Ratings - please read before commenting - Archived

[Philla]

I have a question regarding what happens to the ratings when a fighter changes divisions

The ratings explanation page says:

The ratings are decreased for moving up to higher weight divisions by the square of the reciprocal ratio of the weights limits of the divisions - and they are increased by the same factor for moving down the divisions.

First of all, what happens in the case of the heavyweight division, where there is no weight limit.

Secondly, can someone help me understand this with an example? Say a MW with a rating of 1000 moves up to SMW, what happens to his rating?

160/168 is the ratio from MW to SMW, the reciprocal ratio is 168/160. the square of that is (168/160)^2 which is 1.1025

So if the ratio is decreased by that factor, then the boxer's new rating becomes 1000/1.1025 = 907 pts? Can someone tell me if I understood it right, and if not then what is the right way to calculate it?

Thanks

[computerrank]

I have a question regarding what happens to the ratings when a fighter changes divisions

The ratings explanation page says:

The ratings are decreased for moving up to higher weight divisions by the square of the reciprocal ratio of the weights limits of the divisions - and they are increased by the same factor for moving down the divisions.

First of all, what happens in the case of the heavyweight division, where there is no weight limit.

Secondly, can someone help me understand this with an example? Say a MW with a rating of 1000 moves up to SMW, what happens to his rating?

160/168 is the ratio from MW to SMW, the reciprocal ratio is 168/160. the square of that is (168/160)^2 which is 1.1025

So if the ratio is decreased by that factor, then the boxer's new rating becomes 1000/1.1025 = 907 pts? Can someone tell me if I understood it right, and if not then what is the right way to calculate it?

Thanks

1. a limit of 225 pound is assumed (kind of typical heavyweight weight)

2. correct

[JCS]

Martin.. I'd say the "typical" Heavyweight weight is above 225 these days...

Would doing a year-based ref. weight for Heavies improve prediction/appearance throughout time? In other words...

1960s = 212
1970s = 215
1980s = 220
1990s = 225
2000s = 230

Not exact #s of course.. I would look into it deeper first.

[computerrank]

Martin.. I'd say the "typical" Heavyweight weight is above 225 these days...

Would doing a year-based ref. weight for Heavies improve prediction/appearance throughout time? In other words...

1960s = 212
1970s = 215
1980s = 220
1990s = 225
2000s = 230

Not exact #s of course.. I would look into it deeper first.

Today's median seems to be 240 for top20 and
240,5 for top101 to top120

the parameter was optimized to 225 for best fight result prediction a long time ago ... calculation over all times from pre-1900 to today.

[m1kee50]

For those who really must have a P4P best list, they should just create a parallel ranking based on a vote if the computer is not giving the results they want.

We have one of those as well, Sock runs it very well and the same pudenda whose names appear in here, whinge in that one, and dont even vote in the list.

[JCS]

Martin.. I'd say the "typical" Heavyweight weight is above 225 these days...

Would doing a year-based ref. weight for Heavies improve prediction/appearance throughout time? In other words...

1960s = 212
1970s = 215
1980s = 220
1990s = 225
2000s = 230

Not exact #s of course.. I would look into it deeper first.

Today's median seems to be 240 for top20 and
240,5 for top101 to top120

the parameter was optimized to 225 for best fight result prediction a long time ago ... calculation over all times from pre-1900 to today.

Would make sense to have a different reference weight per time period, no? Or even better, a gradual formula that parallels with that of a sliding average.

[computerrank]

... a time depending value should be considered ...

[JCS]

... a time depending value should be considered ...

I agree

When I was fiddling with a system I used:

180 prior to 1910
185 from 1910 to 1930
gradually increased from 185 to 200 from 1930 to 1970
gradually increased from 200 to 235 from 1970 to 2008

[Asterix]

I've got an idea...

How about a feature for the front page? A result recap of the past 24 hours that listed some of the matches, based off the ratings..

Biggest Upset
Most Noteworthy Matches (Highest cumulative point total)
Biggest Mismatch

Has there been any more thought to this? Just wondering, not meaning to hound you. By the way, I really like the star rating applied to fights. It makes it much easier to skim down the date pages and pick out the big fights.

[JCS]

I've got an idea...

How about a feature for the front page? A result recap of the past 24 hours that listed some of the matches, based off the ratings..

Biggest Upset
Most Noteworthy Matches (Highest cumulative point total)
Biggest Mismatch

Has there been any more thought to this? Just wondering, not meaning to hound you. By the way, I really like the star rating applied to fights. It makes it much easier to skim down the date pages and pick out the big fights.

No idea. JohnShep is the doer and has yet to give me any real power. I just try to be a good "idea man".

[cow]

Manny Pacquiao is a little overrated.. Yes,he may be one of the greatest fighter in this time ,but ,still not good enough to be the number 1 lb 4 lb.. Team pacquiao always pick fighters with low knockout percentage.. The last time the team picked a knockout fighter was when they picked marquez which ended into a controversial win on pacquiao. but, the brilliant minds behind pacman's corner always finds a way to resurrect the filipino champion by again finding yet again a fighter with no knock out skills. Manny's last 3 fights should be against dela hoya,campbell and hatton.. by then he will earn the title as the number 1 lb 4 lb king.

[Philla]

I have a suggestion which has probably been previously made:

For active boxers, there should be a time frame after which points from fights are progressively reduced in ponderance, until they eventually no longer affect a boxer's rating.

I know it completely restructures the ratings, but it gives a more accurate description of who are the best fighters today.

thoughts?

[JCS]

I have a suggestion which has probably been previously made:

For active boxers, there should be a time frame after which points from fights are progressively reduced in ponderance, until they eventually no longer affect a boxer's rating.

I know it completely restructures the ratings, but it gives a more accurate description of who are the best fighters today.

thoughts?

That's pretty much what happens now...

[Philla]

I have a suggestion which has probably been previously made:

For active boxers, there should be a time frame after which points from fights are progressively reduced in ponderance, until they eventually no longer affect a boxer's rating.

I know it completely restructures the ratings, but it gives a more accurate description of who are the best fighters today.

thoughts?

That's pretty much what happens now...

well tbh i haven't read the entire explanation page from A to Z but i was surprised when i saw ODLH's ranking at 154, that is what prompted my comment

[JCS]

I have a suggestion which has probably been previously made:

For active boxers, there should be a time frame after which points from fights are progressively reduced in ponderance, until they eventually no longer affect a boxer's rating.

I know it completely restructures the ratings, but it gives a more accurate description of who are the best fighters today.

thoughts?

That's pretty much what happens now...

well tbh i haven't read the entire explanation page from A to Z but i was surprised when i saw ODLH's ranking at 154, that is what prompted my comment

A close decision loss can gain a boxer points if its against a much higher rated boxer. Oscar got quite a few rating points in his SD loss to Floyd. Furthermore, 154 is a relatively weak division.

[jomothepure]

There isn't by any chance a way you (as in ratings boffins, not as in the general public) can track a boxer's historical rating is there? Specifically Cesar Figueroa. I have it in my head he was somewhere in the 200s before his fight last night. I'm not really questioning or even wondering how he could have moved up (activity possibly?) but just did he actually move up ie. it's my sanity I'm questioning here, not the ratings.

Then I suddenly thought I thought I could find out his rating of yesterday by looking at the movement column in the ratings, but he's listed as "new" there atm. Any help much appreciated.

http://www.boxrec.com/list_bouts.php?hu ... &cat=boxer

[JCS]

... a time depending value should be considered ...

I agree

When I was fiddling with a system I used:

180 prior to 1910
185 from 1910 to 1930
gradually increased from 185 to 200 from 1930 to 1970
gradually increased from 200 to 235 from 1970 to 2008

Martin, Ever have a chance to try this or something similar?

Also, I think the live ratings are broken :).

[computerrank]

... a time depending value should be considered ...

I agree

When I was fiddling with a system I used:

180 prior to 1910
185 from 1910 to 1930
gradually increased from 185 to 200 from 1930 to 1970
gradually increased from 200 to 235 from 1970 to 2008

Martin, Ever have a chance to try this or something similar?

Also, I think the live ratings are broken :).

... there will be a table like your's ... soon ...

There are some other issues to be managed - so it will need some time.

[JCS]

I don't think it'd be such a bad idea to refine that pre-fight inactivity performance adjustment. Vitali at #4 is acceptable, but I would desire a #2 ranking figuring Vitali had more points than Peter did pre-bout before he was inactive... and the result was a dominant one (KO).

There should still be some reduction, but not as drastic, especially considering the comeback performance IMO.

[computerrank]

I don't think it'd be such a bad idea to refine that pre-fight inactivity performance adjustment. Vitali at #4 is acceptable, but I would desire a #2 ranking figuring Vitali had more points than Peter did pre-bout before he was inactive... and the result was a dominant one (KO).

There should still be some reduction, but not as drastic, especially considering the comeback performance IMO.

... already in the pipeline ...

The modified rating for the successful inactive boxer will be the minimum of:

1.) 50% of the opponent's rating
2.) the unreduced rating after the last bout

where factor 2 is new.

The current algorithm sets the modified rating between 50% of the oppent's rating and the fully reduced own rating.

So Vitali Klitschko's modified prebout rating was about 380 with the current algorithm - and would be 540 with the new algorithm.

This would make him #2.

[m1kee50]

There isn't by any chance a way you (as in ratings boffins, not as in the general public) can track a boxer's historical rating is there? Specifically Cesar Figueroa. I have it in my head he was somewhere in the 200s before his fight last night. I'm not really questioning or even wondering how he could have moved up (activity possibly?) but just did he actually move up ie. it's my sanity I'm questioning here, not the ratings.

Then I suddenly thought I thought I could find out his rating of yesterday by looking at the movement column in the ratings, but he's listed as "new" there atm. Any help much appreciated.

http://www.boxrec.com/list_bouts.php?hu ... &cat=boxer

ive asked already to be able to have the ability to see the changing rating over a career. apparently they arent doing it.

[computerrank]

There isn't by any chance a way you (as in ratings boffins, not as in the general public) can track a boxer's historical rating is there? Specifically Cesar Figueroa. I have it in my head he was somewhere in the 200s before his fight last night. I'm not really questioning or even wondering how he could have moved up (activity possibly?) but just did he actually move up ie. it's my sanity I'm questioning here, not the ratings.

Then I suddenly thought I thought I could find out his rating of yesterday by looking at the movement column in the ratings, but he's listed as "new" there atm. Any help much appreciated.

http://www.boxrec.com/list_bouts.php?hu ... &cat=boxer

ive asked already to be able to have the ability to see the changing rating over a career. apparently they arent doing it.

... this would mean to update the current rating for every boxer in every bout in every rating program run every day ... to be consistent with any changes in the records every day

too demanding for the current system

[computerrank]

New release in place:

1.) Promoter home advantage is implemented - matching with the results in Journal of Sports Sciences:

2.) Better adaption of former inactive boxers returning successfully (see Vitali Klitschko)

3.) Time depending table (by decades) regarding the typical weight of the heavyweight boxers - used for division move adaption of the ratings.

[JCS]

The promoter home advantage makes it return! Ought to be controversial.

Martin.. is there a push to make the ratings more objective?

[computerrank]

The promoter home advantage makes it return! Ought to be controversial.

Martin.. is there a push to make the ratings more objective?

... this point was pushed by John ...