Using statistics to predict MMA fights

As you might have seen on the first page of this website we use both quantitative and qualitative analysis. Unlike football, basketball and baseball where you can rely more heavily on statistical models to predict outcomes MMA is the polar opposite. You mostly need to rely on your own eyes and watching tape instead of crunching numbers to spit out the future winner.

Good news for everybody who isn’t good at math, you certainly can be a good capper without using any statistics at all.

But that doesn’t mean statistics is useless, as I said we do a bit of number crunching for time to time when making our predictions. A very obvious stat to look at is age. Fighters age differently so you can’t rely on just looking at the number in a vacuum you also have to see the big picture, is the fighter in question showing his age? Some fighters are done by the time they’re 28 some are still at the top at age 40.

But there are averages and sometimes when a fighter has been away and you don’t have recent data to look at, it’s a good idea then to look at historical data, how have fighters of similar weight and age performed?

When statistics are misapplied

It’s for a less math savvy person to make expensive mistakes relying on statistics. An example would be to draw the conclusion that a person with high % takedowns defended must therefore automatically have good TDD. As you probably should now, not all takedowns, strikes, knockouts whatever you think of – are not the same.

A person that has faced only weak offensive wrestlers will have an inflated TDD stat. A person who has fought weak defensive strikers will have a high number of strikes hit per fight. What do those numbers tell when the data is junk because they’re facing different opponents anyway?

This is why the amateur models you see around the net are crap. It isn’t easy as crunching all the fightmetrics data and then spit out a winner based on those numbers.