📊 Track record

One rate per season, not one rate for everything. A World Cup and a league matchday are different problems: knockout ties are tight and heavily previewed, an August league round is a coin-flip with half the squads still out of shape. Averaging them produces a number that describes neither, so each season is graded on its own and the archive stays visible next to it.

Percentages need a sample. Under 20 graded matches the rate is dimmed and the count leads. A competition sitting at 1/9 and one at 7/10 are, after nine games, mostly telling you about the fixtures they drew — one more result moves either of them by eleven points.

Compared against what. Always backing the home side calls about 45% of outcomes and the bookmakers' favourite about 55%; the single most common scoreline lands around 11% of the time. Those lines are marked on the bars above — a win rate without one means nothing.

A matchday is a recap, not a rate. Five predictions cannot produce a meaningful percentage, so rounds are reported as hits over calls. For direction over time, read the last-20 form strip instead.