We detailed our intentions to add some alternate rating systems for comparison in a previous post.
We got several comments and questions about this, with two common questions being:
1 – I thought the system in use was a version of Elo?
2 – What’s the point of adding alternate rating systems?
Addressing #1, the current rating system does share a few things with Elo, most notably, the core formula that manages the movement. However, it has also borrowed elements from the Glicko rating system and has tons and tons of additional code in place to make it better equipped for MMA, which is still a very disorganized and unpredictable sport, in terms of not only the matches but the matchmaking as well. Systems like Elo and Glicko were made for chess, a sport (game?) where competitors can play dozens of matches in a month where none of them get kicked in the head by their opponent.
So, if Elo wasn’t made for chess, what’s the point of including it? Good question. The standard system in place here has reached maturity. Other than maintenance and bug fixes, it is what it is. Furthermore, it is far beyond mathematics, it is essentially a software program due to the complexity. Elo is mathematically pure and that’s what makes it so interesting. It is also a new challenge — trying to tweak the available parameters while still keeping the formulas that govern the movement as pure as possible… plus it is something new to do and a new addition to the site. The standard system is driven mainly by subjectivity, while using objectivity to help tweak new parameters so they aren’t simply arbitrary updates. The Elo systems we add will be solely driven by objectivity.
I thought I’d post some updates as well. I’m working on two versions of Elo:
“Standard” Elo
The standard version looks at the overall prediction rate for all fights in the database.
“Modified” Elo
The “modified” version (which I may rename), looks at the overall prediction rate for all fights in the database, unless the fight results in a DQ, 1-round technical draw or includes a fighter with 2 or less prior pro fights.
An Elo disclaimer, at least for the time being, is you need to realize that fighters’ reached their ratings based on their fight history and when comparing two fighters, their histories need to be similar. In other words, it is realistic to compare TJ Dillashaw’s rating to Cody Garbrandt’s , because they are on “the same level” in terms of caliber, fight in the same era, same country in the same division. It is not realistic to compare Demetrious Johnson’s rating to Bas Rutten’s, for obvious reasons.
Standard Elo – Top 10 – Active Men / Women
2446 Georges St. Pierre 2402 Daniel Cormier 2367 Jon Jones 2342 Tyron Woodley 2329 Robert Whittaker 2319 Max Holloway 2256 Tony Ferguson 2254 Khabib Nurmagomedov 2239 Ryan Bader 2236 Gegard Mousasi
1906 Amanda Nunes 1881 Cristiane Justino 1833 Rose Namajunas 1772 Jessica Andrade 1731 Joanna Jedrzejczyk 1705 Claudia Gadelha 1674 Karolina Kowalkiewicz 1671 Katlyn Chookagian 1664 Valentina Shevchenko 1659 Tatiana Suarez
Modified Elo – Top 10 – Active Men / Women
2357 Daniel Cormier 2353 Jon Jones 2343 Georges St. Pierre 2280 Max Holloway 2242 Robert Whittaker 2242 Tyron Woodley 2230 Gegard Mousasi 2215 Khabib Nurmagomedov 2194 Tony Ferguson 2178 Stipe Miocic
1887 Cristiane Justino 1854 Amanda Nunes 1826 Rose Namajunas 1762 Jessica Andrade 1735 Joanna Jedrzejczyk 1697 Valentina Shevchenko 1683 Livia Renata Souza 1677 Tatiana Suarez 1661 Claudia Gadelha 1658 Katlyn Chookagian
Δ
This site uses Akismet to reduce spam. Learn how your comment data is processed.
Support Fightmatrix.com and reach thousands of MMA fans by advertising with us! Click for more details.
Betsquare for Aussies