Statistical significance
Is a market result big enough to count as real, or could ordinary chance have produced it?
In plain words
Statistical significance compares an observed result with the size of the random variation that normally surrounds it. A result is called significant when it is larger than what chance would usually produce; a small, unremarkable result is treated as noise.
Why it matters
Most trading ideas are judged from a handful of weeks. Without comparing a move to typical variation, a trader can mistake a single lucky week for evidence of a repeatable edge, and size positions on a pattern that will not persist.
In a real week: BTC, week of 2 June 2025
In the week of 2 June 2025, BTC closed 0.09% higher after a week whose range was 6.4% and whose realised volatility was 31.1% a year. The net change was tiny relative to that week's own swings. Against 381 weeks of history, this return sits at the 46.0th percentile, an ordinary result that chance would routinely produce.
A common mistake
Treating any small p-value as proof, or a large one as proof of no effect. Statistical significance only says a result is unusual under a stated assumption, not that the effect is large or will repeat.