How far below its peak does a 55% strategy fall? Exact numbers
hypothesess.com, AI-assisted writing ยท
A strategy that wins 55% of its trades sounds safe. It is a real edge, and over 250 trades of equal 1% bets it earns about 25% [1]. It also, at some point along the way, falls about 11% below its best level in a typical run, and 18% or more in one run out of ten [1]. Those falls are not a sign that something broke. They are what this edge looks like from the inside, and knowing their size in advance is what lets you keep trading through them.
What luck does to a real edge
The worst fall from a peak, the maximum drawdown, comes from losing streaks clustering more than intuition expects. With a 55% win rate, losing four or five in a row is ordinary, and two such streaks close together are enough to take the account 8% to 10% below its high [1].
Here is the whole distribution, computed exactly for equal bets of 1% of the starting account [1]:
| Win rate | Trades | Typical worst drop | One run in ten | One run in a hundred | Chance of a 20% drop |
|---|---|---|---|---|---|
| 55% | 100 | 7% | 13% | 20% | 1% |
| 55% | 250 | 11% | 18% | 26% | 6% |
| 55% | 1,000 | 16% | 25% | 36% | 29% |
| 60% | 250 | 8% | 12% | 17% | 0.4% |
| 50% (no edge) | 250 | 17% | 30% | 43% | 39% |
Two things stand out. Drawdowns keep growing with time even when the edge is real: over 1,000 trades the typical worst fall is 16%, and a fall of 20% or more happens in more than a quarter of runs [1]. And the edge helps a lot: at 60% the same 250 trades typically fall 8%, and a 20% fall becomes rare [1].
Why this matters before you start
A trader who expects a smooth curve stops trading a working strategy at its first 10% fall, which for a 55% strategy over 250 trades is more likely than not: it happens in 63% of runs [1]. The rule that protects you is set before the first trade: the size of fall your strategy produces by luck, and the size that should make you look for a broken edge.
The same arithmetic also says what a drawdown cannot tell you. A no-edge strategy over 250 trades typically falls 17% [1]; a 55% one falls 11%. The ranges overlap widely, so a single drawdown is weak evidence either way. The win rate over enough trades is the better test, as the entry on telling an edge from luck works through.
Nerd corner
With equal bets the account moves one unit up or down each trade, so the drawdown is the depth of a random walk with drift below its running maximum. Its exact distribution comes from stepping through the trades and tracking that depth: a win brings it one closer to the peak, a loss takes it one further down [1]. A simulation of 40,000 runs gives the same answers within its own noise, for example a median of 11 and a 90th percentile of 17 to 18 bets at 55% over 250 trades [1].
Betting a fixed share of the current account instead of the starting one compounds the falls slightly: a 10-bet drop at 1% a trade becomes a 9.6% fall rather than 10% [1]. The order of magnitude, and every conclusion above, stays the same.
Real strategies have losses of uneven size and streaks that cluster in volatile markets. Both make drawdowns deeper than this model, never shallower.
Check yourself
Over 1,000 trades at 55%, is a 20% fall a sign the strategy is broken? Not on its own: it happens in about 29% of runs with a real edge [1].
Why does a 60% strategy fall so much less than a 55% one? Its expectancy per trade is twice as large, 0.20% against 0.10%, so it climbs away from each loss faster and the streaks have less room to stack up [1].
You halve your bet from 1% to 0.5%. What happens to the typical 11% fall? It becomes about 5.5%: the drop in bets stays the same and each bet is half the size [1].
An example strategy, not trading advice. A lab notebook entry for hypothesess.com, written with AI and checked by hand; the drawdown figures are computed exactly and confirmed by simulation, and the box above scales them to your bet size.