Expectancy and profit factor: what your strategy is really worth

7 min read

Win rate alone does not tell you whether a strategy is good: you can win 70% of the time and lose money, or win 35% and come out ahead. Two figures summarize expected performance far more honestly: expectancy and profit factor. This guide shows how to calculate them, how to interpret them and how far you can trust them.

The expectancy formula

Expectancy is what you win or lose, on average, per trade over the long run. The formula is E = (WR × average win) − ((1 − WR) × average loss), where WR is the proportion of winning trades (win rate), expressed between 0 and 1. If E is positive, the strategy wins on average; if it is negative, it loses, however good it looks.

You can express it in money, but it is more useful in R multiples, where 1 R is what you risk on each trade. That lets you compare systems with different account sizes or position sizes. If E = 0.2 R, then for every 100 USD risked you expect to earn 20 USD on average, after accounting for losses.

First example: winning rarely and still profiting

A strategy wins 40% of the time, makes 2 R on average when it wins and loses 1 R when it fails. E = 0.40 × 2 − 0.60 × 1 = 0.80 − 0.60 = +0.20 R. With 100 trades and 100 USD risked on each, expectancy is 20 USD per trade, or 2,000 USD in total (an average, not a guarantee).

Check it against a sample of ten trades in R: +2, −1, −1, +2, −1, +2, −1, −1, +2, −1. Four win (+8 R) and six lose (−6 R); the net is +2 R, which divided by 10 gives 0.20 R, and the win rate is 40%. Notice there are two pairs of consecutive losses in just ten trades: that is normal for this kind of system.

Second example: winning often and losing money

Now a system that wins 60% of the time but makes only 0.5 R on average and loses 1 R when it fails. E = 0.60 × 0.5 − 0.40 × 1 = 0.30 − 0.40 = −0.10 R. Even though it wins six times out of ten, it loses 0.10 R per trade: with 100 trades and 100 USD of risk, that is −1,000 USD on average.

This is a very common pattern among traders who close winners early out of fear and let losers run. The lesson is that win rate on its own is a misleading number: it only makes sense together with the relationship between average win and average loss.

Relationship with the risk/reward ratio

The minimum win rate needed to avoid losing money (the break-even point) depends on the R:R ratio: minimum WR = 1 ÷ (1 + R:R). With 1:1 you need 50%; with 1:2, 33.3%; with 1:3, 25%; and with 1:0.5 (you win half of what you risk), 66.7%. Any win rate above that point gives positive expectancy before costs.

That is why there is no universal “good” win rate. Trend-following systems tend to have low win rates (30 to 40%) with large wins, and mean-reversion systems high win rates with small wins. Either can be profitable or not: E decides. You can calculate the R:R of each trade with our risk/reward calculator.

The profit factor

The profit factor (PF) is the ratio of gross profit to gross loss: PF = sum of all gains ÷ sum of all losses (in absolute value). A PF of 1 is break-even, above 1 is profitable and below 1 is losing. In the first example: gains = 0.40 × 2 = 0.80 R per trade and losses = 0.60 R, so PF = 0.80 ÷ 0.60 ≈ 1.33. In the second: 0.30 ÷ 0.40 = 0.75.

There is a direct link with expectancy: PF = 1 + E ÷ ((1 − WR) × average loss). In the first example, 1 + 0.20 ÷ 0.60 = 1.33, the same as before. Expectancy tells you how much you earn per trade; PF tells you how much you earn for each unit lost. With real data PF is very sensitive to one or two extreme trades, so do not rely on it alone.

A third example, in money terms

You do not have to use R. If your record shows a 45% win rate, an average win of 150 USD and an average loss of 100 USD: E = 0.45 × 150 − 0.55 × 100 = 67.50 − 55 = +12.50 USD per trade, and PF = 67.50 ÷ 55 ≈ 1.23. That equals 0.125 R if your average risk is 100 USD.

A trend-following case: a 30% win rate, an average win of 3.5 R and a loss of 1 R. E = 0.30 × 3.5 − 0.70 = 1.05 − 0.70 = +0.35 R, and PF = 1.05 ÷ 0.70 = 1.50. Notice that with only three wins out of ten the edge is bigger than in the first example, but losing streaks will be longer: the probability of five losses in a row is 0.70^5 ≈ 16.8%.

Statistical error with small samples

The expectancy you compute from your record is an estimate, not the true value. For the first example (40%, +2 R / −1 R), the variance per trade is 0.40 × 4 + 0.60 × 1 − 0.20² = 2.16, a standard deviation of 1.47 R. The standard error of the mean is 1.47 ÷ √n. With 30 trades, ±0.27 R (a 95% interval of ±0.53 R); with 100, ±0.15 R (±0.29 R); with 400, ±0.07 R (±0.14 R).

In plain terms: with 100 trades, a measured expectancy of +0.20 R is compatible with a true value of −0.09 R or of +0.49 R. In money, with 100 USD of risk, the expected total of +2,000 USD has a 95% range of roughly −880 to +4,880 USD. That is why one good month or one good week proves nothing.

How big a sample you need

For the lower bound of the 95% interval to sit above zero when the true expectancy is 0.20 R, you need n ≥ (1.96 × 1.47 ÷ 0.20)² ≈ 207 trades. To have an 80% chance of detecting that edge, n ≈ (2.8 × 1.47 ÷ 0.20)² ≈ 423. And the smaller the edge, the larger the sample: if the true expectancy were 0.10 R, the number would roughly quadruple.

The practical takeaway: a few hundred trades is a minimum, not a luxury. Until then, treat results as provisional and do not increase size out of enthusiasm. While you build up a sample, a trading journal lets you measure your real E and PF from your own data.

Sensitivity to costs

Costs are subtracted from every trade, winner or loser. In the first example, if each trade costs 0.05 R, winners become +1.95 R and losers −1.05 R: E = 0.40 × 1.95 − 0.60 × 1.05 = 0.78 − 0.63 = 0.15 R, and PF = 0.78 ÷ 0.63 ≈ 1.24. With a 0.15 R cost: 0.74 − 0.69 = 0.05 R and PF ≈ 1.07. At 0.20 R per trade, expectancy is exactly zero and PF = 1.

The lesson is that a tight-stop system can lose its entire edge to commissions and spread. Always measure net E and PF with real costs, and check how well they hold up if costs rise.

Common mistakes and summary

Computing E without costs; mixing trades with different risk without converting them to R; using a sample of 20 or 30 trades; removing “odd trades” from the record; trusting a high PF with few trades; and confusing positive expectancy with the absence of losing streaks. Positive E does not prevent streaks, it only makes the long-run trend upward.

Summary: calculate E = WR × average win − (1 − WR) × average loss in R, and PF = gross profit ÷ gross loss; check that they clear break-even with real costs; estimate the error from the standard deviation and sample size; and demand several hundred trades. You can run these calculations with this site’s expectancy and profit factor calculator, which works in your browser and does not store your data.

Educational content, not financial advice. Risk disclosure

Related tool

Expectancy calculatorFind out how much your system makes or loses per trade on average, its profit factor and the break-even win rate.

More guides