forex-scalping.com

Profit Factor, and What Happens When You Remove the Best Trades

A trader's 50 trades: 18 winners totaling plus 40R and 32 losers totaling minus 24R. Compute the headline statistics.

Profit factor: gross profit divided by gross loss, 40 divided by 24, which is 1.67. Expectancy: net 16R over 50 trades, 0.32R per trade. Win rate: 36 percent. Average win: 40 divided by 18, 2.22R. Average loss: 24 divided by 32, 0.75R.

Interesting: the average loss is below 1R. That suggests many trades were closed early at a small loss, before the stop. It might be good management or inconsistent execution.

Now the robustness test. Remove the two best winners, say plus 5R and plus 4R. Gross profit falls to 31R. Profit factor becomes 31 divided by 24, which is 1.29. Expectancy: net 7R over 48 trades, 0.15R. The edge halved by removing two trades out of 50.

What does it say? Part of the result depends on a handful of outliers, which is typical for trend-following styles. But it also means the statistics are fragile, since a sample without those trades looks much weaker.

Compare with a strategy whose profits are spread evenly. Removing its best two trades would hardly change anything.

Neither profile is wrong, but each demands different patience and different confidence.

Try this: compute your profit factor with and without your best two trades.

The robustness test is the most useful part. If removing the top two winners drops profit factor below about 1.2, the results depend on rare events and the sample may not contain enough of them to trust.

Another angle on the average loss below 1R: it is worth splitting losers into "stopped at full 1R" and "closed early". If the early closes are mostly discretionary exits that saved money, fine. If many of them would have turned into winners, the management is costing edge, and the MFE after exit tells you which.

I ran this exact test on my own record last quarter and it was a sobering hour. Out of 140 trades my profit factor was 1.48, which I was quite pleased with. Taking out the best three winners, all of them from the same violent week around a Bank of England surprise, dropped it to 1.17. The rest of the quarter was barely paying for its costs.

That changed how I think about the quiet months. I used to tweak my rules whenever a few weeks went flat, but those flat weeks were actually what my strategy normally looks like between the outliers. Changing rules then just meant I was less likely to be positioned properly when the next big move arrived.

On the average loss point, I split mine the way the reply suggests and found that my early exits were saving money on about two thirds of occasions. The other third turned into full winners after I left. I kept the early exit but only when price closes back through my entry candle, not on a feeling.

Scrolling a forum or a social feed full of profit screenshots does something to expectations, and the damage is mostly from misreading what is being shown.

A single screenshot and a track record look similar because both display profit. They are different objects. The screenshot is one point picked after the fact from an unknown number of attempts. The track record is the full path, including the stretches nobody posts.

A large win and a repeatable process also look the same on the surface. A big green number can come from a position size nobody sane would sustain, or from one lucky news spike. The same number can also come from a method that works at a normal size. The image does not say which.

Another false twin is a trader who is posting and a trader who is trading. Time spent posting results is time not spent trading them, and the accounts with the most polished presentation are not reliably the ones with the best results.

The comparison hurts in a specific way. It resets what feels like a normal day. After enough screenshots, a plain day with a few pips net starts to feel like failure, and that pushes size up and quality down.

A useful habit is to ask what is missing from any image: the losing trades around it, the account size, the time frame, the costs. If those cannot be answered, the screenshot says very little.

How do you deal with the pull of other people's results on the forums, and has it ever changed how you traded for the worse?

The observation that the average loss is 0.75R is worth investigating further. If the planned stop is 1R and losses average 0.75R, many trades were closed before the stop. That could mean active management, cutting trades that weren't working, or it could mean inconsistent execution, closing trades in fear. Checking the MFE and MAE of the early exits would show which. If those trades often moved further against before recovering, the early exits helped. If they often reversed and would have reached the target, the early exits cost money. Either way, the pattern deserves a closer look.

The robustness test is a valuable habit. Removing the two best trades halves the expectancy from 0.32R to 0.15R. That's not unusual for trend following styles, which depend on occasional large winners. But it means the trader's confidence in the strategy should reflect the possibility that the next 50 trades might not include two such outliers. Over a longer sample, the outliers should appear regularly if they're a genuine feature of the strategy. Over short periods, their absence could make the strategy look much weaker than it is, which can lead traders to abandon it at the wrong time.

It's worth comparing the profit factor with and without outliers across several strategies. A strategy with a profit factor of 1.67 that falls to 1.29 without two trades has a moderate dependence on outliers. One that falls from 1.67 to 1.05 would be heavily dependent. One that barely changes would be robust to their absence. A simple rule is to prefer strategies that remain profitable with the best few trades removed, since that suggests the edge comes from the bulk of the trades. That provides more consistent results over shorter periods, which helps with confidence and adherence.

Computing the profit factor with and without the best two trades, as the exercise suggests, takes only a minute with a trade list in a spreadsheet. Sorting by R and removing the top two rows gives the adjusted figure. It's a good check to run before relying on a strategy's headline statistics. Has anyone here run this test on their own results? I'd be curious how much the profit factor fell and whether the result changed your confidence in the strategy, or how much risk you were willing to take with it.