Profit Factor in Trading: The Most Abusable Number in Any Backtest

Every backtest report you have ever opened leads with profit factor. Every EA vendor quotes it. Every strategy marketplace sorts on it. It is the single most visible number in retail trading software.

It is also the easiest number in trading to make look good without having an edge, which is precisely why it appears so prominently in the material that is trying to sell you something.

Used properly it is one of the most useful numbers you own. This article covers both halves: what it genuinely tells you, and the four ways it lies.

The formula

A profit factor of 1.0 means you made exactly as much as you lost. Below 1.0 you are losing money. Above 1.0 you are making it. There is no ambiguity in the sign, which is the metric’s great virtue.

It is also scale-free. Account size, position size, instrument and currency all cancel out. That makes it directly comparable across strategies in a way that a P&L figure never is, for the same reason R-multiples beat dollars.

Its relationship to win rate and expectancy

Profit factor, win rate and average win are three views of one underlying object. Fix any two and the third is determined:

Which means a target profit factor imposes a hard trade-off. Here is what a profit factor of 1.5 demands of you at different win rates, expressed in R, where 1R is your risk on the trade:

Average loss held at 1.0R throughout.

Look at the third column. Every row has the same profit factor. None of them has the same expectancy. The 30% trader earns more than twice as much per trade as the 70% trader while posting an identical profit factor.

That is the first thing profit factor cannot tell you: how much you make per unit of effort. Two strategies with PF 1.5 can differ by a factor of two in per-trade edge, and if one of them takes 400 trades a year while the other takes 40, the annual outcome is not remotely comparable. Profit factor is silent on frequency, and frequency is half of your income.

This is why expectancy remains the primary number and profit factor is the sanity check beside it.

The four ways it lies

1. One trade did all the work

This is the big one, and it is the reason to distrust any profit factor quoted without a supporting distribution.

Take a 50-trade sample: 20 winners averaging +2.4R, 30 losers averaging −1.0R.

One trade out of fifty was carrying nearly a fifth of the profit factor and half the expectancy. That strategy is not a 1.60 strategy. It is a 1.30 strategy that got one very good fill.

2. The sample is too short to carry the number

Profit factor inherits every sampling problem that afflicts the Sharpe ratio, and for the same reason. Thirty trades tell you almost nothing. A hundred tell you a little. Several hundred, taken across different market conditions, start to mean something.

The specific danger is that profit factor is bounded below at zero but unbounded above, so short samples skew high. A run of luck can produce a PF of 3.0 on forty trades from a strategy whose true long-run figure is 1.1. The reverse rarely happens with the same drama, which is why the eye-catching numbers on strategy marketplaces are almost always short-sample numbers.

3. It cannot see ruin coming

This is the failure that costs accounts rather than opinions.

A martingale or grid system, which doubles into losers and closes baskets at small profits, produces an exceptional profit factor. It closes almost every basket green. Gross loss stays tiny for months. Profit factors of 4, 6, 10 are routine in the marketing material, and they are not fabricated.

They are also not predictive, because the loss that ends the strategy has not been taken yet. Profit factor is a backward-looking ratio of realised outcomes. It contains no information about the size of the loss you are exposed to but have not yet suffered. That question belongs to risk of ruin, and no profit factor, however high, is a substitute for it.

The tell is always the same: an exceptional profit factor combined with an unusually high win rate and an average loss larger than the average win. When you see that combination, you are looking at a strategy that is selling small, frequent gains against a large, rare loss.

4. It gets computed on the wrong numbers

Three quiet corruptions, in order of how often they occur:

What the numbers conventionally mean

Assuming a net-of-costs figure on at least a few hundred trades:

The row that surprises people is the middle one. A profit factor between 1.2 and 1.5 sounds unimpressive next to the 3.4 on somebody’s screenshot. It is, in fact, roughly where a genuine, durable, cost-adjusted edge tends to sit. The 3.4 is usually a short sample, an uncounted cost, or a tail that has not fired.

How to actually use it

Three jobs, and only three.

As a pass/fail gate. Below 1.0, net of costs, over a meaningful sample, stop. Everything else is a conversation about how good, and that conversation is pointless if the sign is wrong.

As a robustness test. Compute it on all trades, then again without your best trade, then again without your best three. A durable edge degrades gently. A fake one collapses. This single exercise will save you more money than any indicator you ever learn.

As a cross-check on expectancy. If expectancy is positive but profit factor is barely above 1.0, your edge is thin and highly dependent on the tail of your winners. If profit factor is comfortable but expectancy per trade is small, you need frequency, which means costs matter more than you think. The two numbers together tell you which lever to pull, which is the sort of thing you find in a proper trade review and nowhere else.

Calculate it

Profit factor is trade-level, so it needs your ticket history rather than an equity curve. Two tools cover the two halves:

Frequently asked questions about profit factor

What is a good profit factor?

Between 1.2 and 1.5 net of costs, over several hundred trades, is a real edge and roughly where most durable professional records sit. Above 2.0 is possible but should trigger verification rather than celebration. Below 1.0 means the strategy loses money regardless of how the equity curve looked over any particular stretch.

Is profit factor the same as expectancy?

They share a sign but not a scale. Profit factor above 1.0 and expectancy above zero always occur together, so either will tell you whether a strategy makes money. Only expectancy tells you how much it makes per trade, which is what you need to project income or size positions. Profit factor is scale-free; expectancy is not, and that difference is the whole reason to keep both.

How many trades do I need before profit factor is reliable?

More than most traders assume. Under a hundred, treat it as directional only. Several hundred trades spanning different market conditions is where the figure starts to carry weight. And regardless of sample size, run the one-trade test, because a large sample does not protect you from a single outlier doing all the work.

Why does my live profit factor differ from my backtest?

Usually costs, slippage and fill assumptions, in that order. A backtest that fills at the exact high or low of the bar, or ignores spread widening around news, will produce a profit factor that cannot be reproduced. The gap is largest on short-timeframe strategies, where costs consume a bigger share of a smaller edge.

Should I include open trades in the calculation?

No. Profit factor is a measure of realised outcomes. Including unrealised positions lets a losing trade sit outside the gross loss figure for as long as you refuse to close it, which converts the metric into a record of your reluctance rather than your results. Closed trades only.