How many replay sessions before a strategy means anything?

The question comes up after the first good session. Three winners, one small loser, the equity curve pointing the right way, and the natural thought: is this it? The honest answer is that four trades tell you almost nothing, and the useful answer is how to know when you have enough.

Why small samples lie

Every strategy has an underlying win rate you cannot see directly. What you see is a sample drawn from it, and small samples wander. A coin that lands heads six times in ten is not a biased coin; it is a normal coin having a normal run. A strategy with a genuine 45% win rate will regularly produce a stretch of ten trades with seven winners, and just as regularly produce ten with three.

If you judge after one session you are judging the run, not the strategy.

A working rule of thumb

There is no exact threshold, but a widely used starting point is around thirty trades per condition you care about before the numbers start to settle. “Per condition” is the part people miss. Thirty trades in total, half of them on gold and half on EUR/USD, taken across two different years, is not thirty trades of one thing — it is fifteen of two things, and neither half is big enough on its own.

So the practical version is: decide what you are actually testing, keep everything else fixed, and collect thirty or more of *that*.

What TradifyFX does to help

The strategy name is the unit of measurement. Every session, trade and statistic is filed under it, so the Analytics page can show you one idea’s profit factor rather than an average of three ideas that describes none of them. Use the same name every time you test the same rules, and a different name the moment you change them.

Then look at the shape of the lifetime P&L curve rather than the end point. A curve that climbs steadily through thirty trades and one that gets there on the back of two big winners finish in the same place and mean opposite things. The drawdown curve underneath tells you whether you could have sat through it.

When results stop moving

You will know you have enough when adding another session changes the profit factor by a little rather than a lot. Early on, each session swings it; later, each session nudges it. That settling is the signal — not a particular number of trades, and not a particular win rate.

Three sessions, three dates

The cheapest discipline is to test every idea on at least three different dates before forming an opinion, and to pick dates you have not studied. One session in a trending week will flatter a trend strategy; the same week will punish a range strategy. Neither result is wrong. It is just one draw from the deck.