JCS83MD wrote:Well its a performance rating, not a traditional ranking. So there's no reason why Ouma can't be ahead of Karmazin.
Except for the fact that when they met, Karmazin beat Ouma. I understand that these rankings try to predict who will beat whom, and I fully recognize that Ouma would likely beat Karmazin if they fight again. That being said, the fact remains that Karmazin beat Ouma in a head-to-head matchup. It's hard to justify ranking Ouma ahead of him, even if Ouma has been more active.
I agree, theres some issues.. but the stats don't lie.
Of course they do. Remember the Twain quote -- "lies, damn lies, and statistics." As was pointed out elsewhere, it's hard to have rankings of this sort because so few fighters in the top 25 fight each other.
Take the heavyweight rankings. These rankings say that Valuev would beat every heavyweight out there. While it's true that in the next year or so Valuev will beat every heavyweight he fights, that's not the same as saying he'd beat every heavyweight. He simply won't fight the heavyweights whom his management knows will beat him. I think Rahman would beat him, for instance, but they are unlikely to meet.
Tye Fields is the same thing. Sure, he'll beat the guys he fights, but he won't be fighting good guys. So these rankings will reward him for that, but we all know that if he fought someone like Rahman or even Chris Byrd, he'd likely lose. But since that won't happen, your rankings will still be accurate in the technical sense, but they won't reflect reality.
All rankings are subjective. To try and get it down to a science is futile since there will always be the human factor at play. For example, since these rankings are trying to show who would beat whom, for them to be accurate in 1990 they would have needed to show that Buster Douglas was ranked ahead of Mike Tyson. Any computer program that would have produced a ranking like that before their Tokyo fight would have been seriously flawed. But Douglas did beat Tyson and no computer rankings could have predicted that.
We optimize for overall accuracy, no matter if its a fight involving Top 10 guys, or an ESPN2 main event.
I know, but I disagree with how you define "accuracy."
This last update was pretty big.
Indeed it was, and it rectified some of the problems I noticed before.
Look, I'm not trying to be a pain in the ass on this. I know it's hard work to do all the rejiggering of the ratings criteria. The last thing you probably want is to have some guy like me sitting at my computer pointing out what I perceive as flaws. I want to see the best ratings possible, though, and if they look strange from my perspective, I'm going to let you know. I do appreciate the work you do, though.
In the end, I agree with an editorial I read in a 1970's copy of Ring magazine. Some guy had approached the editor Nat Loubet and said that he had a computer program for ranking boxers. Loubet explained why this wouldn't be a good idea, since all the variables that go into ranking fighters would be difficult to quantify in a computer program. That's true in my mind. Sometimes you just know Fighter A is better than Fighter B for some undefinable reason, and that's that. No computer program can do this.
