METHODOLOGY
THE MODEL
OctagonIQ uses a Monte Carlo simulation engine to generate fight predictions. For each matchup, the model runs 100,000 fight simulations using 25 input variables derived from each fighter's historical performance data.
INPUT VARIABLES
The simulation considers striking output and accuracy, absorption rates, takedown frequency and success, submission attempts, defensive metrics, finish rates by method, fight pace (rounds fought vs. finished), win/loss recency, opponent quality adjustments, and physical attributes including reach differential and stance matchup history.
SIMULATION PROCESS
Each simulation plays out a fight round by round, using probability distributions built from each fighter's stats. The model accounts for fatigue curves, comeback probability by round, and method-specific finish rates. After 100,000 iterations, the results are aggregated into win probabilities, method of victory distributions, and round-by-round finish likelihood.
ODDS COMPARISON
The model's implied probabilities are compared against real sportsbook odds to identify potential edges — situations where the market price diverges from the model's estimate. A positive edge means the model sees more value than the market; a negative edge means the market agrees with or exceeds the model's assessment.
DATA SOURCES
Fighter statistics are sourced from UFCStats.com, the UFC's official statistics provider. Odds data is provided by The Odds API, covering 15+ US-regulated sportsbooks. All data is refreshed automatically after each event.
LIMITATIONS
No model captures everything. The simulation does not account for injuries, weight cut quality, training camp changes, motivation, or other intangible factors. Fighters with fewer than 3 UFC fights have limited data, and predictions for these fighters carry higher uncertainty. All predictions are for informational and entertainment purposes.