Finishing Propensity Regression

Overview

This pillar analyzes fighters with unsustainable knockout or submission streaks that are statistically likely to regress to their career average. It helps identify overvalued fighters whose recent success is driven by variance, not a permanent skill increase.

What It Does

The model calculates a baseline 'Expected Finish Rate' for each fighter based on their entire career, controlling for opponent quality. It then compares this baseline to their actual finish rate over their last 3 to 5 fights. A significant positive deviation suggests the fighter is over-performing and is a prime candidate for negative regression, meaning they are less likely to get a finish in their next bout.

Why It Matters

Recency bias heavily influences trading markets, causing the public to overvalue fighters on exciting finish streaks. This pillar provides a quantitative counter-argument, highlighting when a fighter's odds are inflated by luck. This creates high-value opportunities to position against them or on the fight going to a decision.

How It Works

First, the system ingests career fight data from multiple MMA promotions. Second, it uses a logistic regression model to compute a career Expected Finish Rate (xFR) for every fighter. Third, it calculates the fighter's actual finish rate in their most recent fights. Finally, it generates a Regression Score showing the difference between recent performance and the career baseline, flagging fighters due for a statistical correction.

Methodology

The core calculation is the 'Regression Score' = (Actual_Finish_Rate_Last_5 - Expected_Finish_Rate_Career) / Standard_Deviation_of_Finish_Rate. The Expected Finish Rate is derived from a model including variables like significant strikes landed per minute, takedown success rate, and the defensive stats of their past opponents. A score above +1.5 is considered a strong signal for negative regression.

Edge & Advantage

This provides a direct statistical edge against the market's recency bias, allowing you to systematically fade fighters whose prices are inflated by an unsustainable finishing streak.

Key Indicators

• Regression Score

  high

  The standardized difference between a fighter's recent finish rate and their career expected finish rate.

• Expected Finish Rate (xFR)

  high

  A baseline finish probability calculated via regression, controlling for opponent quality and fight context.

• Finish Volatility

  medium

  Measures the standard deviation of a fighter's finish rate, indicating how streaky they are naturally.

Data Sources

• UFCStats.com

  Official source for detailed historical fight statistics, including strikes, takedowns, and round-by-round data.

• Tapology

  Crowdsourced fight records, event data, and fighter information across multiple MMA promotions.

• Sherdog

  Comprehensive database of fighter records and event history for professional MMA.

Example Questions This Pillar Answers

• → Will the upcoming fight between [Fighter A] and [Fighter B] go the distance?
• → Will [Fighter on a KO streak] win their next fight by KO/TKO?
• → What is the probability that the main event ends in a finish?

Tags