Log loss
Log loss — a scoring rule that penalises confident errors far more severely than Brier score does.
Log loss takes the negative logarithm of the probability assigned to the outcome that occurred, averaged over all forecasts.
Log loss = −(1/N) × Σ [ o × ln(p) + (1 − o) × ln(1 − p) ]A naive 50% forecaster scores ln(2) ≈ 0.693.Why it matters: as a wrong forecast approaches certainty the penalty approaches infinity. For a model that will be staked on, that is the correct incentive — confident errors are what produce large losses.
Common mistake: comparing log loss across different markets or sports. The scale depends on the base rate, so only same-event comparisons are meaningful.
Related
Check any of this against our record
Every signal CONSENSUS publishes carries the bookmaker odds fixed before the event starts and the settled result afterwards — including the drawdowns and the losing runs. The running total is on the front page and every entry is in the log.