*both systems traded in may & june *both systems trade the US indices *no overnight risk *max daily risk = 2.5% of capital A) 40% winners 60% losers expectancy/contract = 0.6 point made 23% B) 60% winners 40% losers expectancy/contract = 3.25 point made 8%
Chances for the deviation could be explained by: 1) Poor estimation of system expectancy 2) Back luck for system B Assuming your estimation of expectancy is correct, I'll go with B, since the equity should revert to mean. Can't theorize without more data.
Expectancy E = avg. win x prob. of win - avg. loss x probability of loss We only know the probabilities for each system. But Expectancy is an equation with three variables since: prob. of win = 1 - prob. of loss and you need avg. win and avg. loss. Alternatively, given the expectancy and prob. of win or loss, this is an equation with two unknowns, namely avg. win and avg. loss. Thus, the systems cannot be compared. They both belong to a set that contains infinite number of systems. Essentially, "odds" is not known for each system. This is equal to avg. win and it is the amount the system is expected to win when profitable. Expectancy is equal to "edge", the amount won on average. Selection would depend on the ratio of edge/odds. This is also equal to %kelly as Michael Harris shows in the paper below and it is also the ratio that maximizes equity growrh. http://www.tradingpatterns.com/Kelly.pdf Ron
Hi, Could the unexpected result (A better than B) come from the number of positions traded by each system on average, in a month ? If yes, I'd prefer the one with the highest number of trading opportunities. O.
System A uses a 10x bigger positionsize than System B (but with the same risk% of capital). System A trades more than System B.