FRM® glossary: M
15 terms beginning with M.
- Machine learning
- Methods that find patterns in data to predict outcomes without being explicitly programmed with rules. In risk, they raise issues of explainability and model risk. Where it appears →
- Margin call
- A demand to post more collateral when the value of a position moves against the holder. Failure to meet it can lead to close-out. Where it appears →
- Margin period of risk
- The time between the last exchange of margin and the point at which a defaulted counterparty's positions are closed out and replaced. A longer period means larger potential exposure. Where it appears →
- Marginal VaR
- The change in portfolio VaR from a small change in one position. Component VaR splits total VaR into parts that add up to it. Where it appears →
- Mark-to-market
- Revaluing a position at current market prices and settling the change, usually daily. It stops losses from building up unseen. Where it appears →
- Market discipline
- The pressure on banks from investors and counterparties who can see their risks. Basel Pillar 3 requires disclosures to support it. Where it appears →
- Market risk
- The risk of loss from changes in market prices such as interest rates, exchange rates, equity prices and commodity prices. It is measured on trading and other marked positions. Where it appears →
- Master agreement
- A standard contract that governs all derivative trades between two parties and enables close-out netting. The ISDA Master Agreement is the usual form. Where it appears →
- Maturity mismatch
- Funding long-term assets with short-term liabilities. It is the core of banking and the source of its liquidity risk. Where it appears →
- Mean reversion
- The tendency of a variable to drift back towards a long-run level. Interest rates and volatility are often modelled this way. Where it appears →
- Model risk
- The risk of loss from errors in a model or from using it wrongly. Validation and independent review control it. Where it appears →
- Model validation
- Independent testing of whether a model is sound, accurate and suitable for its use. It covers assumptions, data, performance and limitations. Where it appears →
- Monte carlo simulation
- A method that generates many random scenarios from a chosen model and revalues the portfolio in each. It handles non-linear positions but depends on the model and on enough runs. Where it appears →
- Moral hazard
- The tendency to take more risk when someone else bears the cost of losses. Insurance and implicit bailouts are common sources. Where it appears →
- Multicollinearity
- Strong correlation among the explanatory variables in a regression. It inflates standard errors and makes individual coefficients hard to interpret. Where it appears →
Learn these terms in context, with worked examples and questions.
See the FRM course →