By Chamatkar Sandhu
MMA fans might be familiar with FightMetric, the world’s only comprehensive mixed martial arts statistics and analysis system. They provide the official statistics to the UFC which pop up on your screen every time you watch an event whether its strikes landed, takedown attempts, percentage of leg kicks thrown etc. Reed Kuhn earned a research fellowship with FightMetric a few years ago and whilst working there began personalized consultations for professional MMA fighters, including several UFC champions. Reed is also the creator and author of the “Fight Science” feature in FIGHT! Magazine, and has been a writer and researcher for UFC produced televised content. At Sherdog he provides the most comprehensive statistical breakdowns available anywhere for main events on numbered UFC events. At CagePotato he publishes graphical “data bombs” illustrating interesting statistics related to UFC fighter performance or macro fight trends. He also analyzes the implications of betting odds on mixed martial arts for both macro trends, as well as individual matchups at MMA Oddsbreaker, and can be can be heard frequently as a guest on Alchemist Radio. And before and during most UFC events he releases key statistical analysis related to fighter performance and matchups through his “Uber Tale of the Tape” on Twitter
Toward the end of 2013 he released “Fightnomics: The Hidden Numbers and Science of Mixed Martial Arts” his debut book which was a year in the making. Last week I was able to catch up with Reed to discuss his book along with his job and passion for the numbers behind MMA.
What’s your earliest memory of MMA and when did you become a fan?
I was in high school when the UFC launched on pay-per-view, and as a childhood martial arts and action movie fan the draw of a “no holds barred” fight contest was hugely compelling. Those days were the pinnacle of the “who would win, Bruce Lee or Mike Tyson?” questions. My friends got together and ordered one of the early shows, but I honestly can’t remember which one. Later in college, my fraternity brothers chipped during a vacation and we ordered a PPV UFC event and again, I really can’t even remember who fought. I do know that it changed how we viewed fighting pretty much immediately. It wasn’t until I was in business school in 2005 that I reconnected with MMA through the Ultimate Fighter series, and the UFC’s marketing presence became continual. From there on out I was hooked.
How much MMA / UFC do you watch or keep up with?
In the last year since leaving my career in management consulting it’s been a lot. I watch pretty much every televised event now including most prelim fights: all UFC, Bellator, and WSOF cards, and occasional events from BAMMA, Legacy, etc. On top of events, there’s also weekly UFC peripheral shows like UFC Tonight, Ultimate Insider, and of course the Ultimate Fighter series – so it adds up! My wife has been finishing medical school, so fortunately she’s mostly too busy to notice.
What is your background in statistics?
I dabbled in math in college while majoring in Physics, then completed a master’s degree in Systems Engineering while working as a scientific consultant. That second degree was critical, it was basically a program geared towards solving complex problems quantitatively. It included more advanced statistics, risk analysis, data mining, modelling and simulation, etc. on top of the foundation of how to think about structuring problems, and I think that’s the material that I really connected with intellectually. I looked at any given problem or system, at the world really, differently after those courses. When I went to business school afterwards I took their available classes in decision sciences but it was just the icing on the cake at that point. Then I immediately went into management consulting where quantitative analysis was how you communicated your work, analytics is the common language in business and this is where I got the most practice.
When did you first have the idea to write the book?
Once I became a “research fellow” for FightMetric back in fall of 2009, the very first thing I wanted to do was write a book. Immediately after seeing the richness of data available, I had a so many questions that I knew could finally be answered with a little bit of analysis, and I believed the other MMA fans like me would want to know the answers too. But as I started doing the research I was also getting to know to people in the MMA industry, and the opportunity to consult to a UFC fighter popped up. One thing led to another, and my nights and weekends in 2010 and 2011 were spent consulting to fighters and getting more familiar with the data and the possible ways to analyze it. By 2012 I was trying to focus more on research and writing and I was doing more freelance articles for FIGHT! Magazine and online media like CagePotato and Sherdog. I knew there was plenty of material to fill at least one book, I just didn’t have the time to dedicate to it on top of a more than full-time job plus all these hobbies. In January of 2013 I felt like it was “now or never” and I quit my job with the goal of having a book out by the end of the year. I just barely made it.
How has the book been received so far?
So far it’s been overwhelmingly positive. I heard great things from fans, managers, trainers, and even fighters. There’s so much material all in one place in this book that there’s something for everyone to learn, even people who have been in the sport a long time. It’s also entirely new and unique material. No one has ever viewed MMA this way, and there’s insight that’s been hidden in every tiny corner until now. The book covered a lot of ground in one fell swoop, and I hope it’s viewed as an important marker for our improved understanding of MMA.
What would you say has been the most interesting fight stat you have ever researched?
So many, really. Using historical betting lines to re-examine the Tale of the Tape and realizing that the market has been misreading matchups (or at least mispricing and betting incorrectly) was fascinating. And that first had to start with a definitive quantitative assessment of all basic Tale of the Tape advantages, which is hugely important to understanding the sport, and what really matters in a fight. I’m partly amazing it hadn’t ever been done before, but I’m also excited to be the one who did it, and I think we did the analysis justice.
Going deeper I found interesting trends where fighters are making poor tactical decisions based on what historically works in fights. So if ever someone tries to say that stats can’t add to the MMA I’ll point to guys who shoot for too many takedowns, or give up mount for low probability submissions, or just don’t realize what judges are looking for when they score a round. Every subject, industry, business and sport (even “cage fighting”) has benefited from the direct application of analytics and the scientific method. Every insight backed up with evidence takes us from believing something, to knowing it. Occasionally the analysis allows us to let go of a long-held incorrect belief. Good coaches know this, and that’s why analytics are a competitive advantage in sports.
What has been the most difficult stat for you to research?
Probably “Time in Position” or TIP data. I outline all the possible positions fighters can get into during a fight, at least a high level, and every fraction of a second in the UFC gets tagged for one of those categories. But as with the discrete striking data, there’s no sequence to it inside of a round, just a total. So trying to understand, for example, whether a fighter “stole a round” with a takedown right before the bell can’t be done purely from spreadsheets, since I won’t be able to determine when the takedown occurred.
Other difficulties are more macro, but will improve over time. Sample size for many fighters is quite small, so when it comes to looking at matchups or individual performance data I can only focus on more experienced guys. But benchmarking the sport – and divisions within it – is easier because there are thousands of fights. The reporting of basic information like height and reach is also historically volatile. I think they’ve improved this recently, and when it comes to historical macro analysis I can assume that errors either way even out in the long-run. After all, the UFC has no reason to misreport and they do the measurements themselves. But I’ll be interested to re-run analysis later on and see if the patterns get tighter as data accuracy improves.
Are there any stats which you would like to compile but just don’t have the technology or the availability to do so?
The data architecture is quite robust, but it does have its limitations. Even with 70 discrete metrics per fighter per round (excluding TIP data, fighter biographical and anthropometric data, and additional fight-level descriptors like card location and fight order), I still can’t delineate between a kick, punch, elbow or knee. The data also doesn’t track left and right for striking. And all discrete metrics are aggregated for an entire round, so the simple question of “what are his favourite combinations?” can’t really be answered in detail. And yet, there’s so much that can be answered. These are very subtle weaknesses of the data, despite the overwhelming usefulness of what we do have. In the long run, I do think that sensors will eventually be utilized during fights, and the amount of insight we can get from real-time sensors in the gloves would be amazing. Imagine knowing not only how many punches really landed during a round, but how hard they landed would be awesome for analysis, for fans, and certainly for judges. Just like knowing how fast a pitch in baseball was immediately after it hits a catcher’s glove, we’ll know exactly how hard a punch was that wobbled a fighter. The angles we can take analyzing this data is mindboggling.
Are you a gambling man? If so how often do you place bets and do you usually win?
I’ve gotten more drawn into that aspect of the sport since getting involved with MMA Oddsbreaker. The creator of that site is Nick Kalikas, the guy who literally sets the betting odds that are used in most places for MMA, who also offers betting recommendations and a premium betting picks service that has been shockingly profitable. From my perspective, any analyst that does data mining in a competitive system will eventually want to test their predictive capability against the market. That hasn’t been the focus of my research yet, though I did learn a lot from analyzing historical betting lines.
I recently went to UFC 168 and had the opportunity to bet on the fights, I decided to gamble enough to pay for the trip as a test. Fortunately, I hit every single one of my bets and won enough to cover the expensive flights and hotel – and then some. I don’t live near a sports book, but it’s good to know that my numbers aren’t lying to me!
Given the stats you have gathered what would you say is the best discipline in MMA for an amateur to master first? i.e. what discipline tends to give you the most advantages in a fight based on your stats?
Unfortunately there’s nothing in the data that describes styles of fighters. I’ve taken a few steps down that path to try to manually create a new listing of fighters by primary/secondary/tertiary/etc. styles, but so quickly ran into problems that it fell to the bottom of my priority list. I’d love to do a piece on “classic striker-grappler” matchups for example, but I can’t because there’s nothing in the biographical data that says a guy is one and not the other. Guys also reinvent themselves and change their styles with time, so there are fewer and fewer fighters who could even be described as just a striker or a grappler. Maybe one day more biographical data will be collected and we can create that list, then lots of interesting analysis could be run. I’d really like to test the ideas of how the sport evolved over the years. The dawn of MMA was a grappler’s world. Then the wrestlers and ground and pounders came in. Eventually the sprawl and brawl fuelled a resurgence of the strikers. Of course now you can’t really compete at a high level in MMA without being cross-trained.
What was the most shocking stat you came across?
I thought this one was pretty mind blowing: the Southpaw Advantage is so real, it actually works against other Southpaws. When I first analyzed the data years ago I was actually skeptical of any advantage at all. I didn’t think that at the highest levels of competition something as simple as striking stance would have any effect on outcomes. Yet it does, so I went deeper. That’s why we do science in the first place – to be proved wrong and get smarter in doing so. Personally, I think this particular phenomenon should be renamed the “Contra-Southpaw Disadvantage.” There’s nothing about being left-handed that makes you a better athlete or better fighter, but when opponents face a Southpaw stance they all seem to perform worse, which creates the effect we see and drives a meaningful trend. My version of the name just doesn’t roll off the tongue as well despite being more descriptive and accurate. But now we know the effect is real, and we know why.
What was the best part / worst part about writing the book and the whole process?
The research was really an awesome experience. There were nights that I lay away and just couldn’t wait to get up and start cranking out analysis. I had so many ideas in my head I wanted to test, and I knew every bit of it was totally new and would improve our understanding of an often misunderstood sport.
The hardest part was just getting the book out. There’s a hundred little things to worry about and consider, especially since we were going to sell print copies on Amazon. We hit a snag with the printers during the holidays and the books didn’t hit the market until after the big book buying season. That was really frustrating. But now that it’s out in the wild I’m just focused on letting everyone how much valuable information there in there.
Do you have ideas already bubbling for another book in the future?
Perhaps not for MMA, although I might release a “Quantitative Guide to MMA” that focuses on the stats and charts minus all the commentary. That’s a small market to serve though. I’ve had a few other ideas for taking a scientific approach to other popular subject matters, and those would be fun and informative to research as a science writer. But right now I want to carry MMA analytics forward to the larger audiences and make sure everyone is fully understanding and appreciating the fight game. Stats can help us enjoy watching competition, and may also improve the public’s perception of the sport itself as a complex game of human chess, rather than a simple barbaric fight. It’s also thrilling to get in on the ground floor of something you can be passionate about. I now know how much I enjoyed writing the book, so it’s definitely something that could happen again if I find the right project.
If anyone wants to get into the line of work you do what is your best advice for them?
Having a combined science and analytics background is a must, but it doesn’t require expensive degrees. There are so many educational resources available now that anyone can start from zero and get into analytics and systems engineering at their own pace. These are extremely valuable skills in today’s market, and it’s only going to become more valuable during the Information Age. From a professional standpoint, management consultants, investment bankers, and similar business careers have to employ hypothesis driven analytics to prove their points and support complex decision making. The exact same skills that allowed me to research and write the book were honed on the job, not in a classroom. Hopefully there are still people out there interested in learning science, math, and analysis – because the market is going to need a lot of them! If readers of “Fightnomics” enjoy my approach to discovery and problem solving and pursue learning how to do it, chances are good that they will fare well in the job market.
“Fightnomics: The Hidden Numbers and Science of Mixed Martial Arts” is available now on Amazon.
Follow me on Twitter @ChamatkarSandhu
