In 1977, Bill James published the first of his Baseball Abstracts. The newsletter sparked the beginning of an objective revolution in baseball that continues to this day thirty years later.
MMA, up until now, has lacked any resemblance of statistics outside of win/loss, method, and round. That all changed when FightMetric announced that they have scored every fight of the "modern UFC era." Rami Genauer of FightMetric was kind enough to talk to Bloody Elbow about the project, quantitative analysis in MMA, and what it all means going forward.
Bloody Elbow: I guess we should start with an introduction to FightMetric for the uninitiated. What is it? What is it's purpose? How does it work?
Rami Genauer: FightMetric is really just the name of the company. We're the world's first purveyor of comprehensive MMA statistics and analysis. The basis for those stats and analysis is a carefully-crafted statistical system based on thorough research into the way MMA fights really happen.
Bloody Elbow: One of the things that's been a mystery to me is how you originally came up with your system. Can you describe how you got the idea and the process behind it's origins?
Rami Genauer: The origin was really born of the lack of anything previously. I'm kind of a stat head for other sports, specifically baseball. When I got into MMA it struck me as odd that the sport had absolutely no method of quantifying performance other than win/loss, method, and time. So given that absence, I took it as a thought exercise to see if I could think up a statistical system from scratch. It's a daunting challenge, because you have nothing to base your ideas on, but on the other hand, it's a huge opportunity, because you can literally do anything.
The process for the system involved a favorite technique of mine - root cause analysis. The purpose of a root cause analysis is to try and plumb the depths of the thing you're investigating and see if you can understand it in its very essence. That's kind of what we did for MMA, try to boil it down to it's most basic parts and see which ones were likely to be statistically significant. Then we did the research to test those hypotheses and made some changes. After that, we had a pretty robust system that has really only undergone two changes in the 18 months or so that it's been in use.
Bloody Elbow: I'm sure a lot of that data entry is a result of the project you've been working on scoring every UFC fight. What can you tell us about that?
Rami Genauer: We've just recently completed our project to compile complete statistics for every fight in the modern era of the UFC. We call that UFC 28 and onward, because UFC 28 was the first event to use the current Unified Rules. So basically, all fights from UFC 28 until today have been fought on a level playing field. That means that the stats for all those fights are comparable and we can easily make apples-to-apples comparisons.
Now, we specifically designed the system so that it is completely adaptable to all different rule sets, but it's a little more difficult to make exact comparisons when two fights were fought with different rules.
More from Rami in the full entry.
Bloody Elbow: So, what does this mean for fans and sabermetric types? Will we have access to career stats a la baseball-reference.com?
Rami Genauer: That's the goal. The particulars of how that information might be delivered are still being considered, but yes, the goal is to open the entire statistical database to all MMA fans.
The balance on our part is that FightMetric is still a business with a responsibility to those invested in it. So we're going about this in a way where we can make this a viable business while still recognizing the value that this information has to the MMA fanbase as a whole.
Bloody Elbow: If and when the data is opened to the public, what kind of information can we expect to see in terms of raw stats?
Rami Genauer: I guess that depends on the format in which the data is presented. The possibilities are vast.
We track 67 statistical categories for every fight, in addition to a bunch of supplemental information, like judges' scoring, fighter demographics (reach being the hardest to get a definitive answer on), and betting odds. Those 67 statistical categories are raw stats, so they can be combined, cut, summed, divided, or averaged in literally thousands of ways. Of course, not every combination of stats will be terribly meaningful, but there really are enough cuts at the data to keep someone busy for a very long time.
Bloody Elbow: Do you see quantitative analysis becoming more readily accepted by MMA fans than say more established sports like baseball and basketball which have had to contend with not only building superior metrics, but tearing down the more traditional and flawed predecessors?
Rami Genauer: I think that there is a small advantage to not having to break down existing stats, but you also have to contend with the mindset that MMA doesn't have, doesn't need, and shouldn't have stats. I think that can be countered with explanation, exposure, and time, but there really is a bigger challenge.
MMA is very hard to analyze in a simple way because fights have different lengths. If you're thinking about things in terms of volume, of course a guy in a 15-minute fight will land more strikes than someone who finishes his opponent in a minute. So from a statistical volume standpoint, why should you penalize the second guy for doing his job faster?
This problem forces you to do some more creative things to make good comparisons, and the truth is, the more "stuff" you have to do to the data to make it meaningful, the greater the barrier to understanding for people who aren't that into it.
Bloody Elbow: Which I think you've seen in baseball with park and era normalization and the like.
Rami Genauer: Exactly. But speaking of baseball, I like to give the example of imagining that baseball changed its rules so that every park had a sign in the outfield that said, "Hit this sign and win the game." No matter when someone hit the sign or what the score was at the time, that team won the game. If no one hit the sign, then whoever had more runs after nine innings would win.
If that was the case, all the counting stats that we rely on (HR, RBI, SB, etc.) would be meaningless, because you'd need to know how long the games were too. That's exactly the situation that MMA is in. A guy could be losing for 14 minutes and then land the knockout strike - just as Shonie Carter (did to Matt Serra).
Bloody Elbow: In a similar vein, I think another problem you run into is that in stick-and-ball sports, while the ultimate goal of course is to win, you have a very obvious objective - to score points. In MMA, there is no such parallel. You're just trying to end the fight or look more effective in the eyes of three observers.
Rami Genauer: That's true, which is why the very first thing we tried to do was make a hard definition of the goal of MMA and then make a hard definition of "effectiveness." It's very hard to collect metrics without knowing what they're supposed to be measuring toward.
Bloody Elbow: So, what kind of feedback have you gotten from fans and industry types?
Rami Genauer: Very good feedback. I encountered some more skepticism when this thing first started, but that's kind of died down. The nice thing about not being terribly well-known is that I still get good emails from people who are only discovering FightMetric now. I think the industry has been receptive because it feels like a natural progression in the maturation of the sport. Every other major sport has stats, so why shouldn't MMA?
Bloody Elbow: OK, wrapping up. You're finished scoring modern UFC fights. Are you going to move on to another organization, focus on analyzing the new data, or something entirely different?
Rami Genauer: We're always going to keep on scoring more stuff because there's always value in having more data, especially different kinds of data, like from Japan. (It) might be interesting to do some comparisons between UFC data and Pride/Dream/Sengoku data. At the same time, there are a bunch of other projects in mind using the data we've already compiled. That alone would be enough for some, but what can I say? More data is better data.
Bloody Elbow: Last question. Where do you see sabermetric analysis of MMA in 1 year and 5 years in the future? It took the sabermetric revolution over 25 years to get close to fully understanding the game, defensive metrics exluded. Are we still decades away from that type of understanding in MMA?
Rami Genauer: That really depends on what the data shows. There are many different layers of meaning in having data like this. I think we're already seeing people latch-on to the most basic layer, things like takedown success rate, things like that.
But if you're searching for greater meaning, I'll be perfectly honest here, you have to consider the possibility that the data will end up being a bunch of noise. Baseball sabermetricians have (eventually) had success because they were able to make compelling conclusions based on the data. We haven't even started down that path yet in MMA
The game is really still a mystery. That's what makes this so much fun.