Win Rate, Average Win, Average Loss, Expectancy in Four Formulas

[Fairman]

Win Rate, Average Win, Average Loss, Expectancy in Four Formulas

Goal: summary statistics from a single column of R results, in cells A2 to A101.

Win rate: =COUNTIF(A2:A101,">0")/COUNT(A2:A101)

Average win: =AVERAGEIF(A2:A101,">0")

Average loss: =AVERAGEIF(A2:A101,"<0")

Expectancy two ways: =AVERAGE(A2:A101), and the decomposition: =winrate*avgwin+(1-winrate)*avgloss, where the names refer to the cells above. The two should agree, which is a good check for errors.

Test with ten results: plus 2, minus 1, minus 1, plus 1.5, minus 1, minus 1, plus 3, minus 1, minus 1, plus 1. Win rate: four of ten, 0.40. Average win: 7.5 divided by 4, which is 1.875. Average loss: minus 1.0. Expectancy: 0.4 times 1.875 minus 0.6 times 1, which is 0.75 minus 0.6, giving plus 0.15R. The plain average: 1.5 divided by 10, also 0.15.

Add the count of trades and the standard error. Standard deviation: =STDEV(A2:A101). Standard error: =STDEV(A2:A101)/SQRT(COUNT(A2:A101)). A t-statistic: expectancy divided by the standard error.

For these ten trades, the standard deviation is about 1.6, the standard error 0.5, and the t-statistic 0.3, meaning the result is indistinguishable from noise.

Place a note beside the sample size: fewer than 30 trades shows a warning.

Keep breakeven trades in mind. A result of exactly zero counts as neither a win nor a loss in these formulas, which is reasonable but worth knowing.

Exercise: apply the four formulas to your journal and read the t-statistic honestly.

[Shadow Trader]

A system that wins 85 percent of the time sounds incredible. It might be terrible. It might also be a ticking bomb.

Scalping makes high win rates easy to manufacture. Set a target of one pip and a stop of ten, and most trades will win. The losers are rare and large, and one of them can erase weeks of small gains.

The number that matters is expectancy, the average amount gained or lost per trade after all costs. A 45 percent win rate with a good payoff ratio beats an 85 percent win rate with a bad one.

Spread distorts the picture further. A one pip target with a half pip spread has given away half the target before the trade starts. Win rate looks fine while the real edge is thin.

Posting a win rate without the average win, average loss and cost per trade says almost nothing. Anyone asking for opinions on a system should include all four.

What's the lowest win rate that has still produced a decent equity curve?