top-quartile submission winner against a top-quartile knockout winner in ufc finish: +9.25 pts (n=151), confirmed on an era it had never seen. Context for reading a slate — it never adjusts a served probability.
Discovery applies a Benjamini-Hochberg false-discovery correction across every condition tested in this market at once, so a result here is already adjusted for how many conditions were tried. Confirmation re-runs the same test on a chronologically later era the engine had never seen. The eras differ by sport, because each sport has the history it has.
2024–2026: too thin to re-measure
A confirmed trend is context for reading a slate. It never adjusts a served probability, and it is not a bet — the probability you see on a pick comes from the model, and this measurement is not an input to it.
Identifier ufc_finish:grappler_vs_striker. Synced 2026-08-17.