Every forecast we publish gets tracked and scored. We use Brier scores (lower is better, 0 = perfect) to measure calibration. No cherry-picking. Full transparency.
Hit rate = share of resolved forecasts where the predicted direction matched the outcome. The reference is a coin flip at 50%: a directional hit rate below 50% shows no demonstrated edge on direction — consistent with our own finding that 24-hour price direction is not reliably predictable.