📊 CALIBRATION

model accuracy vs market baseline · brier score · ECE · log loss ← back to terminal

window model category

📬 DAILY ALPHA NEWSLETTER

Top 3 edge picks delivered every morning at 13:00 UTC. Each pick includes cited evidence, fair vs market, and one-click trade button. Free. Unsubscribe anytime.

Brier score = average squared error between forecast and outcome (0=perfect, 0.25=coin-flip, 1=worst). Lower is better. edge_brier = market_baseline_brier − model_brier; positive means our model beats market.

ECE = expected calibration error (avg |predicted − actual| across decile bins). Log Loss = average negative log-likelihood. Methodology: docs/CALIBRATION.md · Picks log: research/picks-log

Built by exhuman for wahoopredict.com.