As somebody who enjoys statistical analysis in sports, there are a number of sports websites and blogs I read that have a statistical bent to them. One of these is Bill James Online, the website operated by Bill James himself, the godfather of advanced statistical analysis in baseball. Most of his website can only be accessed with a subscription, but he does make freely available a feature called “Hey Bill,” in which readers ask James random questions about baseball, and James answers them.
One particular question, which is no longer posted on the site, but was there a few weeks ago, asked James about what would happen with baseball if games were played once a week instead of nearly every day. This would obviously cause a lot of dramatic, fundamental changes, but one in particular that James mentioned would happen caught my attention. James said that the value of scouts and scouting would rise in relation to the value of statistics and statisticians, because with dramatically fewer games, there would be much smaller sample sizes to deal with. Take Albert Pujols for example. Pujols currently has a .213 batting average, .248 on-base percentage, and according to Fangraphs, a .240 wOBA (which is terrible) and -0.7 wins above replacement.
At the same time, Pujols has had only 157 plate appearances this season, which means there’s a ton of potential variance with his season’s performance thus far. But what if the 37 games Pujols has played represent an entire season of baseball? Is this really how good Pujols is now, or has he merely been the victim of bad luck or bad variance?
In such a situation, the importance of scouting would become paramount. While in the real world, Pujols will have many more chances to turn his season around, in this imaginary world, scouts would be needed to watch tape of Pujols, break down his stance, swing, and approach, and come to a conclusion about how much his skills have diminished. When we can’t be confident in what statistics are telling us, the best information we can get is what scouts can provide.
This is how I feel about MMA right now. The constant obstacle with statistical analysis in MMA is the small sample sizes people like me have to work with. Think the sample of 157 Pujols plate appearances is small? How about the four takedown attempts of Junior dos Santos? Or the six fights of Matt Mitrione? Heck, what about even the 103 takedown attempts of Georges St-Pierre?
This is why I will never regard SILVA as the be-all or end-all of anything. SILVA estimates how good a fighter is based on his record. From what I’ve seen, SILVA is the best metric at estimating how good fighters are based on this objective data only. It doesn’t mean SILVA can’t be beaten, it just means I think SILVA is the best objective rating system right now. But SILVA is based on a fighter’s last ten fights. Just like everything else in MMA, SILVA has to rely on a small sample of information. Sure, SILVA provides a pretty good estimate most of the time. It has Junior dos Santos as the #1 heavyweight in the world, and Oli Thompson at the bottom of fighters who qualify to be listed right now. On the flip side, it will occasionally do something like rate Mike Russow at #4 at heavyweight.
If you consider SILVA’s objective, its rating of Russow as the #4 heavyweight in the world is defensible. Russow is 10-0 in his last ten fights, which is exceptionally rare at heavyweight, and it includes a few fights in the UFC. Based on statistical data and objective analysis, Russow is an elite heavyweight.
Which is why scouting is so important. I could watch Russow’s fight against Todd Duffee and watch some awful striking on the part of Russow, being treated as a punching bag until landing a KO punch out of nowhere. I could watch Russow’s fight against John Olav Einemo and see Russow grind out a decision against a very low-level UFC opponent. Then I could watch Cain Velasquez demolish opponent after opponent. Velasquez is rated as the #5 heavyweight in the world – one spot below Russow. Bring scouting into the equation, and it becomes abundantly clear that Velasquez is the much better fighter.
I say all this to simply say that while I love statistics and statistical analysis, scouting and tape study have to be part of my analysis and breakdowns of fights. This isn’t to say that SILVA is as good as it’s ever going to be, and I still consider SILVA, Fight Metric, and Fight Matrix to be very useful tools. But as far as analysis of MMA is concerned, that’s all those resources are – tools to be used as part of a larger toolbox.