The only way to trust a model is to track every prediction it ever made. This page aggregates every analysis that has a recorded match outcome and shows the raw performance — good or bad. No cherry-picking.
No graded analyses yet
Accuracy tracks model predictions vs actual results — it needs both an analysis row and an outcome row to grade each pick. The Elo trainer has plenty of historical outcomes loaded; this dashboard lights up once analyses run on matches that subsequently finish.