Win Rate After a Loss: My 24% Problem

About the numbers in this post. Every figure below is one trader's own historical record over a two-week window. It is not a typical result, it is not a projection, and no representation is made that any account will or is likely to achieve similar results. Futures trading contains substantial risk and is not for every investor. Past performance is not necessarily indicative of future results.

Sort your trades by one thing: what the trade before it did.

Not by setup. Not by session. Not by instrument. Just this: did the previous trade win or lose?

I did this with two weeks of my own live trading, April 2–15, on my prop firm account at Topstep. Here's what came back:

• Win rate after a winning trade: 55%
• Win rate after a losing trade: 24%

Same trader. Same model. Same markets. Often the same *hour*. The only variable is what the last trade did to me.

If you trade a prop firm account and you've never run this split on your own data, stop reading and go run it. This post will still be here. I'd bet money your version of these two numbers is the most important thing in your journal, and you've never looked at it.

What the data actually says

Some context, because one stat alone can lie. Over those two weeks on the live account I ran a 29.6% win rate, 5.4 trades a day, and a profit factor of 1.08. On my sim account, with the same model, same setups, same sessions, I ran 61.8%, 2.2 trades a day, and a 1.95 profit factor.

The model wasn't the problem. The model was fine. It kept working on sim the whole time. What broke on live was everything *around* the model, and the after-a-loss split is where the break shows up first.

It gets worse the deeper into a session you go. Trades #9–11 in a day: 0% win rate. Not low. Zero. By the ninth trade there is no model left. There's just a person trying to get back to a number.

And here's what it cost. Two sessions in two weeks landed at −$1,714 and −$2,134, against a $2,000 daily loss limit. One of those went through the limit. My worst run was 11 consecutive losses, and if you know my after-a-loss number, you know an 11-loss streak isn't bad luck. At 24%, streaks like that are the *expected* output. The math of tilt is self-sustaining: every loss lowers the odds on the next trade, which produces another loss, which lowers them again.

One more cut, because it tells the same story from a different angle: my shorts ran 13.3% on live versus 59.3% on sim. That's not a broken short setup. The setup was winning on sim all fortnight. That's me fighting a trend I could see, because being flat while down felt like losing.

The next trade after a loss is a different trade

Here's the thing the split forced me to accept: the trade you take after a loss is not the same trade you'd have taken cold. It looks the same on the chart. It is not the same in your head.

After a win, you're patient. The last trade proved the model works, so you're happy to wait for the next real one. The bar stays where you set it.

After a loss, the red number sits on your screen and it doesn't feel like variance. It feels like a verdict. And there's exactly one way to make it go away fast: another trade. So the bar drops. The next setup is *close enough*. You skip the confirmation you'd normally demand, because waiting means sitting with the loss longer. You entered that trade to escape a feeling, not because the model called for it.

That's the whole mechanism behind revenge trading and overtrading after losses. Nobody sizes up on their weakest idea on purpose. It happens one small rationalization at a time, each one feeling reasonable in the moment, and the data catches all of it: 24% after a loss, 5.4 trades a day from a model that has never produced 5.4 setups a day, 0% on trades #9–11.

Roughly three of my five daily trades weren't the model. They were me, in a chair, needing something to happen.

This is why "just be disciplined" is useless advice. In the moment, you're not the person who wrote the rules anymore. You're the person the rules were written *about*. Willpower is exactly the resource tilt drains first. If your plan for the after-a-loss trade is to make a good decision at that moment, you don't have a plan.

The circuit-breakers that address it

So I stopped trying to out-discipline tilt and started removing the decisions instead. These are the rules I trade with now. Every one of them exists because of a specific number in the data above, and every one is decided *before* the open, when nothing is on the line.

Max 5 trades a day. My model produces two or three real setups a session. Trade six was never going to be one of them. A hard cap is blunt, and that's the point: it doesn't ask me to assess my own state while tilted. It just runs out. The 0% on trades #9–11 means this rule has literally no cost: I'm capping trades that never won.

Three consecutive losses ends the session. At a 24% after-a-loss win rate, three losses in a row isn't a signal to push through. It's the machine that produced my 11-loss streak getting warmed up. Session over, platform closed. The market will be there tomorrow.

Two losses gives back a green day: stop. If I'm profitable and then lose two, I lock the green session and walk. A green day turned red is the tilt trade with extra fuel: now I'm not just recovering a loss, I'm recovering a number I already *had*.

$2,000 daily loss is a hard stop. That's the account's daily loss limit, and I never want the firm's number to be the thing that stops me, but it's written down as the absolute back wall, non-negotiable. The −$2,134 session happened because it wasn't.

Max $200 risk per trade, about 2% of the account. This is the arithmetic layer under the behavioral layer. Sizing is by stop distance, fixed dollar risk: contracts = $200 ÷ (stop points × $2) on MNQ. Wider stop, fewer contracts, never the same contracts and more risk. At $200 a trade, even a full losing day of max trades can't put the daily limit in play on its own. The sizing math has to hold even on the day the discipline doesn't.

Notice what these rules have in common: none of them require judgment at the moment judgment is worst. They're all pre-committed, all mechanical, and all of them fire *before* the point where my data says my decision-making has already left the building.

You can't fix a number you've never seen

Here's the honest part. I had been trading this account for weeks before I ran that split. I knew the losing days felt different from the winning days. I could not have told you it was 24 versus 55. The feeling was vague; the number is not. The number is what made the rules non-negotiable, because I can't argue with my own fills.

This is what a journal is actually for: not screenshotting winners, but surfacing the split you don't want to see. It's also why I built this exact view into Propfy: the behavior section of the analytics shows your win rate after a win versus after a loss, your win rate by trade number in the day (1st, 2nd, 3rd, 4th–5th, 6th+), and your loss streaks, straight from your fills, synced from TopstepX or imported by CSV. It shows you where you stand. What you do with it (the rules, the stopping, the closing of the platform) is still yours.

The uncomfortable conclusion

My model was never the bottleneck. It ran a 61.8% win rate on sim while I ran 29.6% live with the same setups. The bottleneck was the ten seconds after a losing trade, repeated five times a day, for two weeks.

If your prop firm account is bleeding and your sim looks fine, I'd start there. Run the split. Get your own version of the 24. Then build circuit-breakers that fire before the number does, because a correct trade taken at a loss beats a win taken on tilt, and your after-a-loss win rate is the purest measure of which one you're actually taking.

Frequently asked questions

Why does win rate drop after a losing trade?

Because the loss changes how the next entry gets selected. The urge to get back to flat compresses patience, lowers the bar on setup quality, and pushes size up exactly when judgment is worst. The setup criteria haven't changed. The person applying them has. That's why the same trader can post materially different win rates after wins versus after losses in the same fortnight.

How do I know if I'm revenge trading?

Compare your trade count to your model's real setup frequency. If your model produces two or three setups a session and you're averaging five or more trades a day, the extras came from somewhere, and it usually isn't the chart. Then split your win rate by previous-trade outcome and by trade number in the day. If the late trades and after-a-loss trades are dramatically worse, that gap is the revenge trading, measured.

How do I stop overtrading after losses?

Pre-committed circuit-breakers, decided before the session: a hard cap on trades per day, a consecutive-loss cutoff that ends the session, a rule that locks in green days, and fixed dollar risk per trade. The specific numbers matter less than the fact that they're written down before the open and never renegotiated mid-session. Rules you can renegotiate while tilted aren't rules.

What's a good win rate after a loss?

There's no universal number. It depends on your model. The number that matters is the *gap* between your win rate after a win and after a loss. A small gap means your process holds under pressure. A large one (mine was 55% versus 24%) means the trade after a loss is a different, worse trade, and no amount of setup refinement fixes that. The fix is behavioral, not technical.

---

*Educational content for prop-firm futures traders. My numbers are from my own live prop firm account, April 2–15, 2026. Your data will differ, so go look at it.*