Glossary
Quant trading terms, clearly explained
Concise, neutral explanations of common concepts in quantitative trading.
- Algorithmic trading
- Automatically sending trading orders through computer programs that follow predefined rules. Quantitative strategies are often executed this way.
- Backtesting
- Simulating how a strategy would have performed on historical data to test whether an idea holds up. Backtest results don't indicate future performance.
- Circuit breaker
- A protective mechanism that automatically pauses trading or execution during violent market moves or system anomalies.
- Live trading
- Executing trades with real capital in real markets, as opposed to paper trading or test environments.
- Machine learning (in trading)
- Models that learn patterns from historical data to identify market regimes, assess opportunities or estimate risk; they require rigorous validation to avoid overfitting.
- Maximum drawdown
- The largest decline in value from a peak to a subsequent trough, used to gauge the worst loss a strategy may experience.
- Out-of-sample testing
- Validating a strategy on data that was never used during development or tuning — a common check against overfitting.
- Overfitting
- When a strategy or model fits noise in historical data rather than repeatable patterns, causing it to perform poorly on new data.
- Position sizing
- Rules that decide how much capital or exposure to commit to each trade — one of the core tools of risk control.
- Quantitative trading
- A trading approach that makes decisions with data, statistics and explicit rules, usually executed by software, so decisions are verifiable and applied consistently.
- Slippage
- The difference between the expected and actual execution price, common when liquidity is thin or prices move quickly.
- Stop-loss
- A predefined exit condition that closes a position once losses reach a set level, limiting the loss on a single trade.
- Trading signal
- A buy, sell or adjust instruction produced by a strategy from data and rules; typically it must pass risk checks before execution.