Ratings - please read before commenting - Archived

[conan_the_cribber]

I've been thinking about the ratings inflation problem and how to quantify it. If you are adding points to a division because of non-zero sum aspect, then you need to somehow track it during your calculations. Perhaps you could add a test counter that sums up the inflation points i.e. difference between r_a_new + r_b_new and r_a_old and r_old

You could then see how many fights are in the division and then the ratio to come up with an inflation factor per division.

inflation points
---------------
number of bouts

You could then use this ratio to multiply out the ratings in the divisions, to get a better comparison between the divisions.

Just an idea.

conan

Boxers also leave the divisions for inactivity or for ever ...

The launch process (additional points) gives the ratings an initial blow up (degrading with the rating level achieved) and allows some steady state spread of the ratings.

Boxers retiring do reduce the total number of pts in a division, but do not prevent bout inflation. Every bout in a division produces a net positive. The more bouts in a division, the more the boxers profit from a long chain of who beat who and ratings going up.

conan

[conan_the_cribber]

There is a small typo on the Ratings decription page

...

Point 5 is wrong. Correct would be

10 = recent win over opponent with state 9, 11 = 2 recent wins over opponent with state 9, 12 = 3 recent wins over opponent with state 9

...

Also, there is no definition of "recent" in the description.

....

rearranging it....



....

conan

- error in description corrected
- recent means, no losses against inferior opponents in between

all brackets were ok:

Code: Select all

v * cd * (25*n - r_a) * max[6*(n+1), min(r_b, 25*n)] * max(r_a, 18) 
------------------------------------------------------------------- 
25*n  * (max(r_a,18) + max(r_b,18))

Hi. Thanks for the clarification. I'll have a look in detail later. I changed the brackets to make it easier to read.

conan

[conan_the_cribber]

There is a small typo on the Ratings decription page

...

Point 5 is wrong. Correct would be

10 = recent win over opponent with state 9, 11 = 2 recent wins over opponent with state 9, 12 = 3 recent wins over opponent with state 9

...

Also, there is no definition of "recent" in the description.

....

rearranging it....



....

conan

- error in description corrected
- recent means, no losses against inferior opponents in between

all brackets were ok:

Code: Select all

v * cd * (25*n - r_a) * max[6*(n+1), min(r_b, 25*n)] * max(r_a, 18) 
------------------------------------------------------------------- 
25*n  * (max(r_a,18) + max(r_b,18))

Hi. Thanks for the clarification. I'll have a look in detail later. I changed the brackets to make it easier to read.

conan

And what does happen to the launch state in case of an upset loss? So if a launch 11 guy loses to a launch 10 guy, what happens to him?

conan

[computerrank]

Boxers retiring do reduce the total number of pts in a division, but do not prevent bout inflation. Every bout in a division produces a net positive. The more bouts in a division, the more the boxers profit from a long chain of who beat who and ratings going up.

conan

The spread effective also depends strongly on the quality of opponents, not only on the pure number of bouts.

I also think, there is an impact on the rating level in a division.

I once had a normalisation per division depending on the mean rating of the boxers ranked at #8 to #12.

But this showed ugly side effects:

- daily rating jittering
- rating jumps for division movers

I canceled the corrections - as they produced too much trouble.

[conan_the_cribber]

Boxers retiring do reduce the total number of pts in a division, but do not prevent bout inflation. Every bout in a division produces a net positive. The more bouts in a division, the more the boxers profit from a long chain of who beat who and ratings going up.

conan

The spread effective also depends strongly on the quality of opponents, not only on the pure number of bouts.

I also think, there is an impact on the rating level in a division.

I once had a normalisation per division depending on the mean rating of the boxers ranked at #8 to #12.

But this showed ugly side effects:

- daily rating jittering
- rating jumps for division movers

I canceled the corrections - as they produced too much trouble.

Any normalisation attempt (including my suggestion) is just putting bandaids on the fundamental problem i.e. non-zero sum of the results. I'll be putting my thinking cap on soon to come up with a real proposal.

conan

[computerrank]

And what does happen to the launch state in case of an upset loss? So if a launch 11 guy loses to a launch 10 guy, what happens to him?

conan

winner 10 -> 11
loser 11 -> 10

The formula is a bit tricky in detail ...

Play a bit with this Boxrec ratings calculator prototype, if you want ...

http://94.172.160.197/~martin/calc.php

use the Boxrec IDs of the opponents or the known ratings and launch states alternatively ...

[earsjohn]

Why is Ricky Burns ranked at Lightweight when his last fight was at Super Featherweight and it's only his next scheduled fight on 5th November that will be at Lightweight? By way of comparison, Juan Manuel Marquez is also ranked at 135 despite his last fight being at 140 and his next fight, against Manny Pacquiao on 12th November, up at Welterweight.

[computerrank]

Why is Ricky Burns ranked at Lightweight when his last fight was at Super Featherweight and it's only his next scheduled fight on 5th November that will be at Lightweight? By way of comparison, Juan Manuel Marquez is also ranked at 135 despite his last fight being at 140 and his next fight, against Manny Pacquiao on 12th November, up at Welterweight.

The editors assign a boxer to a weight division, it is not the rating program.

So please go to the forum Record Queries & Updates with your question.

[JCS]

Why is Ricky Burns ranked at Lightweight when his last fight was at Super Featherweight and it's only his next scheduled fight on 5th November that will be at Lightweight? By way of comparison, Juan Manuel Marquez is also ranked at 135 despite his last fight being at 140 and his next fight, against Manny Pacquiao on 12th November, up at Welterweight.

The editors assign a boxer to a weight division, it is not the rating program.

So please go to the forum Record Queries & Updates with your question.

This method is truly a mess. I see mistakes all of the time.

When weights are available, bouts should be forced into certain divisions, at least for rating purposes. Though, small allowances should be given when fighters weigh .5 over, for example.

To make this easier, maybe it is only done for bouts in recent years.

[keithmoonhangover]

How can you justify Ezzard Charles' ratin of 3rd best heavyweight of all time. Don't you think something needs tweeking?

[Asterix]

Why is Ricky Burns ranked at Lightweight when his last fight was at Super Featherweight and it's only his next scheduled fight on 5th November that will be at Lightweight? By way of comparison, Juan Manuel Marquez is also ranked at 135 despite his last fight being at 140 and his next fight, against Manny Pacquiao on 12th November, up at Welterweight.

The editors assign a boxer to a weight division, it is not the rating program.

So please go to the forum Record Queries & Updates with your question.

This method is truly a mess. I see mistakes all of the time.

When weights are available, bouts should be forced into certain divisions, at least for rating purposes. Though, small allowances should be given when fighters weigh .5 over, for example.

To make this easier, maybe it is only done for bouts in recent years.

Agreed, big time. There are examples of boxers being assigned a division 2 above/below what they actually weighed in as.

[Blue]

Why is Ricky Burns ranked at Lightweight when his last fight was at Super Featherweight and it's only his next scheduled fight on 5th November that will be at Lightweight? By way of comparison, Juan Manuel Marquez is also ranked at 135 despite his last fight being at 140 and his next fight, against Manny Pacquiao on 12th November, up at Welterweight.

The editors assign a boxer to a weight division, it is not the rating program.

So please go to the forum Record Queries & Updates with your question.

AFAIK; boxers listed on BoxRec are given the weight category they are
campaigning in, which listed by the ABC’s rankings.
http://www.fightnews.com/rankings-2

Or in some cases by their management Faxed/email request or press release.
http://espn.go.com/boxing/story/_/id/70 ... 135-pounds

Scotland's Ricky Burns has vacated his junior lightweight belt and is moving
up to lightweight to face Australian brawler Michael Katsidis for a vacant
interim belt on Nov. 5 in London, promoter allegedly announced.

[JCS]

AFAIK; boxers listed on BoxRec are given the weight category they are
campaigning in, which listed by the ABC’s rankings.
http://www.fightnews.com/rankings-2

Or in some cases by their management Faxed/email request or press release.
http://espn.go.com/boxing/story/_/id/70 ... 135-pounds

Scotland's Ricky Burns has vacated his junior lightweight belt and is moving
up to lightweight to face Australian brawler Michael Katsidis for a vacant
interim belt on Nov. 5 in London, promoter allegedly announced.

I'm not referring to their current setting, I'm referring to the setting assigned to bouts -- often times, it does not actually match the weights in said bouts, which can affect rating calculations.

[Blue]

I was answering the computerrank & earsjohn postings.
(Saving them a trip to Record queries.)

[JCS]

I was answering the computerrank & earsjohn postings.
(Saving them a trip to Record queries.)

[Chauncy]

Hello. My ranking dropped overnight about 15 spots. What happened?

[computerrank]

Hello. My ranking dropped overnight about 15 spots. What happened?

Chauncy,

your rating was 128 until yesterday.

Boxrec rating rules require every boxer to meet a boxer with at least 50% of his own rating within the last 18 months - otherwise the boxer will lose points.

Since the last check on 2010-05-22 your best opponent within the last 18 months was Rob Calloway with 36 before the bout. This was only 28 % of you last rating.

So your rating dropped to 100 - and you lost 28 points.

Best regards
Martin

[Chauncy]

Hello. My ranking dropped overnight about 15 spots. What happened?

Chauncy,

your rating was 128 until yesterday.

Boxrec rating rules require every boxer to meet a boxer with at least 50% of his own rating within the last 18 months - otherwise the boxer will lose points.

Since the last check on 2010-05-22 your best opponent within the last 18 months was Rob Calloway with 36 before the bout. This was only 28 % of you last rating.

So your rating dropped to 100 - and you lost 28 points.

Best regards
Martin

Thank you Martin

[Purse Bid Shakedown]

It's crazy that Jones-Lujan is rated above either Showtime fight tonight. Something is off when Lujan has more points than Moreno, who hasn't lost in 10 yrs, beat much better comp and has a legit title. And Mike Jones is within several points of the No.1 bantam?!

[JCS]

It's crazy that Jones-Lujan is rated above either Showtime fight tonight. Something is off when Lujan has more points than Moreno, who hasn't lost in 10 yrs, beat much better comp and has a legit title. And Mike Jones is within several points of the No.1 bantam?!

Are you comparing Welterweights to Bantamweights? Haha.. That's not how the system works. If a fighter moves up in division, their points get adjusted downward and vice versa.

[Purse Bid Shakedown]

It's crazy that Jones-Lujan is rated above either Showtime fight tonight. Something is off when Lujan has more points than Moreno, who hasn't lost in 10 yrs, beat much better comp and has a legit title. And Mike Jones is within several points of the No.1 bantam?!

Are you comparing Welterweights to Bantamweights? Haha.. That's not how the system works. If a fighter moves up in division, their points get adjusted downward and vice versa.

Other than Lujan and Vic, all of these guys are career BW, WW. Lujan did fight at SWW, with no great success.

Are you saying that lower weight classes are penalized in absolute terms? So that 300 BW points is roughly comparable to 450 WW points, accomplishment-wise?

[JCS]

It's crazy that Jones-Lujan is rated above either Showtime fight tonight. Something is off when Lujan has more points than Moreno, who hasn't lost in 10 yrs, beat much better comp and has a legit title. And Mike Jones is within several points of the No.1 bantam?!

Are you comparing Welterweights to Bantamweights? Haha.. That's not how the system works. If a fighter moves up in division, their points get adjusted downward and vice versa.

Other than Lujan and Vic, all of these guys are career BW, WW. Lujan did fight at SWW, with no great success.

Are you saying that lower weight classes are penalized in absolute terms? So that 300 BW points is roughly comparable to 450 WW points, accomplishment-wise?

They aren't penalized, but if you move UP, your points go DOWN. And vice versa. So you've got it backwards. Points are division-adjusted, not absolute.

[Purse Bid Shakedown]

Are you comparing Welterweights to Bantamweights? Haha.. That's not how the system works. If a fighter moves up in division, their points get adjusted downward and vice versa.

Other than Lujan and Vic, all of these guys are career BW, WW. Lujan did fight at SWW, with no great success.

Are you saying that lower weight classes are penalized in absolute terms? So that 300 BW points is roughly comparable to 450 WW points, accomplishment-wise?

They aren't penalized, but if you move UP, your points go DOWN. And vice versa. So you've got it backwards. Points are division-adjusted, not absolute.

But why are even mention weight jumping? For all practical purposes, here we have career WWs and BWs. I see Lujan hasn't fought above WW in 4 years. Mike Jones is a WW prospect cum contender, with no noteworthy fights above. Moreno, Mares, Agbeko are all established champs at BW, have not moved.

So when you say that points are division adjusted, is this only with respect to moving up/down classes? Or will a cruiser end up with way more points than a flyweight for "comparable" achievement, provided neither has moved up/down?

[JCS]

But why are even mention weight jumping? For all practical purposes, here we have career WWs and BWs. I see Lujan hasn't fought above WW in 4 years. Mike Jones is a WW prospect cum contender, with no noteworthy fights above. Moreno, Mares, Agbeko are all established champs at BW, have not moved.

So when you say that points are division adjusted, is this only with respect to moving up/down classes? Or will a cruiser end up with way more points than a flyweight for "comparable" achievement, provided neither has moved up/down?

Welterweight is a historically stronger division. More fighters, more points. More good fighters, more points.

[Purse Bid Shakedown]

But why are even mention weight jumping? For all practical purposes, here we have career WWs and BWs. I see Lujan hasn't fought above WW in 4 years. Mike Jones is a WW prospect cum contender, with no noteworthy fights above. Moreno, Mares, Agbeko are all established champs at BW, have not moved.

So when you say that points are division adjusted, is this only with respect to moving up/down classes? Or will a cruiser end up with way more points than a flyweight for "comparable" achievement, provided neither has moved up/down?

Welterweight is a historically stronger division. More fighters, more points. More good fighters, more points.

Ok, sorry to keep harping on this. It might be good to factor in the current relative strength of the division, else guys like Sturm and Geale would figure in p4p. And the super 6 guys would be unfairly penalized for the historical weakness of their div. I think bantam has been better, or at least deeper than welter for the past few years.