Introduction

Mixed Martial Arts (MMA) has grown into a global phenomenon, and the Ultimate Fighting Championship (UFC) stands at its pinnacle. As the sport continues to grow, understanding the factors that influence fight outcomes and the overall trajectory of the UFC becomes increasingly intriguing. With the availability of rich data on fighter characteristics, fight outcomes, and betting odds, there is an opportunity to uncover insights into factors that influence fight success and outcomes over time. This report uses statistical analysis and data visualization to investigate key questions about UFC fights. Our findings aim to provide a deeper understanding of the sport’s dynamics and trends, benefiting analysts, enthusiasts, gamblers and stakeholders in the MMA world.

Data & Research Questions

The following analysis utilizes a dataset containing information on UFC fights. The dataset comprises detailed records of 6,478 UFC fights, with 118 features describing fighter characteristics, fight outcomes, betting odds, and event details. Key variables include:

  • Fighter Information: Names, odds, expected value, and attributes of the “Red” and “Blue” fighters in the Red and Blue corners respectively (e.g., height, weight, reach).
  • Event Details: Date, location, and country of the event.
  • Outcome Metrics: Winner, fight duration (in seconds), round of finish, and method of victory (e.g., knockout, submission).
  • Betting Odds: Including decision, submission, and knockout odds for both fighters.

Each row represents a single fight, and columns provide granular data to enable exploration of temporal trends, correlations, and predictive factors in UFC outcomes.

The dataset will be used to answer the following three research questions:

  • How do specific fighter characteristics (e.g., weight, gender, reach) influence fight outcomes?
  • How do pre-fight expectations (odds) correlate with fight outcomes and duration?
  • What temporal trends exist in UFC fights and outcomes over time?

The subsequent sections delve into these questions, supported by interactive visualizations, statistical models, and interpretations that connect these findings to broader narratives within the MMA landscape.

Effects of Fighter Characteristics on Fight Outcomes

To address our first research question, we begin by examining the distribution of fights by weight class. This distribution provides essential context for interpreting the win percentage data, as it helps understand the significance and reliability of the winning patterns observed across different weight classes based on their sample sizes.

The histogram above reveals significant disparities in fight totals across weight classes, with Lightweight being the most active division at over 1050 fights, followed closely by Welterweight at around 1000 fights. The male divisions generally show higher participation numbers than female divisions, with Lightweight, Welterweight, Featherweight, and Bantamweight all having well over 600 fights. In contrast, female divisions have notably smaller counts, with Women’s Strawweight being the most active female division at about 300 fights, while Women’s Featherweight is the least active with fewer than 50 fights. The male divisions show a bell-curve-like distribution centered around the middle weight classes, while female divisions maintain relatively consistent numbers across their available weight classes, except for the notably small Women’s Featherweight division.

Next, we aim to explore whether there are any patterns or biases in fight outcomes based on the weight class and gender of the fighters.

The plot illustrates win percentages across different weight classes in UFC, divided between male and female fighters, with a clear advantage for Red corner fighters across all divisions. In the female divisions, Women’s Featherweight shows the most pronounced Red corner dominance at approximately 62% wins, while other female weight classes (Bantamweight, Flyweight, and Strawweight) maintain a consistent pattern with Red corner wins around 55-58%. The male divisions display a remarkably uniform pattern across all weight classes, with Red corner fighters consistently winning about 55-60% of their matches and Blue corner fighters winning approximately 40-45%. This systematic advantage for Red corner fighters appears across both genders and all weight classes, suggesting a structural factor favoring Red corner placement rather than random variation. As we don’t see a clear relationship between weight class and fight outcome, we explore other fighter characteristics and their relationship with outcomes.

In addition to the plot above, we examine the effects of reach advantage on significant strikes landed and fight outcome to complement our question of interest.

The scatter plot reveals the relationship between reach advantage and significant strikes landed in UFC fights, with fight outcomes color-coded as wins (turquoise) and losses (salmon). The data shows a very little to no correlation between reach advantage and strike success, as fighters with both positive and negative reach differentials achieve varying levels of striking success. Most fights cluster in the range of -15 to +15 cm reach advantage, with the average significant strikes landed typically falling between 2 and 7 significant strikes per round. Winning fighters (turquoise points) appear to be slightly more prevalent in the higher reach advantage range, with some slight prevalence in higher areas of strikes landing, suggesting that reach advantage may be more important for victory than strikes landing alone. The spread of the data points suggests that while reach advantage might offer some tactical benefits, more testing would be required to see if it is a definitive predictor of striking success or fight outcome in UFC competitions.

Effect of Fighter Odds and Other Variables

We begin exploring our second research question by analyzing the distribution of fighter odds by corner.

This plot above shows two overlapping histograms, one for RedOdds and another for BlueOdds, giving an insight into the distribution of both sets of odds in the dataset. It is clear from the plot that red fighters tend to be favorites, as the red histogram has peak around -150 while the blue has its peak around +150. Both seem to have unimodal, normal distributions with most of their values between the -300 and +300 range.

Now that we have the appropriate context, we will analyze the relationship between fighter odds and fight outcome.

The plot above shows a clear negative relationship between the odds of the two fighters: as one fighter’s odds decrease (indicating a stronger favorite), the other fighter’s odds increase. Further, the colors of dots representing the winner of each fight also show that odds appear highly correlated with actual outcomes. For example, the more negative the red fighter’s odds on the x-axis are, the more favored they are to win. This is correlated with most of the red dots (representing red winning) occuring around the region where the red fighter’s odds are more negative (odds favorite). This plot is particularly informative because it visually captures the predictive power of betting odds in determining fight outcomes. By showing the odds for both fighters and the corresponding winner, it highlights whether significant differences in odds align with expected outcomes, offering insights into the effectiveness of odds as a measure of a fighter’s likelihood of winning. This visualization could help fans, analysts, and bettors understand the role of odds in UFC fight predictions and make more informed decisions based on observed trends.

To supplement these findings, we test the relationship between red fighter odds and fight winner (1 if Red, 0 if blue).

## 
## Call:
## lm(formula = Winner_Numeric ~ RedOdds, data = ufc)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -1.3845 -0.4343  0.2327  0.3885  0.8896 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  5.107e-01  6.355e-03   80.35   <2e-16 ***
## RedOdds     -6.111e-04  2.128e-05  -28.71   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4639 on 6257 degrees of freedom
##   (219 observations deleted due to missingness)
## Multiple R-squared:  0.1164, Adjusted R-squared:  0.1163 
## F-statistic: 824.3 on 1 and 6257 DF,  p-value: < 2.2e-16

The linear regression analysis between RedOdds and the fight outcome (coded as Winner_Numeric) shows a statistically significant relationship, with a p-value less than our significance level of 0.05 for the RedOdds coefficient (< 2e-16). The negative coefficient for RedOdds (-0.0006116) indicates that as the RedOdds increase, the likelihood of a “Red” win decreases, suggesting that higher odds for the “Red” fighter are associated with a lower probability of winning. Given the strong significance of the model and the large sample size, we can conclude that there is a meaningful relationship between the odds for the “Red” fighter and the likelihood of winning, though other factors not captured in this model likely also contribute to the fight outcomes. We must note that the model may produce biased or imprecise results if the relationship between our variables is nonlinear or if there exists other confounders that may explain this relationship.

Lastly, we analyze the relationship between fighter odds vs their current win streak, to see if fighters with better odds are performing relatively better.

This plot explores the relationship between the current win streak and fighter odds for red fighters, as well as whether they won or lost. The first noticeable trend is that red fighters are often favored to win (represented by lower odds), with a significant number of them having lower odds and winning. On the other hand, fighters with higher odds (indicating they were less favored) tended to lose the fight. Additionally, the plot suggests that as a fighter’s win streak increases, the odds become more balanced, indicating that fighters on a winning streak may face stronger opponents. This results in odds that are less skewed, possibly reflecting a more evenly matched contest.

Results

  1. Fighter Characteristics and Fight Outcomes Analysis of win percentages by weight class and gender revealed a consistent advantage for Red corner fighters, who won 55-60% of matches across all divisions. This advantage was even more pronounced in female divisions, particularly Women’s Featherweight, where Red corner fighters achieved a 62% win rate. Examining reach advantage as a potential factor showed no strong correlation with significant strikes landed or fight outcomes, suggesting that while physical advantages may be useful, they are not definitive predictors of success.

  2. Odds and Predictive Accuracy Betting odds proved to be a strong predictor of fight outcomes. Fighters with lower (more negative) odds were more likely to win, as demonstrated by the correlation between RedOdds and fight winners in regression analysis (p-value < 2e-16). Interestingly, fights involving strong favorites tended to end faster, especially when favorites won. However, underdogs who triumphed could have employed high-risk, decisive strategies, leading to shorter fight durations, especially among Blue corner fighters.

  3. Temporal Trends in UFC Fights The UFC experienced consistent growth from 2010 to 2023, with the number of fights increasing steadily before stabilizing around 500 events annually. Red corner win rates, historically hovering between 55-65%, exhibited a declining trend post-2015, reflecting a potential increase in parity among fighters or evolving matchmaking strategies. Peaks in Red corner success during 2014, 2015, and 2020 were notable but short-lived, highlighting dynamic shifts in competitive balance over time.

Conclusion

This analysis highlights several key findings about UFC fight dynamics. Even though corner color assignments appear cosmetic in nature, the Red corner consistently held a structural advantage (relating to odds) across most weight classes and genders, although this dominance has been waning in recent years. Temporal analysis reflects the UFC’s evolution into a global sports organization with increasing competitive balance. Betting odds emerged as a valuable tool for predicting outcomes, with strong favorites generally outperforming expectations and underdogs showcasing unique patterns in fight duration.

Future research could delve deeper into how other characteristics like age, experience, and styles relate to fight odds and outcome. Moreover, one could analyze striking and grappling metrics like takedowns and submission attempts to reveal more about in-fight factors that may lead to victory. Lastly, location and audience could be examined to see if home-court advantage or time zone differences could influence outcomes and betting patterns.