Six losses in a row. The system is broken — obviously. Except a system that wins 45% of the time will hand you six straight losses roughly every 55 trades. If you trade a few times a day, that's a normal fortnight. Nothing broke. That's just what a 45% win rate looks like from the inside.
Edge is the average. Variance is everything else.
Your expectancy describes what happens on average over many trades. It says nothing about the path. Variance is the scatter around that average — and in the short run, variance is far louder than edge.
Over ten trades, variance dominates completely; the edge is invisible. Over a thousand, edge dominates and variance is a texture. The whole discipline of systematic trading is surviving the first regime long enough to reach the second.
In the short run, your results are mostly variance. In the long run, they're mostly edge. The trouble is you live in the short run.
The streaks are not optional
Some rough intuition for a 45%-win system: six losses in a row happens about every 55 trades. Eight in a row, every 250 or so. Ten straight will happen if you trade long enough. None of that is a malfunction — it's the arithmetic of a coin that lands your way slightly less than half the time but pays more when it does.
If your plan doesn't survive a ten-loss streak, your plan doesn't survive your strategy. That's not a variance problem, it's a sizing problem, and it's the one that actually ends accounts (risk of ruin).
Telling the two apart
So how do you know if it's noise or death? Honestly: not quickly, and not from the streak itself. What helps:
Know your streak distribution in advance. If you've simulated it (Monte Carlo), a six-loss run is a shrug rather than an emergency. Expected pain is manageable pain.
Watch the mechanics, not the P&L. Are the setups still appearing? Filling normally? Behaving as designed? A strategy that's still doing exactly what it was built to do and losing is probably in a drawdown. One that's firing in situations you don't recognise has genuinely changed.
Check the sample, not the feeling. A bad week is 20 trades. Your sample needs hundreds. The week has almost no informational content, however loud it is.
Look for a reason, not a result. Regimes do change and edges do decay. But the evidence should be structural — the conditions your edge depends on have gone — not just "it lost".
The failure mode
The classic account-killer isn't a losing streak. It's the response to one: change the system after six losses, catch the next six losses of the new version, change again. You end up with a strategy assembled entirely from reactions to noise, and every piece of it was added at the exact moment you were least able to judge it.
The counter-discipline is unglamorous. Decide in advance what would make you stop. Make it structural and specific, and larger than any streak you should expect. Then let the system run through the pain it was always going to produce.
See also: what is drawdown and emotional discipline.