What a realistic profit factor looks like (and why 3.0 means you have a bug)

July 31, 2026 · 8 min read

Profit factor is the first number most people quote about a strategy and the one they understand least. It is easy to compute, easy to inflate, and almost meaningless without the trade count next to it. Here is what it measures, what range is actually plausible, and what a very high value is usually telling you.

What it measures

Profit factor is gross profit divided by gross loss — the sum of every winning trade over the absolute sum of every losing trade. A profit factor of 1.0 means the wins exactly paid for the losses; below 1.0 you lost money; 1.30 means you took in $1.30 for every $1.00 you gave back.

It is a ratio of totals, which is the source of most of the confusion. It says nothing about how many trades produced those totals, nothing about the path, and nothing about whether one outlier carried the whole thing. Two strategies with identical profit factors can have completely different risk of ruin.

How it relates to win rate

Profit factor decomposes cleanly. With w as the win rate and R as the ratio of the average win to the average loss:

profit factor = R × w / (1 − w)

This is why "high win rate" and "good strategy" aren't the same claim. A 70% win rate with an average win half the size of the average loss gives a profit factor of 1.17. A 35% win rate with wins three times the size of losses gives 1.62. The table below is that formula evaluated — it shows the payoff ratio you need to reach a profit factor of 1.30 at each win rate:

Win ratePayoff ratio needed for PF 1.30Interpretation
30%3.03Trend / breakout profile — rare wins, large
40%1.95Typical of a breakout with a wide target
50%1.30Wins must be meaningfully bigger than losses
60%0.87Can survive wins smaller than losses
70%0.56Mean-reversion profile — frequent small wins
Direct evaluation of the formula above (payoff ratio = PF × (1 − w) / w), not a backtest. It shows what has to be true arithmetically, not what any particular market delivers.

What range is plausible

There is no universal "good" profit factor, because the number depends heavily on how often you trade and how much of the edge costs eat. But the shape of the relationship is consistent and worth internalising:

  • The more frequently you trade, the lower a plausible profit factor gets. A strategy taking several trades a day is competing against costs on every one; sustained profit factors well above 1.5 in that regime are unusual, and the ones that exist tend not to be public.
  • Longer holds tolerate higher ratios. A swing strategy holding for days pays costs once per multi-point move, so a higher profit factor is arithmetically easier to reach — and correspondingly less impressive.
  • Out-of-sample is always worse than in-sample. Not sometimes. If your held-out period shows the same profit factor as your fitting period, be suspicious of how the split was done rather than pleased.

The useful mental model: for an intraday index-futures strategy, net of realistic costs, a profit factor modestly above 1.0 that holds up out of sample and across many trades is a real result. A profit factor of 3.0 on a few hundred intraday trades is not a better strategy — it is a different kind of object, and almost always a bug.

Why 3.0 usually means a bug

When an intraday backtest comes back at 3.0, the cause is nearly always on this list, roughly in order of frequency:

  • No costs, or costs applied once instead of per round turn. The single most common cause. Check whether removing costs entirely changes your number — if it doesn't, they were never applied.
  • Fills at the signal bar's close. You cannot transact at a price you only know once the bar has finished. On breakouts this alone can double the apparent edge.
  • Stops filled at exactly the stop price. Real stops slip, and they slip when the market is moving fastest against you. A backtest that never slips a stop is systematically removing the worst part of every losing trade — which lands directly in the denominator of the profit factor.
  • Both sides of an intrabar range being hit. If a bar's high touched your target and its low touched your stop, bar data cannot tell you which came first. Engines that default to "target first" turn losers into winners at exactly the wrong moments.
  • Parameter selection over a small sample. The best of forty variants on 200 trades will produce an impressive profit factor by construction. It is a maximum over draws, not an estimate.
  • One or two outlier trades. Delete the single best trade and recompute. If the profit factor falls from 3.0 to 1.1, you have one trade, not a strategy.
The one-line diagnostic: a profit factor that survives realistic costs, a one-bar entry delay, deletion of the best trade, and a held-out period is worth reporting at any value. One that doesn't survive those is worth nothing at 3.0.

Report it with company

Quoted alone, a profit factor is unfalsifiable. Report it with the four numbers that make it interpretable: trade count, win rate, average win and average loss, and the cost assumption per round turn. With those, a reader can reconstruct your result and check whether it hangs together. Without them, "profit factor 2.4" is a claim about nothing in particular.

Trade count matters most of all — a profit factor over 40 trades carries essentially no information, which is the subject of how many trades a backtest needs. For the biases that inflate the ratio in the first place, see why your backtest lies.