Core terms
Expectancy
The average result per trade a process produces over many trades: (win rate × average win) − (loss rate × average loss), after costs, often expressed in R.
Expectancy combines how often a process wins with how much it wins and loses. A process that wins 40% of the time at +2R, loses 45% at −1R, scratches 15% at 0R and pays 0.1R in costs per trade has an expectancy of 0.40 × 2 − 0.45 × 1 − 0.1 = +0.25R per trade (Illustrative).
A high win rate can hide negative expectancy if losses are larger than wins, and a low win rate can carry positive expectancy if wins are larger. Costs (commissions, fees, spread, slippage) belong inside the formula; for scalps they can be a large share of the average win.
Measured expectancy is an estimate with error. Over 10 or 20 trades the range of plausible values is wide; see trading-sample-size.
What traders should verify
Compute expectancy per setup tag, in R and after costs, from your own log. Report it with the number of trades it is based on.
Do not conflate
Trader usage; there is no official definition, and tools calculate it differently (gross vs. net of costs, scratches included or excluded). A positive expectancy estimate from a short sample can be noise. Expectancy describes an average, not the next trade.
Primary source
Trader usage: there is no official definition, and platforms and tools may calculate it differently.