Percentile rank
How do you place one unusual week against every week that came before it?
In plain words
A percentile rank states where a value falls inside the distribution of all earlier values of the same kind. At the top of the ranking, the week beat every prior week in the sample; near the middle, it was unremarkable.
Why it matters
Raw percentage moves mean little without comparison. A large weekly return may be routine for one asset and unheard of for another. Percentile rank normalises the move against the asset's own history, so a reader can tell an ordinary week from a genuinely exceptional one, and can judge whether recent conditions have shifted.
In a real week: ETH, week of 17 December 2018
In the week of 17 December 2018, ETH returned 52.37%, ranking at the 100th percentile against its 44 weeks of history. No earlier week in that sample produced a larger gain. The same week's range was 56.33%, its annualised realised volatility 146.0%, and it finished 30.68 percentage points above BTC.
A common mistake
Reading a high percentile as a forecast of more gains. It only says where the completed week sat relative to a short history, and a sample of only a few dozen weeks can make ordinary events look unique.