I remember when I did previews for UFC 139, the show last November featuring Dan Henderson and Mauricio “Shogun” Rua. As a statistics-minded person, I often look to see what various available statistics have to say about an upcoming fight. For the bantamweight fight between Urijah Faber and Brian Bowles, the stats just didn’t tell the whole story.
Fight Metric stats at the time seemed to suggest that Bowles was the better fighter, and should have been favored to beat Faber. The striking statistics were fairly similar, and the wrestling statistics significantly favored Bowles. On top of this, SILVA picked Bowles to win, suggesting that he should be favored due to a stronger recent fight history.
But the betting public disagreed, making Faber about a -180 favorite to win the fight. And sure enough, when the fight took place, Faber not only won the fight, he beat Bowles in emphatic fashion.
I feel the upcoming flyweight championship match between Joseph Benavidez and Demetrious Johnson represents a similar situation. A look at the Fight Metric matchup preview seems to indicate that Johnson should be the favorite to win. Johnson has advantages in significant strikes landed per minute, striking accuracy, takedowns landed per 15 minutes, takedown accuracy, and takedown defense. Benavidez’s advantages in strikes absorbed per minute and striking defense are relatively slight; his only significant statistical advantage is in submission attempts.
Now, unlike when Brian Bowles fought Urijah Faber, the fighter with an advantage in Fight Metric stats does not also have the advantage in SILVA. SILVA favors Benavidez to beat Johnson, giving the Team Alpha Male fighter a score of 86.30, whereas Johnson’s SILVA score is 72.32.
But if I was on a desert island, and all I had to make a prediction was Fight Metric stats, I would have to pick Johnson to win, since the stats suggest he’s a better wrestler and striker.
And once again, the betting public disagrees, and by a greater margin this time, as Benavidez is about a -300 favorite. That suggests that not only will Benavidez win, he’ll win decisively.
I think there are a few reasons for this. One is that, while Fight Metric’s numbers themselves are objective, the way they’ve chosen to collect them is subjective. Their definition of what is considered a takedown attempt or a significant strike might differ from somebody else’s definitions.
As an example, in Benavidez’s last fight against Yasuhiro Urushitani, there was a sequence of clinching against the fence that led to Benavidez landing a takedown in the first round. This sequence involved Benavidez pulling at Urushitani’s leg a few times. I believe Fight Metric considers each pull as its own takedown attempt, meaning that even though the sequence ended with Benavidez completing a takedown, he’s considered to have “missed” on a few takedown attempts at the same time. An educated observer of MMA might reasonably conclude that the sequence was just one overall attempt; this is ultimately a matter of opinion, not a matter of fact.
The second reason is that, if I’m not talking about a fighter like Chuck Liddell, Matt Hughes, or Georges St-Pierre, when I look at Fight Metric stats, or stats of any kind, I’m dealing with a small sample. Whether it’s a small sample of opponents or a small sample of techniques, there’s a large margin for error in these statistics. Combine a high margin for error with a strong dose of subjectivity, and what you get is a tool that, by itself, is inadequate in terms of predicting fights.
Don’t get me wrong; I love Fight Metric, and think it’s a terrific resource. But I also have grown to consider Fight Metric more of a descriptive tool than anything. It’s fantastic at describing what took place during a fight, and its proprietary Effectiveness Algorithm seems to be a very good means of judging who the better man was in a fight (in my opinion). I just don’t know if it’s good as a predictive tool.
This is also why scouting is so important in MMA. When dealing with a sport with small samples, quality scouting has to take precedence over statistics in terms of analyzing and evaluating fighters. From what I’ve seen, I think Joseph Benavidez is a better wrestler, and a better fighter, than Demetrious Johnson. Tape study is simply mandatory when it comes to breaking these fights down. And that’s why stats don’t tell the whole story with Benavidez and Johnson.