Profit factor: the backtest number that hides your worst month

You are reading a strategy report. The equity curve climbs from left to right, and the line your eye lands on says the profit factor is above three.

Six weeks later the same rules are running on a live account and the balance is lower than when you started. Nothing broke. The report was arithmetically correct the entire time.

Profit factor is one of the most quoted numbers in trading and one of the least interrogated. By the end of this you will have three checks that separate a figure worth acting on from a figure that only decorates a sales page.

How is profit factor calculated?

Gross profit divided by gross loss. Total up every winning trade in currency terms, total up every losing trade, divide the first by the second. Above 1.0 the set made money. Below 1.0 it did not.

That is the entire formula, and the simplicity is where the trouble starts. The ratio flattens everything that happened between the first trade and the last: the order results arrived in, how long each position tied up your capital, how large each one was against the account balance at the time.

It is also currency-weighted, not trade-weighted. One position taken at four times your normal size counts four times as much toward the verdict. If your sizing has ever been emotional, the metric has quietly recorded that emotion as strategy performance.

What is a good profit factor?

No threshold survives contact with a different sample. The honest answer is that the same value can be strong or worthless, and what decides it is the sample underneath, not the digit.

Judge the sample first. Then read the number.

A modest figure that passes all six rows beats a spectacular one that fails two. I would rather trade the boring number I understand than the impressive one I cannot explain.

Which single trade is doing all the work?

Delete the largest winner and calculate it again. That is the fastest quality test available on any trade record, and it takes about a minute in a spreadsheet.

If the figure barely moves, the result came from the process. If it falls under 1.0, you have not measured a strategy at all. You have measured one lucky position that the other trades happened to be standing near.

Profit factor, win rate or expectancy: which question does each answer?

Each one is a different question about the same trades, and reading only one of them is how traders end up confidently wrong.

Expectancy became a retail staple largely through Van Tharp’s writing in the late 1990s, and it pairs well here. Profit factor tells you whether the account grew. Expectancy tells you what one more trade is worth. Neither says a word about whether you could sit through the middle.

Why does a backtest profit factor collapse live?

Three causes, and they usually arrive together.

The first is cost. Spread widens exactly when a fast strategy wants to enter, and swap grinds away at anything held overnight. A test that assumes a fixed spread has already inflated the profit column and shrunk the loss column.

The second is overfitting. Bailey, Borwein, Lopez de Prado and Zhu made the argument formally in the Notices of the American Mathematical Society in 2014: run enough variations over the same history and you will always find one that looks excellent in-sample while carrying no expected out-of-sample return at all. The impressive number is a property of how many attempts you made, not of the market. This is worth understanding properly before your next backtest .

The third is the trade log itself. If the signal you recorded only settled into position after the candle closed, the entry price in the log was never available to you. That is the mechanism behind backtests that repaint their way to profitability , and it corrupts the profit factor at the source rather than at the edges.

What does profit factor never tell you?

How bad it felt. The ratio is blind to sequence, so a run of eleven consecutive losses and a run of alternating results can produce an identical figure, while only one of them empties your patience and possibly your account.

It says nothing about depth or duration of drawdown either. That is what the ulcer index was built to capture, and why the two belong on the same page.

It also ignores what happened inside each trade. A position that closed for a small gain after travelling almost to your stop is scored the same as one that never went against you, which is precisely the gap maximum adverse excursion exists to fill.

Read alone, profit factor answers one narrow question well and every other question badly.

What has to be true before you trust the input?

The metric is arithmetic. Arithmetic cannot detect a dishonest input, and the input here is your list of entries and exits.

So the requirement is unglamorous: the signal that produced each logged entry has to be the signal that was on the chart at the moment you would have acted. If an arrow settles into a better location once the bar completes, every downstream figure inherits a price that never existed.

RelicusRoad Pro is built around signals that lock at candle close and stay where they were put, on MetaTrader and on TradingView, which is what makes a trade log admissible as evidence later. That is the claim, and it is a narrow one.

It will not raise your profit factor. Nothing on a chart decides your position size, your patience, or whether you take the trade the rules asked for. What a fixed reference point buys you is a record you can measure, and once your risk per trade is set from measured numbers rather than remembered ones, the metric finally describes something real.

Frequently asked questions

What is profit factor in trading? It is the total currency value of your winning trades divided by the total currency value of your losing trades, over any set of closed trades. A result of 1.0 means wins and losses cancelled exactly. Because it works on totals rather than averages, one oversized position weighs more heavily on the verdict than a properly sized one.

What is a good profit factor? No single threshold holds across different samples. The same value can be strong or meaningless depending on trade count, whether costs were deducted, whether settings were tuned on the data being scored, and whether one trade supplied most of the profit. A modest figure from hundreds of costed trades beats a spectacular one from forty trades in a trending quarter.

How many trades do you need before profit factor means anything? Enough that adding the next batch barely moves the figure, and enough calendar time that the trades were not all taken in one market condition. Recalculate after every twenty or thirty new trades. If the number swings by a third each time, you are reading noise rather than a strategy.

Is profit factor better than win rate? They answer different questions and neither is sufficient alone. Win rate tells you how often you were right and ignores the size of each outcome. Profit factor accounts for size but hides how often each result occurred and in what order. Read them alongside expectancy and your worst drawdown .

Why is my live profit factor lower than my backtest? Usually three things at once: costs understated in testing, settings tuned to a period that will not repeat, and logged entry prices that were never available because the signal only settled after the candle finished.

Run the outlier test on your own trade journal this week, and if you want entries that stay where they were printed, start with RelicusRoad Pro .