How many trades before your numbers mean anything
By Axelrod
A trader finishes thirty trades, counts nineteen winners, and now believes he has a sixty-three percent system.
He has thirty trades. What he has is a number that will move a lot, in a direction he cannot predict, for a long time yet.
Why the number moves
Take a trader whose real edge wins fifty-five percent of the time. Not a hope, an actual edge, the kind most people would be pleased to own.
Run twenty trades. Roughly a quarter of the time, that trader will finish those twenty with nine winners or fewer, which reads on the page as a forty-five percent system. Same edge, same trader, nothing broken. Twenty trades is simply not enough to tell fifty-five from forty-five.
The same arithmetic runs the other way, which is the more dangerous direction. Twenty trades can just as easily hand you sixty-five percent and a strong feeling that you have found something.
Streaks do the same work on your confidence. In a hundred trades at a coin-flip win rate, a run of five losses in a row is ordinary. It is not a sign the edge has gone, and it will feel exactly like one.
The number nobody quotes
There is no single figure for how many trades you need, because it depends on what you are trying to tell apart. Distinguishing a good system from a bad one is quick. Distinguishing a good system from a slightly better one takes hundreds, sometimes thousands.
The practical version: your win rate and your expectancy are estimates, and estimates need volume. Until you have a few hundred trades under one set of rules, treat both as provisional and stop re-tuning the system every fortnight on the strength of them.
That sounds like a counsel of despair for anyone six weeks into a funded account. It is not, because the estimates are only half of what a log holds.
The half that reads early
Here is the part that gets missed. Some things in your record are not estimates at all. They are counts of what you did.
How many trades did you take outside your plan? How many times did you move a stop away from your entry? How many trades came within ten minutes of a loss? What was the biggest position you took and what came just before it?
None of those is a prediction. They are the trades themselves, and there are three reasons that makes them usable long before a win rate is.
You do not need the rate, you need the instances. Four trades that broke your rules are four trades you can open and read tonight. Whether your true break rate is eight percent or twenty-two does not change what you do about those four.
The action is available either way. Knowing your win rate is fifty-five rather than forty-five decides whether you keep the system. Knowing you moved a stop four times decides nothing about the system. It tells you to stop moving stops, and that is correct at any rate.
And one of them is yours to change. An estimate you cannot act on has to be precise before it is worth anything. An observation you can act on this afternoon does not.
This is why a new trader gets more out of a rules column than a statistics page. The statistics need a year. The rules column needs a fortnight.
What to do with a small sample
Three things hold up when the numbers are still young.
- Judge the process, not the results. Twenty trades will not tell you whether the edge is real. They will tell you, precisely, how often you followed your own plan.
- Change one thing at a time, and mark the date. If you re-tune constantly, you never accumulate a clean sample of anything, and after a year you still have thirty trades of every version.
- Write down what would make you abandon this system, before the drawdown. A rule written while calm survives a losing streak. A decision made during one is the streak talking.
The trader with thirty trades and a sixty-three percent win rate is not wrong to be encouraged. He is wrong to plan around the number.
What he can plan around, today, is the other column: the four trades in those thirty that he took outside his own rules, which he can count on one hand and fix this week.