What is actually winning

Returns are simple price returns over the aligned window. Dividends, coupons, storage costs and rolling costs on futures are not included, so income-paying assets are understated against those that pay nothing.

Volatility is the standard deviation of daily returns annualised by the square root of 252 trading days. Crypto trades every day of the year, so its annualised figure is slightly conservative under this convention; using two different conventions in one table would be worse.

Per risk is return divided by volatility. It is not a Sharpe ratio, because no risk-free rate has been deducted. Every figure here describes the recent past and carries no predictive claim whatsoever.

Series are aligned on the dates every market shares, so a holiday on one exchange removes that day from all of them. That is why the window is shorter than the calendar period it covers, and why adding a market with an unusual calendar shortens it further.

102 trading days, roughly 5 months, is long enough to show what has happened and still too short to show what tends to happen. A correlation measured over one period can reverse over the next, particularly between assets with no structural link to each other.

Best day and worst day are here because a total return conceals its own shape. Two markets can finish the window level while one drifted there and the other did it in a single session, and those are not the same experience to hold.

Days up counts sessions that closed higher, regardless of size. A market can rise on most days and still finish lower, which is what a small number of large falls does to an otherwise steady series.

Everything here is descriptive. Nothing on this page ranks, recommends or forecasts any of these markets, and past behaviour over a window this size carries no information about the next one.