Losing streak probability calculator — what cold streak should you expect?

A 60% win rate sounds bulletproof. A 55% win rate sounds like a real edge. Until you run a hundred trades and hit eight losses in a row, watch your account drift down for a fortnight, and start doubting the whole thing. The reaction most traders have at that point is "the strategy must be broken." The math says: actually, no — that streak is fully expected, and a system that's never delivered one is a system that hasn't run long enough.

This calculator does the version of the math nobody shows you on the strategy-selling websites. Type in a win rate, a trade count, and a streak length. Get the probability of seeing at least one streak of that many consecutive losses across the run. The answers are going to surprise you in one direction every single time: the streaks are longer, and more probable, than your gut thinks. Pair it with the win-rate confidence calculator for the other half of the picture — how reliable your measured win rate even is at your sample size.

Why the answer is bigger than you think

Most traders carry an unspoken model in their head: a 60% win rate means losing streaks shouldn't really exceed three or four. After all, four losses in a row at 40% loss rate is 0.4⁴ = 2.56%. Sounds rare. So a streak of six or seven feels almost impossible.

The flaw in that intuition is the difference between "what's the probability of the next four trades all being losses" and "what's the probability that somewhere across 100 trades there's a four-in-a-row stretch." The first is a single-roll question. The second has to account for every possible four-trade window inside a 100-trade run, plus the way streaks chain when one window is a loss and the adjacent one extends it. There are roughly 97 different starting positions where a four-loss streak could begin in a 100-trade run, not one — and the math compounds accordingly.

Concretely: at a 55% win rate over 100 trades, the probability of seeing at least one streak of six losses in a row is around 25%. Bump streak length to seven and it's still ~12%. These aren't tail events — they're the median experience for someone running a strategy long enough.

What this changes about position sizing

If a six-loss streak is one-in-four likely across a hundred trades, then your risk plan has to survive it. Six losses at 2% risk per trade is a 12% drawdown — uncomfortable but recoverable. Six losses at 5% risk per trade is a 30% drawdown — the kind of loss most retail accounts don't psychologically come back from, even when the strategy itself was fine.

The arithmetic of recovery is unforgiving. A 30% drawdown requires a 43% gain to break even, not 30%. A 50% drawdown requires 100%. The tool's main job isn't to tell you whether to run the strategy — it's to make the worst plausible streak visible before you size positions, so the worst plausible streak is also a survivable one. For what to actually do during a losing streak — the cortisol curve, the 30-90 minute recovery window, why the "I feel fine" feeling lies — the recovery post covers the behavioural side the math doesn't.

The professional convention is sizing such that even an extreme streak (say, the 99th-percentile bad run) keeps drawdown under 20-25%. Plug in your strategy's actual win rate and a streak length one or two longer than what the calculator gives at 99.9% probability, and use that as the worst-case anchor.

What it doesn't model

This calculator assumes each trade is independent — every coin-flip is fresh, your win rate doesn't shift mid-run. That's a defensible approximation for short timeframes where market regime is roughly stable, but breaks down in two cases:

• Regime change. A trend-following strategy that wins 65% in trending conditions might win 30% in chop. If you're in chop, the "true" win rate during that period is much lower, and your streak probability balloons. Most multi-month drawdowns are regime drawdowns, not random ones.
• Correlated trades. If your strategy is taking BTC long, ETH long, SOL long, and XRP long simultaneously and they all stop out together, that's effectively four losses on one bad market day. The "trades" aren't independent — they're fused. The math here treats them as independent; reality treats them as one big bet.

Both effects make actual streak durations longer than this tool predicts. Treat its number as the optimistic version: in a stable regime with uncorrelated trades, here's what you'd see. In the messier real version, longer streaks are common.

What "win rate" actually means here

The win rate input is fraction-of-trades-that-close-positive. It treats a +5R win and a +0.1R win identically (both count once), and a -1R loss identically to a -0.1R loss. For most strategies this is fine — risk-per-trade is roughly fixed and the win/loss split dominates the math. For strategies with highly variable trade sizes, the streak-probability picture is the same but the financial damage from a streak shifts based on whether the losses inside the streak were the small ones or the large ones. The calculator gives you the count; sizing your trades consistently is what makes that count translate into predictable drawdown.

FAQ