Rankings in MMA have become a major talking point recently. Rankings in MMA are generally created by one or more media members for a particular news website. According to reports in the media, the 'official' rankings for the UFC are determined by the votes of a worryingly large pool of people ranging from established names from major websites to bloggers. There appears to be few to no rules established for the creation of each member's rankings. This resulted in the now much discussed rant by UFC president Dana White about Chael Sonnen's middleweight ranking. After reviewing the various points being made about the difficulty with these rankings I determined to create my own method for establishing and maintaining MMA rankings. There are a few key obstacles to overcome that are particular to MMA; or perhaps combat sports in general.
The first obstacle to MMA rankings is confusion as to the purpose of a ranking system. The UFC matchmakers insist that their job is to build contenders, rather than trying to showing who beat who. Therefore this new ranking system has been dubbed the Contender Generator. The second problem to overcome is the all too common circular MMA Math problems. For instance, Dan Henderson defeated Mauricio 'Shogun' Rua, who defeated Lyoto Machida, who defeated Dan Henderson. Third is the small amount of data available overall. Simply stated fighters don't fight very often, or many times in total. The solution to these problems with MMA rankings is to develop and maintain a formula-based point system.
Many rankings seem to be based on a combination of 'who beat who' and 'gut feelings' systems. In the example above Dana White was animated about how Chael Sonnen had defeated multiple fighters ranked above him at middleweight. It feels like a number of people who develop rankings get stuck in that mindset. But if the mission of UFC matchmakers is to build (or generate...) contenders, then having Sonnen ranked at 185 is senseless. To the extent you believe anything Chael Sonnen says, once he declared the move to the light heavyweight division his ranking at middleweight was forfeited. That isn't to say he couldn't be ranked if (when?) he returns to 185. The other half of the equation is based on personal feelings. For instance, some opposed a high ranking for Michael Bisping because of a perception of how he would have performed against Anderson Silva. Other examples include needing to get a signature win, being too boring, or runaway hype trains. The Contender Generator uses a number of criteria to assess the impact of each fight on the participants. It also has rules regarding eligibility pursuant to it's purpose of supplying contenders for titleholders. So stay put, Chael.
The fact that MMA Math has become part of the lexicon of the sport should make something clear. Useful rankings cannot be held hostage by MMA Math. They must be able to respond to the progress in a division. The Contender Generator establishes a point value for each fighter when they enter a division. It then uses weighted variables to manipulate each fighter's score, generally as a direct result of a fight. A convincing win should generally place a fighter above their opponent in the rankings but that result cannot be allowed to limit future changes. The current state of the 205 pound division is clear evidence that leapfrogging is the reality of the sport.
Unfortunately we cannot artificially increase the number of matches a fighter participates in (sorry UFC Undisputed). But we can have an effect on the value of each win or loss in a fighter's career. After spending some time contemplating various fights I came up with a number of factors for the first edition of the Contender Generator. These factors are all used to provide a net change to a fighter's score. This means that wins and losses can be enhanced or tempered by other factors than getting 'the W' or not. They are listed and briefly discussed below.
Win/Loss: If a fighter wins, this factor is positive, if they lose, it's negative. Draws or no contests are neutral.
Finish/Decision: If a fighter finishes their opponent, this factor is positive, if they win via decision, it is *neutral. *Please see next factor before complaining.
Dominance: This factor is graded on a scale and is meant to reward the fighter that seems to be the dominant fighter over the course of the match. For example, Yushin Okamai was dominant against Tim Boetsch for two solid rounds before succumbing to a third round charge. In a fight that is considered reasonably even this factor does not impact fighter score.
Expected Difficulty: This factor is graded on a scale and is meant to both slow the rise of fighters winning fights they are supposed to, and reward fighters who win fights against heavy favorites. In a fight that is expected to be reasonably even this factor does not impact fighter score.
Inactivity: Many in the industry believe in the concept of ring rust. So does the Contender Generator. A layoff has to be of a certain length before this factor has any impact. The impact builds slowly over time after that.
Title Loss: Losing a title fight has additional negative impact on a fighter's score. In many cases this provides the opportunity for a new contender to a successful defending champion. A long reigning champion will generally have sufficient score to indicate the need for a rematch. See: Anderson Silva vs. Chris Weidman II.
X Factor: This factor is generally not used, but basically includes anything else that affects the impact of a match. The impact of these is generally small, but every bit counts. This factor will be used less as the Contender Generator expands to accommodate more factors. Punishment for fouls and cheating are within the referee's discretion, their impact on rankings is at the Contender Generator's. Missing weight is part of this factor. Another example is a fight where both fighters somehow come off looking very impressive (tough?) regardless of the official result. If you've watched a lot of MMA, you've seen some of those fights. Again, this factor hasn't seen extensive use and the actual impact when used has been minor.
The goal of the Contender Generator is in the name. A sport that is trying to reach the elusive 'casual fans' around the world needs a reasoned approach to rankings and determination of title contenders. The rules and scoring factors used by the Contender Generator provide that approach. Of course the same reasoning mind that developed a spreadsheet to quickly and easily calculate a series of complex scoring factors in order to rank MMA understands the reality of matchmaking in the fight business. But the Contender Generator could be useful in the continual matchmaking and contendership conversation. This first edition focuses on the UFC's flyweight division. The reasons for that will be explained in the followup paragraph below, along with a few more particulars.
The reason for choosing to exclude non-UFC fighters was quite simply based on my limited spare time and the availability of data. The UFC (thanks to FightMetric and others) keeps close tabs on a lot of data from every fight. While the data isn't perfect it's more than anyone else has. The reason for selecting the flyweights is similar. I wanted to see if I could develop formulas that resulted in reasonable rankings without spending months in the attempt. There are a relatively small number of flyweights in the UFC, meaning fewer fighters to track. I was initially worried about the relatively short division history, but the flyweights are fairly active and with weighted starting scores based on prior experience gave solid evidence that the system works as intended. This system was developed with the idea of expanding it to include all UFC divisions, given the time to input the fight history. Doing so would provide an additional tool for the MMA community in the always moving matchmaking conversation. I would be very happy to discuss the idea with any interested sites.
7 Timothy Elliott
10 Chris Cariaso