"Fair" is the model's no-vig price for that exact outcome; "book" is the best live sportsbook price for it (hover for the book). A green book price is longer than fair — value if you trust the model. Method prices update with each odds sync.
The model makes Rongzhu a narrow 51% favorite. On the feet, Rongzhu sets the tempo — 5.4 significant strikes a minute to 4.3 — so Garcia needs moments, not volume. Garcia tends to settle in for the long haul — 9 of 12 tracked fights reached the judges. Rongzhu's most likely path is on the scorecards (37%), and the fight goes the distance 71% of the time. The live danger for Rongzhu: Garcia finishes 42% of wins by submission and Rongzhu has been submitted in 67% of losses.
Model: recent-fight output & defense rates (ESPN box stats), career method splits, durability, layoffs, and physical edges through a bounded logistic; weights tuned by walk-forward backtest on 1,500+ UFC fights — 61% winner accuracy and 32% exact winner-plus-method on 204 held-out fights (vs 26% picking favorite-by-decision). Probabilities are discounted in boom-or-bust matchups (both fighters' bouts ending fast) and squeezed toward the betting market's implied odds where a line exists — divergence beyond 10 points counts half. Experimental — size responsibly.