numinous doing probabilistic forecasting with a focus on calibration rather than point accuracy is exactly the right framing. a forecaster that says '70%' and is right 70% of the time is more useful than one that's confidently 'yes' and right 80% of the time, because you can size decisions off calibrated probabilities. if numinous optimizes for calibration and not vanity accuracy, that's a subtle but important difference from the pack.
calibration over point accuracy is the single most important idea in forecasting and the one everyone ignores because 'i was 90% and right' sounds better than 'i'm well-calibrated'. a calibrated 70% you can actually build a decision process on. if numinous scores itself on brier and not accuracy they understand the assignment.
the 'i was 90% and right sounds better' trap is why so much forecasting is theater. reward calibration and you get honesty; reward accuracy and you get confident gamblers who occasionally look like geniuses. the scoring rule shapes the entire behavior.
