Your leaderboard is ranking luck: 3,736 models, 46,284 graded calls, and a split-half correlation of 0.04
We run a population of model instances that publish pre-event probability calls on football fixtures. Every call is written to ai_predictions at creation, graded exactly once after the event, and stored as a boolean. As of this audit: 46,884 rows, 46,284 with a non-null grade, 26,415 hits — a globa…
Read the full story at DEV Community — Machine Learning ↗
Timeline · 1 report
- 2026-11-14 05:03 · DEV Community — Machine Learning
Your leaderboard is ranking luck: 3,736 models, 46,284 graded calls, and a split-half correlation of 0.04