Country Breakfast - Matt Hughes at one point was 42-4 what category does he fall in?
Like Wanderlei, he is an all-time great whose Elo never exceeded 2000 because of the all-around lower average Elos of earlier eras.
Country Breakfast - Matt Hughes at one point was 42-4 what category does he fall in?
Like Wanderlei, he is an all-time great whose Elo never exceeded 2000 because of the all-around lower average Elos of earlier eras.
Anek -triforce -Anek -triforce -How the fuck was Anderson Silva a "international master" and not a GM?
He was a single solid victory away from Grandmaster status when he lost to Weidman.
He was on a 17 fight win streak against the highest rated middleweights in the world and undefeated in the UFC.
Iām sorry but you fucked up the elo calculations.
Rafael Dos Santos is a Grandmaster but ole Anderson Silva in his prime is stuck at an IM
And if youāre going by chess standards, itās not just a simple rating. You need to get "GM norms" meaning you have to beat at least 1 GM in 3 seperate tournaments, not just attaining "2500".
The calculations are being done by computer, and the same program thatās calculating Silvaās rating is calculating RDAās, so itās hard to imagine it systematically over-rating RDA while under-rating Silva. Itās not like Elo is complicated. I have it on GitHub if you want to take a look.
Iām aware of the norms requirement in chess, but achieving 2500 in MMA is clearly hard enough by itself without adding further requirements.
-Did you start from UFC 1 where each fighter was ranked at a provisional 1200?
-How do you rank fighters that come from outside the UFC? If Ben Askren entered the UFC today would he be a 1200? Wouldnāt a win over him be pretty significant in terms of ratings, not just getting a 1200ā¦?
-Do fighters have provisional rankings (like chess?).
-What about fighters that go up and down weightclasses? In 2005 did BJ Penn have his same elo rating when he went up 3 weightclasses to fight Machida? How do you adjust for this? Itās not like he would be a1200 with all of his experience/skillā¦ and itās not like he would be rated as high as he would at his usual weight class because of the signficant size difference.
All Iām saying is that thereās a lot of subjectivity to the way this is done.
Anek -Osbot -You can't apply this system to MMA in this manner.
You're basically rating people at their peak inperpetuum. Anderson Silva that lost to Weidman, wasn't peak Silva.
The Jose Aldo that lost ot Holloway, wasn't the Aldo of 5 years ago.
According to your system, you could have some random shitlord fight Silva and Weidman when they are in their 50's and become a grand master or whatever the fuck.
Ratings decline as soon as the fighter begins to lose. A random shitlord who beat Silva and Weidman in their fifties wouldn't get much of a ratings boost because Silva and Weidman would themselves have low ratings in their fifties.
But what if Jon Jones retired today and decided to make a comeback when he's 50?
triforce -Anek -triforce -Anek -triforce -How the fuck was Anderson Silva a "international master" and not a GM?
He was a single solid victory away from Grandmaster status when he lost to Weidman.
He was on a 17 fight win streak against the highest rated middleweights in the world and undefeated in the UFC.
I'm sorry but you fucked up the elo calculations. Rafael Dos Santos is a Grandmaster but ole Anderson Silva in his prime is stuck at an IM And if you're going by chess standards, it's not just a simple rating. You need to get "GM norms" meaning you have to beat at least 1 GM in 3 seperate tournaments, not just attaining "2500". </p> </blockquote> The calculations are being done by computer, and the same program that's calculating Silva's rating is calculating RDA's, so it's hard to imagine it systematically over-rating RDA while under-rating Silva. It's not like Elo is complicated. I have it on GitHub if you want to take a look. I'm aware of the norms requirement in chess, but achieving 2500 in MMA is clearly hard enough by itself without adding further requirements.</blockquote> <p> -Did you start from UFC 1 where each fighter was ranked at a provisional 1200? -How do you rank fighters that come from outside the UFC? If Ben Askren entered the UFC today would he be a 1200? Wouldn't a win over him be pretty significant in terms of ratings, not just getting a 1200...? -Do fighters have provisional rankings (like chess?). -What about fighters that go up and down weightclasses? In 2005 did BJ Penn have his same elo rating when he went up 3 weightclasses to fight Machida? How do you adjust for this? It's not like he would be a1200 with all of his experience/skill... and it's not like he would be rated as high as he would at his usual weight class because of the signficant size difference. All I'm saying is that there's a lot of subjectivity to the way this is done. </p>
The program works by giving provisional ratings to players who don't have a calculated rating and using them for an approximate rating. However, after using the program to calculate enough ratings, you will have ratings for every opponent the fighter you're rating has fought, and each of those will have ratings for every opponent they've fought, and so on, until the only provisional ratings left are for debut fighters.
I have a function called "deep_calc" that rates fighters three levels down. I've used it enough that by now,, I probably do have all the fighters from UFC 1 rated correctly. That's how I accumulated a database of over 13,000 fighters.
The program rates fighters from all orgs. Askren himself is listed as a "Candidate Master".
I don't do anything about weightclasses, so that's definitely a weakness, but to be fair, Fight Matrix works the same way, and seems to work pretty well.
triforce -Anek -Osbot -You can't apply this system to MMA in this manner.
You're basically rating people at their peak inperpetuum. Anderson Silva that lost to Weidman, wasn't peak Silva.
The Jose Aldo that lost ot Holloway, wasn't the Aldo of 5 years ago.
According to your system, you could have some random shitlord fight Silva and Weidman when they are in their 50's and become a grand master or whatever the fuck.
Ratings decline as soon as the fighter begins to lose. A random shitlord who beat Silva and Weidman in their fifties wouldn't get much of a ratings boost because Silva and Weidman would themselves have low ratings in their fifties.
But what if Jon Jones retired today and decided to make a comeback when he's 50?
There is a rating system designed to handle this called the Glicko rating system. It works by having both a rating and a number representing the uncertainty of the rating. During periods of inactivity, the uncertainty of the rating increases. If a fighter with a low rating beats a fighter who has a very high rating--an unexpected result--but the rating uncertainty is very high, he won't get much credit, because the uncertainty of the rating and the unexpectedness of the result are taken as evidence that the high rating is no longer legit.
I did try to implement Glicko for MMA to deal with ring rust and retired fighters making comebacks, but it doesn't work very well, because it requires a high number of matches (15-20 per rating period), which is great for chess but not for MMA.
Cool thread OP. Is there a ranking of the grandmasters? GSP has the highest rating ever in MMA but where do the others stack up?
Jeon Sa -Cool thread OP. Is there a ranking of the grandmasters? GSP has the highest rating ever in MMA but where do the others stack up?
Thanks.
If it wasn't clear, the graphs in the second post represent the Grandmasters' Elo ratings over time.
The max ratings achieved are:
Chris Weidman:
2514
Max Holloway:
2551
Stipe Miocic:
2501
Daniel Cormier:
2514
Tyron Woodley:
2545:
Rafael Dos Anjos:
2529
Jon Jones:
2556
Georges St. Pierre:
2604
I've been looking into the Elo inflation problem, and it's apparently because fighters with bad records are more likely to quit early in their careers, so the average rating increases over time by about ten points per year.
When I finish modifying the program to give inflation-adjusted ratings, we will be able to compare fighters between eras.
Anyone who wants to use or improve it can find the program here:
<a href="https://github.com/austinlorenz/MMA_Elo_Ratings">https://github.com/austinlorenz/MMA_Elo_Ratings</a></p>