A strategy's win rate is not its edge. Trade expectancy combines how often its trades win with the size of its average wins and losses, producing an average result per trade—not a promise about the next one.
The calculation behind trade expectancy
Trade expectancy is the average profit or loss per closed trade, estimated from a set of results. A common form is:
expectancy = (win rate × average win) − (loss rate × average loss)Use win and loss rates as fractions, and enter average loss as a positive magnitude. Keep every input on the same basis: same trade sample, same unit, and a consistent treatment of costs. ThinkCapital's expectancy guide and TradeReview's calculator guide describe the same relationship between the two rates and average trade outcomes.
Why the win rate cannot answer the question
Win rate counts winning trades, but it ignores how large each win or loss was. Expectancy adds those missing sizes, so a trader can compare systems that win at different frequencies without treating hit rate as a verdict. Both ThinkCapital and TradeReview frame expectancy as an average across the wins and losses, rather than a synonym for win rate.
For a simple hypothetical, imagine a rule set with a 40% win rate, an average win of 2R and an average loss of 1R, where R is the example's chosen unit of risk. Its estimated expectancy is (0.40 × 2R) − (0.60 × 1R) = +0.20R per trade. Those inputs are illustrative, not market observations or a claim about a real strategy. Change the payoff or the loss rate and the result changes; the arithmetic does not certify that either estimate will persist.
What the average leaves out
A positive historical expectancy does not describe the order of outcomes. A run of losses can arrive before the winners, and an average does not show how deep that run might be. ThinkCapital explicitly distinguishes the average from losing runs and says historical expectancy does not promise future profit; TradeReview likewise cautions against judging a system from a single lucky streak. Read both limits alongside the formula: ThinkCapital · TradeReview.
The estimate also inherits the choices behind its inputs. Which trades count? Are partial exits grouped as one position? Are costs included in each result or deducted afterward? Changing those conventions can change the calculated average without changing the underlying signal. State the convention before comparing two expectancy figures.
How you'd actually test it
Turn the concept into a reproducible report, not a target to chase:
- Freeze the entry, exit and trade-grouping rules before collecting results.
- Calculate win rate, average win and average loss from the same closed-trade sample and in the same units.
- Record whether each result is before or after spread, commission, swap and slippage; do not mix bases or deduct a cost twice.
- Keep the trade sequence. Review losing runs and drawdown separately instead of asking expectancy to answer a path-risk question.
- Recalculate on data not used to design the rules, then compare the estimate with the original sample.
A larger trade count alone does not settle whether an estimate is dependable. Our explainer on how many trades a backtest needs covers sample uncertainty; the methodology page shows why a reproducible test needs its assumptions visible.
Frequently asked
Is positive expectancy proof of a profitable strategy?
No. It describes the average of a defined historical sample. It does not establish that the estimate will hold in new data or describe the sequence of losses.
Can a strategy with a low win rate have positive expectancy?
Yes, if its average wins are large enough relative to its average losses. The formula makes that trade-off explicit; it does not predict future results.
Takeaway
Expectancy is a useful summary of the trade log, not a substitute for reading it.