Fantasy Fights

Intelligent, unique MMA analysis

ELO’s Accuracy at Predicting UFC Fights

As you may or may not know, I’m working on the development of a statistical model that will attempt to predict the winners of MMA fights using a variety of different statistics and methods. This will include ELO ratings, SILVA scores, and Fight Metric data, and may include other variables as well (in particular, I’m considering adding years of experience as a variable).

Of the components I’m planning to throw into the model, the only one that I have complete data for is ELO ratings. I dumped these ratings into SPSS and told it to build me a model of how well it could be used to predict UFC fights.

As it turns out, ELO ratings correctly predict UFC fights about 60.5% of the time. This success rate is comparable to how well SILVA scores did when I was running those. It does not, however, beat the betting odds, or most MMA experts. My goal for this model is to be able to predict UFC fights correctly at least 65% of the time, and ideally closer to 70%.

The idea is for the model to end up with the following components:

-ELO ratings: a representation of how well a fighter is doing in his recent career
-Early-career SILVA scores: a prior “best guess” for how successful a fighter will be in his career
-Fight Metric data: rating how a fighter performs within the actual fights; going beyond simple win/loss/draw output
-(Possibly) duration of MMA career: indicating whether a fighter is more likely to improve or decline in future fights

Hopefully with a well-rounded data set, I’ll have a model that is successful at predicting fights. Right now, I only have one of the three or four elements, and getting the others will just take time. When I have new data to present, I’ll let you know.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: