Trade expectancy is the average profit or loss per trade after combining how often a strategy wins with how much it wins, and how often it loses with how much it loses.

Trade expectancy is the average profit or loss per trade after combining how often a strategy wins with how much it wins, and how often it loses with how much it loses. In plain terms, it answers a simple question: what does one trade earn or lose on average once both frequency and magnitude are accounted for.[1]

A standard way to express expectancy is:

Expectancy = (probability of win × average win) − (probability of loss × average loss).[1]

This mechanism matters because win frequency alone is incomplete. A strategy can win often but still have poor expectancy if its losses are larger than its gains. The reverse can also be true. A strategy can win less often but still have positive expectancy if its average win is large enough relative to its average loss.[1]

An illustrative calculation shows how the parts combine. Suppose a set of trades has a 60 percent win probability, an average win of 2 units, a 40 percent loss probability, and an average loss of 1.5 units. The expectancy is:

(0.60 × 2.0) − (0.40 × 1.5) = 1.2 − 0.6 = 0.6 units per trade.

That result means the average trade produces 0.6 units after accounting for both the frequency and the size of wins and losses. If the same win rate were paired with larger average losses, expectancy would fall. If the same payoff sizes were paired with fewer wins, expectancy would also fall. Expectancy therefore works as a compact summary of trade distribution, not just hit rate.[1]

For researchers, expectancy should be treated as an estimate from sample data rather than a fixed truth. A backtest can overstate edge when model choices are heavily tuned to historical data. Sonar’s backtest overfitting audit is designed to examine whether backtest results are robust or inflated by data mining and selection effects, which is directly relevant when interpreting estimated expectancy from a historical sample.[2] A robust expectancy estimate is one that remains credible after accounting for the risk that in-sample results were optimized too aggressively.[2][3]

This is also why expectancy is best read alongside tools that evaluate statistical reliability. Sonar’s research glossary discusses the Deflated Sharpe Ratio as a way to adjust performance assessment for multiple testing and non-normal returns, reinforcing the broader principle that apparent backtest quality can be overstated if selection bias is ignored.[3] Expectancy tells you the average trade outcome in a compact form. Robust validation helps determine whether that estimate is likely to generalize beyond the sample used to calculate it.[2][3]

Covered in depth in the Strategy research fundamentals pillar hub.