Is It Real, or Is It Luck?

Flip a fair coin ten times and you can easily get seven heads. Nobody would call that a biased coin. Yet a trader with seven wins in ten trades will often call it a proven edge and size up.

Results from a small sample are dominated by chance. As a rough guide, fewer than 30 trades is noise. Around 100 trades per setup is where a pattern starts to mean something. Two hundred to four hundred gives you real confidence. This is not a precise rule, it is a reminder that conviction should scale with sample size, not with the warmth of a recent streak.

Most edges that evaporate were never edges. They were one of these.

• Curve-fitting: tuning a strategy so tightly to past data that it describes history instead of predicting the future.
• P-hacking: testing dozens of variations and keeping the one that looked best by chance.
• Recency bias: trusting the last few trades far more than the full record.
• Small-sample illusion: mistaking a short hot run for a durable advantage.

All four share one fingerprint. They look backward and call it forward. The defense is a validation process that deliberately tests the edge on data it has never seen.

Take an edge from idea to live size one rung at a time. Skipping rungs is how lucky runs get funded.

1. Backtest: does it show positive expectancy over a long history?
2. Walk-forward, out of sample: does it hold on data the design never touched?
3. Forward or paper test: does it survive in current conditions with no money on the line?
4. Live, small: does it survive real fills, real costs, and real emotion at minimum size?
5. Scale: only once the edge has held through each rung do you grow the size.

At every rung you are asking the same question. Is this still positive when it cannot be the result of fitting the past? An edge that clears all five rungs has earned your size.