Total gross profit divided by total gross loss over a given period. A profit factor above 1.5 indicates a profitable strategy; above 2.0 is strong.
also: PF, Gross profit ratio
Updated
For every unit of money the strategy lost, how much did it make? Above one it took in more than it gave back; below one it did not. It is a ratio of totals, so it says nothing about the order the wins and losses arrived in.
Profit factor is the fastest way to tell whether a strategy has an edge at all: it collapses every trade into one number with a hard threshold at 1.0. It is also the number most easily inflated, which makes it a screening tool rather than a verdict.
Profit factor = Gross profit ÷ |Gross loss| Both sums cover closed trades only, over the same window. Costs belong inside the figures, not beside them: a profit factor computed before spread and commission is not the one the account will experience.
No single number describes a strategy. These metrics change how this one should be read.
An EA closes its winning trades for $8,000 in total and its losing trades for $4,000 in total over the same window. The ratio of those two sums is the profit factor.
Calculation 8,000 ÷ 4,000 = 2.00
Result Profit factor 2.00 — the strategy earned two units for every one it lost
Illustrative figures — not the record of any listed EA.
A profit factor is only as good as the sample behind it. Read the trade count, the cost model and the window before the number.
| Range | What it means |
|---|---|
| Fewer than about 100 trades | Statistically thin. A handful of large wins can carry the ratio, and removing the single best trade often drops it below 1.3. |
| Computed without spread or commission | Not the figure you will trade. Costs land entirely in the denominator, so realistic costs always lower it. |
| In-sample (optimised window) | Reflects the tuning as much as the strategy. Useful only as the top of a range whose bottom is the out-of-sample figure. |
| Out-of-sample or forward | The figure worth quoting. A moderate drop from in-sample is normal; a collapse through 1.0 means the edge was fitted. |
| Recomputed from a published trade ledger | The only version you can audit. If the seller ships the closed-trade list, you can re-add gross profit and gross loss yourself — and compute the ratio over a window the seller never quoted. |
The fourteen runs here span 1.27 to 1.52: nothing reaches the 1.8–2.5 band above, and three sit below the 1.3 line this page calls marginal. All fourteen ran 0.1 lots on a 10,000 USD deposit — the fixed-lot input the builds ship with, LotSize on ten, InpLots on two, numbered LotSize inputs on the two bundles — and that input is the one the ratio ignores: raise it and profit factor does not move. Each EA page prints the ratio beside the window, trade count and tester model, and ships the closed-trade ledger. The post-listing block on those pages reports return and drawdown but not profit factor — recomputing that one from the ledger puts two of the thirteen post-listing windows below 1.0.
The ratio has no drawdown term. A strategy that averages down into every loser posts a high profit factor for years — unrealised losses never close, so they never enter the denominator — until the one basket that does close erases the record.
A profit factor from a zero-spread backtest and one from a real-spread backtest measure different things. Match the cost model first, then compare.
Profit factor is a statistic, and small samples are noisy. Twenty trades can produce almost any figure; the number only starts carrying information in the hundreds.
It means the sample was profitable, not that the account survives the path. A 1.2 profit factor alongside a 40% drawdown describes a strategy most accounts cannot sit through.
Sorting by the ratio picks the strategy whose backtest window suited it best, not the one most likely to repeat. The second-best backtest among the fourteen EAs here — 1.526 across 232 trades — turned in 0.72 across the 21 trades that closed after it was listed, and its published figure fell to 1.43 once those trades counted. On 21 trades that reversal is noise, not a law; noise is exactly what a ranking column cannot tell you apart from an edge.
Listings whose published record is judged on Profit factor.
