What is profit factor? Gross wins ÷ gross losses — and one big catch

Profit factor is the ratio of everything a strategy made to everything it lost — gross wins divided by gross losses. It is the fastest way to size up a trading record, and one of the easiest numbers to be fooled by, which is why our desk publishes a simulated 0.82 of its own. Research and education only — not financial advice.

The formula, in one line

Profit factor (PF) = gross profit ÷ gross loss. Add up the profit from every winning trade. Add up the loss from every losing trade, as a positive number. Divide the first by the second. No averages, no annualizing, no adjustments — just two sums and a division.

A purely hypothetical example: a system closes 10 trades. The 4 winners make $500, $300, $200 and $100 — gross profit $1,100. The 6 losers cost $200 each — gross loss $1,200. PF = 1,100 ÷ 1,200 = 0.92. For every dollar this hypothetical system gave up on its losers, the winners recovered 92 cents. It is a losing system — even though "a $500 best trade" would look great in a screenshot, and even though nothing about a 40% win rate sounds alarming on its own.

Reading the number, as a description of a past sample and nothing more:

PF (on a given sample)	What it says about that sample
Below 1.00	The sample lost money gross. Costs make it worse.
1.00	Break-even before fees and slippage — which means losing after them.
1.01–1.25	A gross edge thin enough that real-world costs often erase it.
Roughly 1.3–2.0	The range backtesters commonly treat as workable — if the sample is large, the profit isn't concentrated, and it survives costs.
Above ~2.5 on a small sample	Frequently a warning sign: overfitting or one outlier trade doing the lifting.

What 0.82 looks like in the wild — our own number

Definitions stick better with a real, published example, so here is ours. When we ran our raw scanner "traded blind" — take every signal it fired, no filters, no human judgment — the hypothetical backtest produced 161 simulated trades, a 46.6% win rate, a profit factor of 0.82, and an expectancy of roughly −2% per simulated trade. In plain terms: for every simulated dollar the losers cost, the winners clawed back 82 cents.

Two things are worth unpacking from that simulated set. First, PF below 1 and negative expectancy are the same fact wearing different clothes — if gross wins don't cover gross losses, the average trade must be a net loser. Second, you can reverse-engineer the shape of the trades: at a 46.6% simulated win rate, a PF of 0.82 implies the average simulated winner was only about 0.94× the size of the average loser. Winners that are both less frequent and slightly smaller than losers — that is the anatomy of a slow bleed, and no single trade on the list would have told you.

We publish that losing number on purpose. A raw scanner output is a candidate list, not a strategy, and 0.82 is the honest simulated baseline that everything else — catalyst checks, liquidity screens, adversarial review, exits — has to justify itself against. The full workup, losers included, lives at the record.

0.82 vs 1.41: what the gap actually costs

Now take 1.41 as an illustrative benchmark — a hypothetical system that books $1.41 in gross wins for every $1 of gross losses. Closing the gap from 0.82 is not a polish job. There are exactly two levers:

Make winners bigger relative to losers. Holding that 46.6% simulated win rate fixed, the average winner would have to grow from ~0.94× to ~1.62× the average loser to reach a hypothetical 1.41. That is exit engineering — where profits are taken and where losses are cut.
Win more often. Holding winners roughly equal to losers in size, a hypothetical PF of 1.41 needs a win rate near 58–59%. That is selection — which signals get traded at all.

Every stop, take-profit level, time-stop and filter is an attempt to move one lever without wrecking the other, and the honest way to test whether a rule change does that is a properly run backtest — with all the traps that come with one.

Why profit factor alone misleads: the 61% problem

Profit factor is a ratio of two sums, and sums have no memory of how they were assembled. One monster trade can carry the entire numerator while the metric smiles back at you.

Our desk hit this exact wall. While grid-testing 21 exit-rule variants over the same simulated signal set, the best-looking cell showed +362 simulated units — comfortably the prettiest result on the board. Our own statistical audit rejected it: a single ticker accounted for 61% of that cell's simulated profit, and the result failed significance testing. Strip out one lucky name and the "edge" mostly evaporates. The grid and the rejection were published together, because a best cell you haven't tried to kill is marketing, not research.

Concentration is only the first blind spot. On its own, profit factor also tells you nothing about:

Sample size. A PF of 3.0 across 12 trades is noise with good posture. 161 simulated trades was enough for us to trust a negative verdict; small samples rarely justify a positive one.
Sequencing and drawdown. PF ignores order. Two records with identical PF can be a smooth climb or a stomach-dropping 60% drawdown followed by a rescue.
Costs. Most quoted PFs are computed before slippage, spreads and fees — precisely the things that turn a thin gross edge negative.
The future. An in-sample PF is a description of what happened, not a forecast of what will. No PF, ours or anyone's, implies future results.

How to actually use profit factor

Compute it yourself from trade-level data. If a room or vendor quotes a PF but won't show the trade list behind it, the number can't be audited and shouldn't be trusted.
Run the concentration check. What share of gross profit comes from the single best ticker and the single best trade? If one name carries most of the profit — like the 61% case above — you are looking at luck wearing a lab coat.
Check the sample size and slice by time. A PF that holds across months and market conditions means more than one built in a single hot streak.
Recompute net of realistic costs. Especially for options and fast-turnover systems, the spread alone can move a PF below 1.
Then paper trade it forward. An out-of-sample run on a model desk — no real money — is the cheapest lie detector available; see what is paper trading.

Standard caveat, stated plainly. Every performance figure on this page from our desk — 161 trades, 46.6% win rate, 0.82 profit factor, −2% expectancy, the +362-unit rejected cell — is hypothetical/simulated output from a paper/model desk trading no real money. Backtested and simulated results have inherent limitations and do not indicate or promise future outcomes.

Profit factor grades trades after the fact; sizing and exit discipline are what shape them beforehand. Our beginner handbook Options, In Plain English builds that discipline around one real trade, mistakes included — a free chapter is at Options, In Plain English (EN/ES/PT/FR).

The 30-second recap

Profit factor = gross wins ÷ gross losses. Above 1, the sample made money gross; below 1, it lost.
Our published example: a raw scanner traded blind scored 0.82 across 161 simulated trades — 82 simulated cents recovered per dollar lost.
Reaching a hypothetical 1.41 from there requires winners ~1.6× losers at the same win rate, or a win rate near 59% — exits or selection, nothing else.
PF is a ratio of sums: one ticker carried 61% of our best-looking simulated grid cell, and the desk's own audit threw it out.
Always pair PF with sample size, concentration, costs and an out-of-sample paper run — and treat any PF as history, never a promise.

Common questions

How do you calculate profit factor?

Sum the profits of all winning trades (gross profit), sum the losses of all losing trades as a positive number (gross loss), and divide gross profit by gross loss. Hypothetical example: $1,100 in total wins against $1,200 in total losses gives a profit factor of 0.92 — a losing sample.

What is a good profit factor?

There is no universal threshold. Above 1.0 means the sample was profitable before costs; backtesters commonly treat the mid-1s as workable if the sample is large, the profit is not concentrated in one or two trades, and it survives realistic fees and slippage. A very high PF on a small sample is more often an overfitting warning than a strength. Any PF describes past data only.

Can a system with a profit factor above 1 still lose money?

Yes. Quoted PFs are usually gross — slippage, spreads and fees can push a thin edge below break-even. Concentration is the other trap: if one outlier trade or ticker supplies most of the gross profit, the rest of the system may be losing. Our own audit rejected a +362-unit simulated result for exactly that reason — one ticker was 61% of the profit.

Is profit factor better than win rate?

They measure different things and only work together. Win rate is frequency; profit factor is magnitude. Our published blind-scanner backtest shows why you need both: a near-coin-flip 46.6% simulated win rate paired with a 0.82 simulated profit factor — winners slightly smaller and less frequent than losers, which no win rate alone would reveal.

Why does ClaudeQuantAlgo publish a losing profit factor?

Because it is the honest baseline. The 0.82 simulated PF is what our raw scanner produces traded blind, with no filters or review — proof that a signal list is not a strategy. Publishing it, alongside the rejected best-looking grid cell, lets anyone audit the gap between raw output and the disciplined process. Details are on the public record at /record/.

See the process with your own eyes. The desk posts trigger-based cards to a public, timestamped record — losses included — and published the backtest where its own raw scanner loses. The scoreboard is free to watch. Join the floor →

Free to join · paid floors optional · research and education only

Last updated 2026-07-11 · ClaudeQuantAlgo Research Desk · research and education only.

Disclosures. ClaudeQuantAlgo is a research and education community. Nothing on this page constitutes financial, investment, legal, or tax advice, or a recommendation to buy or sell any security or derivative. No profit promises are made, ever — trading stocks, options, and forex involves substantial risk of loss; options positions can lose 100% of their value. The public scoreboard reflects a model ("paper") desk — no real money. Past performance — real, paper, or simulated — never guarantees future results. Hypothetical and simulated results have inherent limitations and no representation is made that any account will or is likely to achieve similar profits or losses. ClaudeQuantAlgo is not a registered investment adviser or broker-dealer. You are solely responsible for your own trading decisions. Never risk money you cannot afford to lose.