Profit factor is the ratio of your total gains to your total losses. A profit factor of 1.4 means that for every dollar lost, you made $1.40. It is the measure that best sums up a system in one number, because it weighs trades instead of counting them.
What the ratio compares
Contents
Profit factor (PF) answers a question neither win rate nor P&L asks: for one dollar of loss, how many dollars of gain does this system produce? It combines the frequency of wins and their size into one number, which makes it the first figure to look at when judging a strategy.
It is independent of account size and trade count, which makes it comparable across periods and across traders. That is also why practically every trading journal and backtesting package displays it.
Its real value shows up when it is read per setup rather than globally: that is where you find one configuration carrying the whole result while another eats it.
Profit factor = sum of winning trades ÷ |sum of losing trades|
The denominator is taken in absolute value: losses are negative, and the ratio must stay positive.
The calculation runs on the net, fees included. A profit factor computed on gross is systematically more flattering than reality, and the more so the higher your frequency.
If no loss was recorded over the period, profit factor is mathematically infinite. That is not a result, it is a sample that is too short.
| Profit factor | What it means |
|---|---|
| below 1 | The system loses money. No amount of discipline will flip that sign. |
| 1 to 1.2 | Break-even or close. Fees alone are enough to push the period into the red. |
| 1.2 to 1.8 | A credible and liveable system. This is the order of magnitude of most methods that hold up. |
| 1.8 to 3 | Very good, provided the sample exceeds a hundred trades. |
| above 3 | Almost always an artefact: one exceptional trade, too short a period, or an over-optimised backtest. |
The first useful reflex is not to look at the value, it is to recompute it without your biggest winner. If it collapses, it was not a system, it was one lucky hit, and no conclusion holds. That check takes two minutes and it prevents building a strategy on an accident.
The second is to break it down by setup, by market and by time window. A global profit factor of 1.3 can hide one configuration at 2.1 and another at 0.6: dropping the second is often enough to transform the account, without learning anything new.
Profit factor = 1.39, which is good. But removing the single biggest winner drops it to 1.04: this quarter rests on one trade, and that two-line check is worth more than the headline figure.
This is the original use of profit factor, and the one backtesting software makes of it: a rule set is tested over history, and profit factor sums up in one number whether the whole thing wins. It lets you compare two versions of the same strategy without getting lost in curves.
Watch for the classic backtesting trap: a profit factor that rises with every extra parameter no longer describes a method, it describes the past. That is over-optimisation, and it is detected by testing on a period the parameters never saw.
This is the most profitable use on a real journal, and it is rarely practised. Compute profit factor per setup, per market, per time window, per day of the week. A mediocre global result almost always hides one or two clearly profitable configurations and one that eats them.
This reading requires no new skill and no learning: dropping the configuration at 0.6 is often enough to move an account from red to green, with nothing else changed.
Most tools display one profit factor, computed over all your trades, and that is exactly what makes it unusable. A global figure mixes setups that have nothing in common, markets with different fees, and periods where your position size was not the same. It produces an average that describes none of those situations.
Second common flaw: computing on the gross. Removing commissions and spread changes the result by several tenths on a high-frequency style, always in the flattering direction. A profit factor of 1.25 announced on gross can be 0.95 on net, that is a losing strategy displayed as profitable.
Third flaw, the quietest one: no robustness check. No tool spontaneously tells you that your profit factor rests on a single trade. That is nonetheless the most frequent case on samples under a hundred trades, and recomputing without the biggest winner is all it takes to see it.
The classic mistake. Celebrating a profit factor of 4. On a real sample, a figure like that almost always comes from one huge trade or a run that is too short. On a backtest, it more likely signals a parameter set fitted to the past, which will not repeat live.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| Profit factor | Σ gains ÷ |Σ losses| | What you earn per $1 lost | Judging a whole system, or one setup |
| Win rate | winners ÷ total | How often you win | Understanding the psychological comfort |
| RRR | average win ÷ average loss | How winners weigh against losers | Setting targets and stops |
| Expectancy | net P&L ÷ trade count | What an average trade earns | Estimating what more volume will earn |
Compute your profit factor per setup. If one sits below 1 on a sufficient sample, stopping it improves the whole without you having learned anything new. It is the fastest lever and the only one that requires no practice.
The denominator is what you control. If your average loss exceeds your planned risk, a handful of trades with no recorded stop explain almost the entire gap. Fixing those raises profit factor without touching a single entry.
Below a hundred trades, profit factor moves by several tenths on a single result. Many traders change method on a figure that did not yet mean anything, and lose the one thing producing a result: repetition.
It is the ratio of a strategy's total gains to its total losses. The profit factor definition fits in one sentence: how many dollars you earn per dollar lost. Above 1, the system is profitable over the measured period.
Between 1.2 and 1.8 for discretionary trading on a real sample. Below 1.2 fees alone can tip the period over, and above 3 you should be suspicious: on real history, a figure like that almost always comes from one trade or an over-optimised backtest.
Add up all your winning trades, add up all your losing trades in absolute value, then divide the first total by the second. The calculation runs on the net, fees included, otherwise the result is systematically flattered.
Profit factor is a unitless ratio: it says what you earn per dollar lost. Expectancy is an average per trade, in currency or in R: it says what one repetition earns, so it multiplies by the number of trades. Both describe the same reality from two angles.
No, and that is its most dangerous blind spot. It compares totals, not trajectories: two systems at 1.6 can have one with 6% maximum drawdown and the other with 38%. The second is unliveable and profit factor says nothing about it. It is always read alongside maximum drawdown and recovery factor.
In Tradoshi, profit factor is computed on the net, commissions and swaps included, and it appears in the Reports screen next to win rate, expectancy and maximum drawdown. It recomputes automatically on every sync with your broker, with nothing for you to type in.
Sorting by setup, by market and by time window happens in the same screen: that is the reading described above, the one that points at the configuration to stop. Trades imported from a statement and those pulled by automatic connection enter the same calculation.
To go deeper on this: read the full guide. Every term is gathered in the glossary.