A 55% win rate sounds comfortable until you count the coin flips. Over a few hundred trades, a system that good will still hand you a run of seven, eight, sometimes ten losers back to back — not because it broke, but because that is exactly what a 55% edge does over enough attempts.
Streak probability from your win rate
Start with the building block. If your probability of losing any single trade is L (so L = 1 − win rate), and trades are independent, the chance of losing k in a row is:
P(k losses in a row) = L^kAt a 55% win rate, L = 0.45. A run of five losers is 0.45^5 ≈ 1.8% on any given starting point. That feels rare. But you are not taking one shot at it — you are rolling the dice on every trade, all year. The right question is not “how likely is a streak here?” but “how long a streak should I expect somewhere across all my trades?”
Work in R-multiples so the streak math maps to real damage. A string of losers at −1R each costs a predictable amount; a string that includes a few −1.5R breaches hurts more. If you haven’t standardized outcomes this way, the r-multiple concept is the prerequisite for everything below.
The longest streak expected over N trades
The expected longest losing streak over N trades grows with the logarithm of N — slowly, but relentlessly. A workable approximation is:
longest expected streak ≈ log(N) / log(1 / L)Run it for a 55% win rate (L = 0.45):
| Trades (N) | Approx. longest losing streak |
|---|---|
| 100 | ~6 |
| 250 | ~7 |
| 500 | ~8 |
| 1,000 | ~9 |
The lesson: more trading guarantees longer streaks. A trader who takes 1,000 trades a year should plan to eat a nine-loss streak, not be shocked by it. Drop the win rate to 45% and those numbers climb sharply — every extra point of loss probability lengthens the worst run you’ll meet. Trend systems that win less than half the time but pay large R survive precisely because their operators expect double-digit streaks and size for them.
Why streaks feel like broken edges but aren’t
Eight losers in a row feels like the edge died. Your brain, wired for patterns, reads the run as information — “the market changed, the setup stopped working.” Usually it is just the log-N math playing out on schedule.
The tell is whether the streak length is within what your win rate predicts. If your model says a 45%-win system should throw a run of nine somewhere in 1,000 trades and you hit nine, that is not a broken edge — it is a working edge on a bad day. The genuine warning sign is a streak well past the expected longest, or a change in the character of losses (bigger, faster, at odd times), which is a different signal than length alone.
This is where a journal that records the context of each trade earns its keep. Shibiki auto-journals every trade and tracks live edge health per strategy with a Wilson confidence interval, so a streak shows up as the interval widening rather than as a gut-feel emergency. If your true edge is intact, the confidence band absorbs the streak and re-tightens; if the edge is actually degrading, the band drifts and tells you with statistics instead of panic.
Streaks, drawdown, and prop-firm daily limits
A losing streak is a drawdown with a schedule, and that is where prop rules bite. Most firms enforce a daily loss limit and an overall max drawdown, and both change between firms and programs — always confirm the exact figures with the firm before you trade them. The point here is structural: a normal streak must not, by itself, breach either line.
Two ways streaks kill funded accounts:
- Same-day clustering. Take enough trades per day and part of your expected streak lands inside one session, stacking losses against the daily limit.
- Compounding into the max. A long streak drags the account toward the overall drawdown floor, and because many firms trail that floor on your equity peak, the room you have shrinks exactly when you need it. See the trailing drawdown explainer for why the floor moves.
Model it before it happens: run your expected worst streak against a firm’s thresholds with the prop-firm drawdown calculator, using a firm like FTMO as a reference for how daily and overall limits stack.
Sizing so a normal streak survives the rules
The fix is arithmetic, not willpower. Take your expected longest streak for the number of trades you’ll place, assume the losses land close together, and set risk-per-trade so that pileup still leaves the account inside every limit — with margin for the streak that runs one or two past expectation.
- Compute expected longest streak for your real trade frequency.
- Multiply by your per-trade risk to get the streak’s worst-case cost.
- Confirm that cost clears the daily limit and the overall max, with a buffer.
- If it doesn’t, cut size or cut trades per day — those are the only two knobs.
Shibiki turns that math into a hard risk limit enforced at the broker, so the size that survives your streaks is a rule the EA holds even mid-slump. Cross-check the raw numbers against your edge using the expectancy calculator — a system with strong expectancy can weather longer streaks because the winners between them are larger.
Mental prep: pre-committing to the plan
The hardest part of a streak is that the correct action — keep executing the plan at unchanged size — feels most wrong exactly when it matters. So decide before the streak, in calm conditions, what you will do during one.
Pre-commit in writing: the size you’ll trade, the daily stop that ends the session, the streak length that is still “normal” versus one that triggers a review. When trade eight loses, you are not making a fresh decision under stress — you are following a plan you already trust. Auto-journaling makes that trustworthy after the fact, because you can look back and confirm the streak was ordinary variance rather than rewriting the story in hindsight.
Related: Trailing drawdown · Prop-firm drawdown calculator · FTMO overview