FanPost

Elo Rating and Tracking GSP's Dominance

Ranking fighters within a division has got to be one of the harder things to do in the business. You're expected to come up with a concrete linear list of the best fighters in any given division where, for the most part, fighters do not often compete against each other more than once or ever do to physical or contractual limitations. Yet we still try to do it in order to appease this part of brain that likes to categorize all phenomena in our life into a simple understandable structure. Odds are there isn't one list that 'correctly' captures the nature of fighters status in the division and any attempt to do so will only generate contempt from fans about how poorly your rankings are a reflection of what the division actually is like. Nevertheless, we like to try.

Along with the countless subjective opinions that people have for rankings there exists a number of statistical algorithms that those who attempt to rate fighters can try to employ in order to add some objective data into the mix. The one i'm concerned with and using here is Elo Rating. Elo ratings basically start every fighter out with a certain score, in this case 1000 and when players compete with one another they gain or loss points (when they win or lose respectively) proportional to their opponents score. Elo rating's are used by some MMA media outlets which claim it to be a completely unbiased method which is a load BS but it can be pretty cool for tracking the rise and fall of certain fighters numerically.

Below is a graph of the Elo ratings of the top 7 rated fighters based off of records I scraped from wikipedia using code like this. The legend has in order the highest rated players with their updated score as of yesterday.

R9ka4f0_medium

via i.imgur.com

As you can see GSP is a beast. He has beaten basically every top Welterweight out there and the only two dips in his score come at his losses early in his career. And we can basically can agree that based on record alone GSP is a beast. That being said a lot of different rating algorithms will give you GSP on top so it's not like we should be advocating for Elo just based on that.

The downside of this algorithm is that if two fighters are marked as welterweights and they fight each other then the fight will be counted even if the fight isn't at the 170 division. While this didn't matter to much for the welterweight division doing the same thing for lightweights pulls up some interesting results, namely Jose Aldo being ranked high. Which I actually thought was pretty interesting in trying to asses his chances in a potential super fight with Pettis. Aldo has fought a ton of former lightweights at 145 and has put all of them to shame.

N4qanxm_medium

via i.imgur.com

The rankings themselves are pretty interesting but what I think is really cool is being able to look at other information that can be inferred from the data based on their numerical ratings. For instance check out some of the biggest upsets that occurred at welterweight and lightweight based on the difference between ratings of fighters.

Welterweight Upsets

  1. Matt Serra def. GSP on 2007-04-07 with a rating difference of 373
  2. Charlie Brenneman def. Rick Story on 2011-06-26 with a rating difference of 331
  3. Thomas Denny def. Tony Frykland on 2006-11-17 with a rating difference of 325

Lightweight Upsets

  1. Sregey Golyaev def. Takanori Gomi on 2008-11-01 with a rating difference of 432
  2. Hiroyuki Takaya def. Joachim Hansen on 2010-05-29 with a rating difference of 373
  3. Toby Imada def. Jorge Masvidal on 2009-05-01 with a rating difference of 301

One of the downsides of Elo ratings is that it is biased towards fighters who compete more often. This may or may not be a good thing based on what you consider to be important for ranking a fighter but we can sort of mitigate this downside by looking at the average rating of the fighters whom a single fighter competes against sort of like a strength of schedule score.

Welterweight Strength of Schedule

  1. Johny Hendricks with an average of 1204.846
  2. Martin Kampmann with an average of 1200.583
  3. Rory MacDonald with an average of 1194.182

Note that GSP isn't here. Not because he hasn't fought the best but because his score is so damn high that anyone who fights him will probably have a higher Strength of Schedule than him. Also Martin Kampmann has a strong schedule but he also usually gets wrecked by those fighters.

Lightweight Strength of Schedule

  1. Frankie Edgar with an average of 1201.375
  2. Gray Maynard with an average of 1181.000
  3. Benson Henderson with an average of 1179.706

Benson Henderson both has a stellar schedule and good win record which is why he maintains the top spot even though he has lost twice to the current champ.

Again I want to say that using Elo ratings to make holistic assessments about fighters or make predictions is probably a crappy idea. There are too many biases that go un into forming these ratings for them to to be used in the competitive MMA landscape by themselves. That being said I think that using elo-ratings in conjunction with other information could be useful as it does at least capture some of the nature of a fighters career.

\The FanPosts are solely the subjective opinions of Bloody Elbow readers and do not necessarily reflect the views of Bloody Elbow editors or staff.

X
Log In Sign Up

forgot?
Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join Bloody Elbow

You must be a member of Bloody Elbow to participate.

We have our own Community Guidelines at Bloody Elbow. You should read them.

Join Bloody Elbow

You must be a member of Bloody Elbow to participate.

We have our own Community Guidelines at Bloody Elbow. You should read them.

Spinner.vc97ec6e

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_5349_tracker