The idea, simply put: most of the numbers trading platforms display mean nothing in isolation. A Sharpe ratio without the max drawdown next to it, a win rate without the expectancy, a backtest without real costs: so many traps. This glossary gives you the exact meaning of each term, and above all how to read it without being misled.
Quantitative trading means making decisions from rules and data rather than gut feeling. The downside is a dense vocabulary — often used to impress rather than to clarify. At Sextant, a transparent quant platform, the guiding principle is the opposite: every technical term must be explainable simply, otherwise it is hiding something.
The terms are grouped into six families. Those that already have a detailed entry or a full article are linked; the rest are coming over the next few weeks.
Browse by family
Risk-adjusted performance ratios
Sortino ratio
A variant of the Sharpe that only penalises downside volatility. Logical: it is the drops that hurt, not the gains.
Read the entryCalmar ratio
Annualised return divided by the maximum drawdown over 3 years. How much a strategy returns relative to the worst it put you through.
Read the entryProfit factor
The sum of gains divided by the sum of losses. Above 1, the strategy wins more than it loses — regardless of the win rate on display.
Read the entryExpectancy
The average gain expected per trade. It, not the percentage of winning trades, decides whether a strategy is profitable.
Read the entryPayoff ratio
The average winning trade divided by the average losing trade. Paired with the win rate, it tells the whole story of a strategy's viability.
Read the entryMAR ratio
Entry coming soonA cousin of the Calmar, but computed over the strategy's whole life rather than a rolling 3 years. On a young strategy it can look too favourable.
CAGR (annualised return)
Entry coming soonThe compound annual growth rate of capital over a given period. Always read it alongside the risk taken.
Risk & losses
Drawdown
The worst decline recorded from a peak, before recovery. The risk you actually live through — the one that makes people quit at the worst moment.
Read the full articleVolatility
The standard deviation of returns: a measure of noise. Useful for comparison, but blind to the depth and sequence of losses.
Read the entryRecovery time
Entry coming soonThe number of months needed to recover capital after the worst drawdown. A -20% recovered in 6 weeks is nothing like a -20% recovered in 14 months.
Value at Risk (VaR)
Entry coming soonThe maximum loss expected over a given period, at a given confidence level. Useful, but it says little about the remaining extreme scenarios.
Risk of ruin
Entry coming soonThe probability of losing a large enough share of capital to never recover. The blind spot that average returns almost always hide.
Trade performance
Win rate
The percentage of winning trades. The most advertised number… and the most misleading on its own: a '90% win rate' can hide a losing strategy.
Read the entryEquity curve
Entry coming soonCapital over time. The most honest visual read of a strategy: a smooth, steady curve is worth all the ratios.
Execution & costs
Slippage
The gap between the expected price and the price actually obtained at execution. A very real cost, often forgotten by over-optimistic backtests.
Read the entrySpread
The difference between the best bid and the best ask. A cost paid on every entry and every exit.
Read the entryTrading fees
Entry coming soonThe commissions the exchange charges on each order. Decisive on an active strategy — an honest backtest always includes them.
Position sizing
Entry coming soonThe amount of capital committed to a trade. Sizing well to volatility often matters more than the exact entry timing.
ATR (Average True Range)
Entry coming soonA measure of an asset's recent volatility. Used to adjust position size: the choppier the market, the smaller the position.
Model & artificial intelligence
Probability calibration
The adjustment that makes a model's stated confidence match its real success frequency. A well-calibrated model that says '70%' is right 70% of the time.
Read the full articleDrift / PSI
Entry coming soonA model's drift as the market changes and no longer resembles its training data. The PSI measures that gap — a signal it is time to retrain.
Market regime
The market's general state — bullish, bearish or neutral. At Sextant it is read by a transparent rule based on trend and volatility.
Read the guideOverfitting
Entry coming soonWhen a model fits past data so tightly it learns the noise rather than the signal — and collapses on new data.
Backtest
Entry coming soonTesting a strategy on historical data. Reliable only if it includes real costs (fees, slippage, spread) and is validated on unseen data.
Walk-forward / out-of-sample
Entry coming soonTesting a strategy on periods it was never built on. The best defence against overfitting.
Feature (market indicator)
Entry coming soonAn input variable the model uses to decide. Sextant's engine tracks 22 market indicators.
Random Forest
Entry coming soonAn AI model made of a forest of decision trees whose votes are aggregated. At the heart of Sextant: a calibrated AI model that knows when to abstain.
Abstention (no-trade rate)
The share of time a system deliberately chooses not to trade, for lack of a reliable enough signal. Staying out of the market is a decision in its own right.
Read the full articleMethod
Paper trading
Trading in real conditions but with fictitious capital. The best way to test a strategy without risking a cent until confidence is established.
Going further
These terms make full sense once you see how they fit together in a method. Our guide lays the foundations of quant reasoning, step by step.
The quant guide for beginnersFrequently asked questions
What is a quant glossary?
It is a dictionary of the terms used to measure and compare quantitative trading strategies: performance ratios (Sharpe, Sortino, Calmar), risk metrics (drawdown, volatility), execution costs (slippage, spread) and model concepts (calibration, drift, market regime). Understanding these terms lets you read honestly what a trading system really does — instead of trusting a single isolated number.
What is the difference between the Sharpe ratio and the Calmar ratio?
The Sharpe measures return per unit of total volatility — a strategy's noise. The Calmar measures return relative to the maximum drawdown endured — the real pain. A strategy can show a great Sharpe and a poor Calmar if it suffers a rare, sharp collapse. The two are complementary, never interchangeable.
What is a good Sharpe ratio?
By convention, a Sharpe above 1 is decent, above 2 good, above 3 excellent. But the figure depends on the calculation window and the asset class. In crypto, which is very volatile, a Sharpe should always be read with the maximum drawdown and the recovery time alongside it — never on its own.
Why is a high win rate not enough?
Because profitability depends on expectancy, not on the percentage of winning trades. A strategy that wins 90% of the time but loses big on the remaining 10% can be a net loser. Conversely, a win rate of 52 to 58% can be highly profitable if average gains exceed average losses. The win rate is never read without the payoff ratio.
Do these metrics guarantee gains?
No. They are tools for reading and comparison, not promises. A metric computed on the past describes what happened — it does not predict the future. Trading involves a risk of capital loss.