How to Calculate Trading Expectancy and Test Whether Your Edge Is Real

Asking 'did I make money this month?' tells you very little about whether a strategy works. A good month can be variance, one outsized trade or luck. A bad month can be the same in the other direction. Expectancy is the average result per trade, and it is the first number to compute from your trade log.

The formula

Expectancy = (win rate x average winner) - (loss rate x average loser)

Win rate is wins divided by total trades, loss rate is 1 minus that, and the average loser is entered as a positive number. Example, with hypothetical numbers: 55% win rate, $300 average winner, $200 average loser.

(0.55 x $300) - (0.45 x $200) = $165 - $90 = $75 per trade.

Use net figures: subtract commissions and fees from every trade before averaging, and include realistic slippage if you are working from simulated trades.

Expectancy per dollar risked

Dollar expectancy depends on size. Divide by the average amount risked to compare across sizes: expectancy ratio = (win rate x (average winner / average loser)) - loss rate.

Same example: (0.55 x 1.5) - 0.45 = 0.375. Each $1 risked returned an average $0.375, so $200 risked per trade matches the $75 above. Strictly, the R-multiple version divides by the planned risk per trade, so it matches the average-loser version only when losers are close to the planned stop.

A related fact: with a 1.5 average winner to loser ratio, the break-even win rate is 1 / (1 + 1.5) = 40%. Below that win rate before costs, the strategy loses money on average, and costs push that threshold higher.

How big is good?

No universal threshold exists. A small positive number can disappear after commissions and slippage, so it matters that your costs are included. A very large number from a backtest deserves suspicion, because overfitting and unrealistic fills produce attractive figures that rarely survive live trading. The practical question is whether the estimate stays positive after costs and across fresh data.

Calculating it from your log

1. Export trade history from NinjaTrader or your broker as a CSV with entry, exit, quantity and direction.
2. Compute each trade's net P&L in dollars.
3. Win rate = winners / total trades.
4. Average winner = sum of winning P&L / number of winners.
5. Average loser = sum of absolute losing P&L / number of losers.
6. Apply the formula. For NinjaTrader users, the Trade Performance report is the place to pull the figures from.

The edge calculator will do the arithmetic for you.

Sample size and uncertainty

The standard error of a win rate is the square root of (p x (1 - p) / n). At a 55% observed win rate with 30 trades, that is about 9 percentage points, so a rough 95% range is about 37% to 73%. With 50 trades it is about 7 points, giving roughly 41% to 69%. Even at 50 trades, the true win rate could be below the 40% break-even from the example above, and the average winner and loser vary too.

So 30 trades is a first look and 50 is only somewhat better. Neither proves an edge, and 50 simulated trades are not live proof either, because simulation does not reproduce queue position, slippage or the way real money changes behavior. More trades narrow the range, and a different market regime can still change the result.

Expectancy versus profit factor

Profit factor is gross profit divided by gross loss. With a 55% win rate and a 1.5 winner to loser ratio, the hypothetical numbers give (0.55 x 1.5) / 0.45, about 1.83, and an expectancy ratio of 0.375. They describe the same record from different angles. Profit factor is quick to read, and expectancy feeds sizing decisions. Track both, because neither captures drawdown.

Using the number

Continue or stop. If a decent live sample shows expectancy near zero or negative after costs, you have no evidence of an edge. Find out whether execution, market conditions or a sample that was too small and lucky explains the earlier results.

Sizing. The Kelly formula uses these inputs: fraction = win rate - loss rate / (winner to loser ratio). For the example, 0.55 - 0.45 / 1.5 = 0.25. Treat this as an upper bound, not a recommendation. It assumes your estimates are exact and the outcomes are fixed and independent, and errors in a small sample inflate it. Fractional Kelly is commonly used for that reason, and variable dollar risk or drawdown limits need simulation instead of the simple formula.

Comparing approaches. Win rate alone misleads. A hypothetical scalping style with a 70% win rate and a 0.5:1 reward to risk has an expectancy ratio of (0.70 x 0.5) - 0.30 = 0.05. A hypothetical swing style with a 45% win rate and 2.5:1 has (0.45 x 2.5) - 0.55 = 0.575. The lower win rate wins here, though both are made-up figures and real costs would shrink the first one most.