Fifty trades. A fair coin-flip edge, 50% win rate, even payouts. Run that sequence and there's a 97% chance you hit at least five losses in a row somewhere in there.
Not a bad month. Not a broken strategy. Just math.
I used to think a losing streak meant something was wrong with the system. It took two years of live trading for me to accept that the streak isn't the exception. It's the baseline.
Here's the number that changed how I size positions. Researchers who ran the combinatorics on a 50-trade sample at a 50% win rate found a 71% chance of six consecutive losses somewhere in the run, and a 29% chance of seven. At a 40% win rate, which is closer to what most systematic strategies actually run since they lean on a bigger average win instead of a high hit rate, those streaks get longer and show up more often.
Most retail bots bet the same size on trade fifty as they do on trade one, win streak or lose streak. That's fine in theory if every trade is truly independent and the edge is stable. In practice, a losing streak tells you one of two things: you're in the statistically normal bad stretch every strategy has, or the edge has actually degraded. You can't tell which one you're in until the streak is already over.
The asymmetry is what makes this expensive. Lose 20% of an account and you need a 25% gain just to get back to even. Lose 50% and you need 100%. The deeper the hole, the more the math works against you, and a system that keeps betting full size through a losing streak digs that hole faster than it needs to.
A streak-defense rule flips this around. Cut position size by a fixed amount, say 25%, after three consecutive losses, and restore it once you get a win or size returns to baseline. You're not trying to predict when the streak ends. You're just making sure that if trade six, seven, and eight in that unlucky stretch are also losers, they cost less than trades one, two, and three did.
Most people get this backwards, and I did too for a while. The instinct after a loss is to trade bigger to make it back faster. That's a martingale, and it's how a bad week turns into a blown account. The instinct after a win is often to trade smaller out of fear of giving it back, which just caps the upside for no statistical reason.
The streak-defense rule only cares about one thing: recent losses, counted mechanically, not emotionally. It doesn't ask whether you feel confident. It doesn't care that you're sure the next trade is the one that turns it around, because that feeling shows up most right before the seventh loss in the streak, not before the first.
I also see people confuse this with a stop-loss or a daily kill switch. Those answer "when do I stop trading." This answers a different question: how much should I risk while I'm still trading through a stretch I can't yet tell is normal or broken. You need both, and they're not the same lever.
If you're running anything systematic, go find your max historical losing streak at your actual win rate, not the streak you remember seeing, the streak the math says is coming. Then build one rule: after N consecutive losses, cut size by a fixed fraction until you get a win.
It's maybe ten lines of code. It won't fix a broken strategy, but it means a normal bad stretch costs less than it would have. That's the gap between riding out the streak and needing a 100% gain to recover from it.
This is one of maybe four rules that separate systems that survive five years from systems that look great for five months. We run ours live and publish the drawdowns, good and bad, at v33systematic.com.