A high hit rate does not mean someone is making money
Every leaderboard, including the one here, ends up rewarding hit rate because it is easy to measure.
But hit rate on its own says nothing. Ninety percent wins with one catastrophic loss is a losing record, and forty percent wins at good reward-to-risk is a fine one.
I am not saying the number is useless. I am saying it is half a number, and people quote the half that flatters them.
Comments 5
Comments marked "Example" were written by staff to show how a discussion goes. They are not real users.
Log in to comment- Level 1 PrivateStopFirstExample. Written by staff as a worked example of discussion. Not a real user, and not counted in accuracy or rankings.8/17
Agreed, and the two are usually in tension. Tighten the stop and your hit rate falls while your expectancy can rise. People optimise the visible half.
- Level 0 RecruitMeanRevertExample. Written by staff as a worked example of discussion. Not a real user, and not counted in accuracy or rankings.8/17
Counter-trend styles look terrible on reward-to-risk and fine on hit rate, trend styles look the opposite. Comparing two people on either number alone is meaningless.
- Level 1 PrivateCountingItExample. Written by staff as a worked example of discussion. Not a real user, and not counted in accuracy or rankings.8/17
Which is why I log R rather than win/loss. "Won" without a size is not a data point.
- Level 1 PrivateByTheRulesExample. Written by staff as a worked example of discussion. Not a real user, and not counted in accuracy or rankings.8/17
The practical version: publish both, or publish neither. One number invites the wrong optimisation.
- Level 1 PrivateWeekOneChartsExample. Written by staff as a worked example of discussion. Not a real user, and not counted in accuracy or rankings.8/17
So when I look at someone's accuracy here I should be checking their average reward-to-risk next to it. Noted.
The analysis and predictions in this post are the author's own opinion and are not investment advice.