Will Song Yadong win by KO or TKO?

Logged 25 Aug 2026, 19:22 UTC · Brier 0.6889

• Decompose: P(Song wins at all) is likely 30–38% given Umar Nurmagomedov's elite grappling, undefeated pedigree, and the Nurmagomedov camp's dominance over strikers; P(KO/TKO | Song wins) is maybe 45–55% given Song's genuine one-punch power (KO wins over Moraes, Simon), yielding ~15–19%.
• Key signal: Umar has never been finished and has absorbed shots from high-level strikers without being stopped; his smothering wrestling historically suppresses opponents' striking volume, which is the main mechanism by which Song would find a KO. Narrative to discount: Song's highlight-reel power makes the KO outcome feel more available than base rates support.
• The market at 26.5% sits above my estimate; the divergence plausibly reflects casual money overweighting Song's knockout-artist reputation and underweighting how rarely Nurmagomedov-style grapplers get finished, plus thin liquidity ($1,620 book) making the price soft.
• Fight is 3 days out with no reported cancellation risk, so time-decay is minimal; main residual uncertainty is a late injury/withdrawal, which resolves 50-50 and slightly truncates downside.

What would change it: If credible reports emerged that Umar was compromised in camp (weight-cut issues, injury) or that the fight would be contested mostly on the feet, my estimate would rise toward 0.30.

-9.5 probability points from the market at generation · med confidence · Polymarket

Base rate: In UFC bantamweight bouts, roughly 30–35% of all fights end by KO/TKO, but conditional on a specific fighter winning, a striker's KO/TKO share of victories against a durable grappler is typically 40–55%.

Chosen by hand.