Trading Expectancy vs Win Rate: What the Numbers Mean

Winning often and making money are different measurements

A trader who wins nine small trades and loses heavily on the tenth has a high win rate. Whether the sequence made money depends on amounts. Win rate is wins divided by the number of evaluated trades. Expectancy is the sample’s average outcome per trade, expressed in currency, percentage return or a defined risk unit. It brings frequency and magnitude into the same calculation.

Keep the trade definition stable. Is one trade an entire position from entry to final exit, or is every partial exit a separate observation? Splitting one successful position into five winning exits while counting a losing position once changes the reported win rate without changing the account outcome. State the grouping rule before calculating either metric.

CME’s explanation of trading expectancy provides the general framework. The numerical examples and reconciliation process here are illustrative calculations, not observed TradeCopier performance or evidence for a particular strategy.

The expectancy formula

For a two-outcome model, expectancy equals win probability multiplied by average win, minus loss probability multiplied by average loss. Average loss is entered as a positive magnitude. If the win and loss amounts exclude costs, subtract average cost per trade. If they already include all costs, do not subtract those costs again.

For a historical sample, use observed frequencies rather than calling them known future probabilities. A direct cross-check is total net result divided by the number of trades. With consistent trade grouping and complete cost records, that average should reconcile with the frequency-weighted calculation.

Breakeven trades require explicit handling. If they count as observations, the probabilities of wins, losses and breakevens sum to one. A trade with zero price movement may still lose money after commission. Calling it breakeven before costs and then excluding it from the sample understates the drag from activity.

Example: a 70% win rate that loses money

Imagine 100 closed trades: 70 win $40 each and 30 lose $120 each before costs. Gross winnings are $2,800 and gross losses are $3,600. The gross result is −$800, or −$8 per trade. A $2 round-trip cost on each trade produces another $200 of expense, leaving −$1,000 in total and net expectancy of −$10 per trade.

The arithmetic is 0.70 × $40 − 0.30 × $120 − $2 = −$10. The high win rate does not rescue the unfavorable relationship between gain size and loss size. A promotional claim that highlights only the 70% figure omits the quantities that determine the result.

Example: a lower win rate with positive sample expectancy

A second hypothetical set has 40 wins averaging $180 and 60 losses averaging $80. Before costs, the result is $7,200 minus $4,800, or $2,400. At the same $2 cost per trade, net result is $2,200. Net expectancy is $22 per trade even though the sample win rate is only 40%.

Sample	Win rate	Average gross win	Average gross loss	Net expectancy
A	70%	$40	$120	−$10
B	40%	$180	$80	+$22

This comparison demonstrates arithmetic, not a recommendation to seek a particular win rate. Sample B could have a painful losing sequence, concentrated exposure or unrealistic fill assumptions. It also says nothing about whether these averages will persist. A distribution contains more information than its mean.

Calculate the breakeven win rate correctly

With average gross win W, average gross loss L and a constant per-trade cost C, solve pW − (1 − p)L − C = 0. The breakeven win rate is (L + C) divided by (W + L). For the second example, (80 + 2) / (180 + 80) is about 31.54%. Using gross figures but ignoring costs would give 30.77%.

This threshold assumes the same average amounts and cost convention. If costs depend strongly on whether a stop or target fills, model outcome-specific costs instead. If losses can exceed the planned stop, historical average loss may differ from the planned risk in a chart drawing. The breakeven win-rate calculator is an arithmetic aid, not a probability estimator.

Currency expectancy and R expectancy

Dollar averages become difficult to interpret when position size changes. A $300 gain with $600 initially at risk and a $60 gain with $30 initially at risk represent different trade outcomes. Expressing each result as an R multiple can help compare decisions, provided the original risk amount was recorded consistently.

Do not calculate total profit divided by today’s usual risk amount and call that an average R per trade. Compute each trade’s net result divided by its own documented initial risk, then average those values. Keep the currency totals alongside them. Average R and average dollars can move differently because they answer different questions about sizing and execution.

A position whose initial risk was not defined cannot honestly receive a precise R value after the outcome is known. Label it missing and investigate the process failure. Filling the gap with the eventual loss would make a badly controlled trade look like an ordinary one-R loss.

Check sample quality before trusting the average

Start with completeness: include all completed trades, financing, commissions and adjustments under the stated grouping rule. Keep rejected or cancelled requests in a separate operational record; an unfilled request is not a completed trade with zero P&L. Reconcile the sample with the account statement. If deposits funded a recovery, keep them separate from trading profit. If several trades were legs of the same idea, document their relationship instead of assuming independent observations.

For example, suppose ten intended instructions produce eight completed trades with a combined net result of $24, and two instructions never fill. The completed-trade average is $24 / 8 = $3. Dividing the observed total by all ten instructions gives $2.40 per intended instruction, a different operational measure that must be labeled separately. Neither calculation tells you what the rejected trades would have earned. Report the rejection count alongside realized performance and investigate its cause without inventing counterfactual fills.

Next inspect concentration. Suppose a 50-trade sample earned $1,000 but its largest trade earned $1,400. The remaining trades lost $400. That does not justify deleting the winner, but it changes the question: how much evidence exists that the process can repeatedly capture comparable opportunities? Report the full sample and this sensitivity observation together.

Then examine stability across preselected periods or market conditions. Avoid inventing dozens of subgroups and advertising only the profitable one. A distinction such as planned versus unplanned trades can be operationally useful if it was logged at the time. A label added after seeing the outcome is vulnerable to hindsight.

Why copied accounts can have different expectancy

A follower can receive the same direction yet have a different entry, exit, size or commission. Small trades may round to a minimum size or be skipped because they cannot be represented within a risk limit. Currency conversion can also alter account-currency results. The master’s expectancy is therefore not a guaranteed follower statistic.

Compare matched trade identifiers and timestamps using copying activity logs, then reconcile actual follower fills with its statement. Separate omitted trades from filled trades with different prices. Combining these two problems into one average slippage figure makes troubleshooting harder.

The slippage cost calculator can illustrate a price difference in monetary terms. Record whether quoted commission is one-way or round-trip. A factor-of-two cost error can materially change expectancy for a strategy with a narrow historical margin.

Turn measurement into a review routine

1. Export complete trades for a fixed period and apply a consistent position-grouping rule.
2. Reconcile net totals and record costs without double counting.
3. Calculate win rate, average win, average loss, expectancy and the number of observations.
4. Inspect concentration, drawdown and differences between source and follower execution.
5. Write down what requires another controlled test before changing live settings.

Use the expectancy calculator to reproduce a scenario and the journal workflow to preserve the underlying evidence. A sample average supports investigation. It does not establish a stable edge, tell you how much personal capital to commit, or remove the possibility of loss.