More than most people think. A coin with no edge wins 5 trades in a row 1 time in 32 — so in any crowd of traders, plenty of "hot streaks" exist by luck alone. Small samples prove nothing: even 7 wins out of 10 happens by chance about 1 time in 6. Skill only shows up in large numbers.
What are the odds of a winning streak with zero skill?
Surprisingly good — a streak of 5 happens 1 time in 32 with no edge whatsoever. Model a skill-free trader as a coin flip: 50% chance of winning each trade. The chance of 2 wins in a row is ½ × ½ = 1 in 4. Each extra win halves the odds again:
| Streak length | Chance by pure luck |
|---|---|
| 3 wins in a row | 1 in 8 |
| 5 wins in a row | 1 in 32 |
| 7 wins in a row | 1 in 128 |
| 10 wins in a row | 1 in 1,024 |
One in 32 is not rare. It's roughly the chance of rolling double sixes — annoying to wait for, guaranteed to show up if you keep rolling. And markets contain a lot of people rolling.
Why do so many "hot" traders exist if streaks are luck?
Because the crowd is enormous, and rare-for-one is common-for-many. Suppose 1,000 people each start trading with zero skill — pure coin flips. About 31 of them (1,000 ÷ 32) will open with five straight wins. Give the crowd 10,000 members and about ten will hit ten straight wins — with literally no edge. Those ten will feel like geniuses, post their streaks, and attract followers, while the thousands of coin-flippers who lost quietly disappear from view. This is called survivorship: you only ever hear from the streaks. So when you see an impressive run — someone else's, or your own backtest's — the first question isn't "what's the secret?" It's "how many people were flipping?"
Why don't 10 trades prove anything?
Because almost any result is reachable by luck in a sample that small — the arithmetic is quick. Flip a fair coin 10 times: there are 2¹⁰ = 1,024 possible win/loss sequences. Count the sequences with 7 or more wins: 120 ways to get exactly 7, 45 ways for 8, 10 ways for 9, and 1 way for 10 — 176 in total. So a no-edge strategy scores an impressive-looking 7-out-of-10 (or better) with probability 176 ÷ 1,024 — about 17%, roughly 1 time in 6. You would not call a die skilled for rolling a six. A 10-trade backtest with a 70% win rate deserves exactly as much applause: none yet.
How do you tell luck from skill?
Two ways: demand more trades, and demand success on data the strategy has never seen. More trades first — luck thins out as samples grow. Winning 60% of 10 trades happens by chance about 1 time in 3; winning 60% of trades happens by chance only about 3 times in 100. Same win rate, completely different meaning. Second, out-of-sample behavior: a strategy tuned on last year's data can "win" by memorizing the past, so test it on a later period it never saw, or run it forward in simulation before funding it. Skill persists on new data; luck doesn't. Liquid Edge — non-custodial automated algorithmic crypto trading and portfolio management — is built around that discipline: lets you rack up trade counts on historical and live-simulated data before committing funds. The honest limit: no sample size ever skill — markets change, and yesterday's genuine edge can be gone tomorrow. Big samples don't grant certainty; they just make luck a worse explanation.
