- [Vintage Year](https://mahouse.world/glossary/vintage-year.md): A vintage year is the year in which a private fund made its first investment or first called capital — the year that fixes the prices it bought at, the market it will exit into, and therefore the only fair comparison set: a 2021 fund is judged against other 2021 funds, not against a 2016 fund that bought cheaply and sold at the top. Investors spread commitments across vintages for the same reason winemakers do not rely on one harvest.
- [Virtual Account](https://mahouse.world/glossary/virtual-account.md): A virtual account is an account number — an IBAN, a local account number — that holds no funds itself but routes payments into one real account while identifying who paid and for what, so a company can give each customer, invoice or subsidiary its own number and reconcile automatically. Fintechs use virtual IBANs to collect, to offer local account details abroad and to run multi-currency treasury; regulators watch who is behind each number.
- [Void](https://mahouse.world/glossary/void.md): A void is the cancellation of an authorised card payment before it has been captured — the merchant drops the transaction, a reversal goes to the issuer, the hold is released, and no money ever moves. It costs nothing beyond the authorisation fee already paid; a refund returns money that was captured and settled, takes days, and costs fees on both legs. A merchant that voids instead of refunding saves money and complaints.
- [Warranty and Indemnity Insurance](https://mahouse.world/glossary/warranty-and-indemnity-insurance.md): Warranty and indemnity insurance is a policy, usually bought by the buyer, that pays for losses from breaches of the warranties a seller gives in a sale agreement — so that the buyer claims against an insurer instead of the seller, and the seller can walk away from the deal with the price and little continuing liability. It has become standard in private equity exits and common in mid-market fintech deals.
- [WealthTech](https://mahouse.world/glossary/wealthtech.md): WealthTech is technology applied to investing and wealth management: robo-advisers that allocate portfolios automatically, digital brokers that made trading free and mobile, the platforms independent advisers run their businesses on, fractional and direct-indexing tools, access to alternatives and private markets, and the software and data that serve all of them. The consumer wave of the 2010s learned that assets grow slowly and fees on them are thin; the businesses that scaled served advisers or monetised flow.
- [Webhook](https://mahouse.world/glossary/webhook.md): A webhook is a call the payment provider makes to a URL the merchant registered whenever an event occurs — a payment succeeded or failed asynchronously, a dispute opened, a payout sent, a subscription renewed — carrying the event's data so the merchant's system can act without polling. It is how an integration learns what happened after the API call returned, and the part most often built badly.
- [Weighted Average Cost of Capital](https://mahouse.world/glossary/weighted-average-cost-of-capital.md): The weighted average cost of capital is the return a company must earn on its investments to satisfy everyone who funds it — its lenders and its shareholders — blended in proportion to how much of the company each finances, with the cost of debt taken after tax. It is the rate at which a discounted cash flow valuation discounts the company's future cash flows.
- [White Knight](https://mahouse.world/glossary/white-knight.md): A white knight is a friendly acquirer that a board brings in to defeat a hostile bid: a company the board would rather be bought by, which offers a higher price, better terms or a more compatible plan than the unwelcome bidder, and receives the board's recommendation and its cooperation in return. It is the defence that ends with the company sold — to someone else.
- [Working Capital Adjustment](https://mahouse.world/glossary/working-capital-adjustment.md): A working capital adjustment is the change made to an acquisition price for the difference between the target's working capital at completion and a target level — the "peg" — agreed at signing. It stops a seller stripping the business of receivables and inventory before handing it over, and a buyer paying twice for the cash the business needs to run.
- [Yield Curve](https://mahouse.world/glossary/yield-curve.md): The yield curve is the line through the yields of government bonds from the shortest maturity to the longest at one moment. Its normal shape slopes upward, because lenders want more for lending longer; it flattens when the central bank raises short rates; it inverts, short above long, when markets expect rates to fall. Every loan, deposit and valuation is priced off a point on it.
- [Zero-Shot Learning](https://mahouse.world/glossary/zero-shot-learning.md): Zero-shot learning is a model performing a task it was never explicitly trained on, from a description of the task alone, without labelled examples — asking a language model to classify support tickets into categories it has never seen, or CLIP to recognise a document type from its name. Few-shot learning supplies a handful of examples in the prompt. Large pre-trained models made zero-shot practical, and it decides whether a fintech needs a labelled dataset or a well-written sentence.
- [x402](https://mahouse.world/glossary/x402.md): x402 is an open payment protocol, launched by Coinbase in May 2025, that uses the HTTP 402 "Payment Required" status code so that a server can request payment inside an ordinary web request and a client — often an AI agent — can pay it, usually in stablecoins, with no account, subscription or API key. A facilitator settles the payment.