Trading Statistics Explained: The Numbers That Describe a Record

A trading record compresses into a handful of numbers, and every serious decision — keep the strategy, fix the exits, cut the size — should be made by reading them together. Here is the set, what each one means, and the traps in each.

R multiples: the common unit

R is the money risked on a trade — entry to stop, in cash. Results expressed as multiples of it (+2R, −1R) make every trade comparable regardless of instrument or size, and make the rest of the statistics meaningful. A record kept in dollars with wildly varying risk is a diary; a record kept in R is data. (The arithmetic lives in position sizing.)

Win rate

wins ÷ decisive trades. Alone it says almost nothing — a 90% win rate loses money if the tenth trade gives back more than the nine paid. Scratches (≈0R) are best excluded from the ratio and counted separately, or they flatter it.

Average win, average loss

The mean R of winners and of losers. Their ratio is your payoff shape, and it trades off against win rate: trend methods run low win rate with big average wins; mean reversion the reverse. The dangerous pattern is the drift — average win shrinking (profits taken early) while average loss grows (stops given "room").

Expectancy: the headline

expectancy = (win rate × average win) − (loss rate × average loss), in R per trade. Positive expectancy after costs is the definition of an edge. Two caveats: it needs a sample — treat any expectancy from under ~50 trades as a rumour — and it is a mean, so one outlier can carry it; look at the distribution, not just the number.

Profit factor

gross profit ÷ gross loss. Below 1.0 the system loses; around 1.5 is workable; well above 2 on a large sample is strong (and on a small one, suspicious). It compresses the same information as expectancy from another angle — when they disagree with your impression, the record is telling you your impression is wrong.

Drawdown and streaks

Maximum drawdown and longest losing streak are the record's pain numbers — the cost of earning the expectancy. They set position size boundaries and, known in advance, are what let you sit through a normal bad run without abandoning a working method.

Read them as one instrument

Each number alone can lie; together they triangulate. Rising win rate with collapsing average win is exit fear, not improvement. Healthy expectancy with a deepening drawdown is a sizing problem. Great backtest numbers with poor live ones, on the same rules, is an execution gap — and the on-plan percentage from your discipline record will usually name it.