Ratings - please read before commenting - Archived

[Finkel]

The trouble with basing it on predictions of past fights is at the top of the heavyweight division all four of Wilder's results against top 15 box rec opponents involved some form of A-side advantage.

Ortiz 1 (ring doctor interference) WIN
Fury 1 (one very corrupt scorecard) DRAW
Ortiz 2 (9 count KO) WIN
Fury 2 (points deduction) LOSS

So the prediction is based on the existence of an advantage that the number 2 heavyweight (Wilder) as a non belt holder likely no longer has, and which he relied heavily upon in his most high profile fights.

So if he was to face the number 3 (Anthony Joshua) he would be stripped of that artificial boost inherent in the new ranking calculation.

[SportsRatings]

The WHR ratings winner prediction power for all bouts between connected boxers from 2009 to 2018, with only regarding the bouts BEFORE the respective bout date, is 83.08 percent - and with including 16 additional parameters it is 84.02 percent.

I think what he's getting at is that that prediction rate is entirely for bouts that have already happened. In other words, you've tailored a system that makes sense of past results (since there have been few if any bouts since the system went live). But the percentage at which it will predict future bouts seems uncertain because it has yet to be tested live in that context.

Well, no, he answered my question. I was concerned after hearing about all the backward-looking-smoothing aspect of the ratings that the "predictions" were not real predictions, but they are.

Given his description, the going-forward prediction rate should remain solid, and if I recall worked out to several percentage points better than the old system.

[computerrank]

So you get a guy like Tyson Fury having more points in his career after beating Seferi following 2.5 years away than he did right after beating Klit? How does this work

Klitschko was in decline ...

[computerrank]

I didn' adjust the WHR ratings themselves. All parameters and resulting values are undisturbed. There is a correction over-rule for DISPLAY purpose only and for a t most 1 year after the bout to set the winner above the loser. And that is only for pro boxing, not for amateur boxing, whose ratings are calculated too now, and will be published soon.

Why? The WHR recognised that a boxer’s record was more than the latest fight and that the winner did not always jump the loser...something I always thought was wrong with the previous boxing rankings (I know I’m in the minority) but now that seems to have changed so that a Buster Douglas will always jump over a Tyson? Takes away from the new system if that’s what’s happened. If both systems were available you wouldn’t have to change that element because the old system would still provide what people are used to seeing. Just seems like you have changed the bravest element at the first sign of poster pressure....that’s if I’ve understood you correctly.

This was a Boxrec decision due to existing pro boxing usances. A winner should always be ranked above the loser. But when you look at the bout ratings, you will see the undisturbed ratings.

[JCS]

I've probably said it before.. but.. WHR seems perfect for the olden days of boxing... where a loss or even multiple losses didn't derail a boxer on the way to THE title OR for up and coming boxers that have 4+ matches per year.

For the top talent, WHR seems counterproductive. The most recent match should always hold precedence over everything.

[computerrank]

I've probably said it before.. but.. WHR seems perfect for the olden days of boxing... where a loss or even multiple losses didn't derail a boxer on the way to THE title OR for up and coming boxers that have 4+ matches per year.

For the top talent, WHR seems counterproductive. The most recent match should always hold precedence over everything.

And where is the problem? Boxrec has it exactly so now. Those top talents can walk on top the top ratings of their defeated opponents - if they manage. But WHR as the basics is much more sensitive for the career status and for the prediction power.

[margaret thatcher]

So you get a guy like Tyson Fury having more points in his career after beating Seferi following 2.5 years away than he did right after beating Klit? How does this work

Klitschko was in decline ...

And what about 2.5 years layoff and Seferi being even older than Klit and way way way worse? I just am struggling to wrap my head around how the points progress now.

Whyte has been losing big points over his last fights as well despite winning all of them (against way better opponents than Seferi and not 2.5 years inactivity)

[computerrank]

Klitschko was in decline ...

And what about 2.5 years layoff and Seferi being even older than Klit and way way way worse? I just am struggling to wrap my head around how the points progress now.

Whyte has been losing big points over his last fights as well despite winning all of them (against way better opponents than Seferi and not 2.5 years inactivity)

Fury's increase in points at time of Seferi bout wasn't because he defeated Seferi. But Fury was up to defeat Wilder. Fury's rating at time of Seferi bout was nearly independent of that not relevant episode.

Fury' rating at time of the Pianeta bout and his current rating would only be 20 points lower without the Seferi bout.

[margaret thatcher]

What about Whyte?

[computerrank]

What about Whyte?

Whyte had good wins against Parker and Chisora. But in spite of the wins against Rivas and Wach his rating became more questionable, because those opponents had a high rating uncertainty of 5+ and were rated low in relation to his own rating. So the uncertainty of Whyte's rating increased too and his rating degraded.

Code: Select all

2019-12-07 Heavyweight          W   UD  569964 Dillian Whyte                  309799 Mariusz Wach                    198.5 4.215 6.625 5.898
2019-07-20 Heavyweight          W   UD  569964 Dillian Whyte                  503964 Oscar Rivas                     256.7 3.277 56.89 5.297
2018-12-22 Heavyweight          W   KO  569964 Dillian Whyte                  381739 Dereck Chisora                  312.3 2.636 124.3 2.665
2018-07-28 Heavyweight          W   UD  569964 Dillian Whyte                  613846 Joseph Parker                   305.7 2.440 141.4 2.623

[Finkel]

I'm sure you can explain how a model you have designed acts. But at the moment as an outside observer it seems like a black box of code that only you as the designer understand. We can only assess it on its outputs. And it doesn't make sense to most of us on this thread.

Your results may well be internally valid, but you are doing a very bad job of making a case for its external validity.

i.e. at the moment your argument as to why these results better reflect reality is "because the results"

And external validity is surely the whole point of an unbiased ranking site.



Is there a page where you publish:
The goals of your model;
Along with the methodology;
And also what benchmarks you use to assess its accuracy?

I would very much like to read it.

[Finkel]

So I went back through this thread to see what discussions led up to the change to this new system of WHR. As I wanted to have a better understanding of the arguments for and against the change.

JCS was strongly against the idea of the change, citing multiple issues the rankings were throwing up.

Cobwebcat is the main supporter who I think is fair to say was encouraging the change.

Btw I just want to say I can see both sides of the argument. I personally side with JCS, but I think it's a problem when the main supporter of the change Cobwebcat acknowledges an obvious flaw just before the new system goes live:

A boxing system whereby the winner of the bout must jump the loser is ridiculous to me. It might be the way it’s always been done but it doesn’t make it right. If Fulham beat Man City in the cup tomorrow do we say they are the better team and promote them?

I know the argument is boxing is different because fighters don’t fight very often I do get that but if you have a system that produces a better prediction AND looks just as good on the eye as the current one why not use it?

Ruiz jumped above AJ after his win but hardly anyone believed him a better fighter he just fought better that one time as Fulham might. Sure enough the rematch was a wide point victory for AJ. No one fight should have such a huge impact as to promote a rank outsider (not meaning Ruiz here) above a fighter with an impeccable record who has an off night.

If the algorithm promotes the underdog above the loser that’s fine but not every time for the sake of it...If Federer loses in the first round to a no hoper the winner isn’t immediately ranked above him.

I don’t get the argument other that’s what people expect to see but that can’t be the only reason for doing it. The WHR are objectively better than the current system for present fighters and they look fine as a bonus. The all-time one can’t be measured so is subjective I admit.

I respect your view and you might be in the majority but I disagree. On your MMA rankings I always go straight to modified elo.

Publishing both rankings is a good idea but I get the impression that one of the reasons the WHR is being considered is its speed of calculation vs current system. I could be wrong though.

Soccer is not boxing. Neither is tennis. Those sports are a much better fit for a pure mathematical system.

Ruiz above AJ after the first fight was "right". It wasn't even a flukey win... I can't see another argument. Did I favor him in the rematch? Of course not. Who deserves to be ranked higher and who is better are two different questions

The only time I'd support usage of WHR in a "standard" ranking is a situation in which there is insufficient data... like a midwest freak show who's pounding out a nobody per month... or a former champion who hasn't fought in three years that needs a placement. Something like that.

...

I notice common doubts even inside the boxing community, whether an upset winner should be regarded a valid and respected rank winner, as long as he did'nt confirm his upset with a won return bout. So he may be the champ, but he will not be the respected boxer until then.

This won return bout would let him pass his opponent in the WHR ratings too.

I think that's fair enough. He could even prove it wasn't a one off by improving his record against other top fighters too which, in a way, would be even better as it would show it wasn't just that he had the style to beat that one fighter.

No doubt there will be lots of people unhappy at the change though. I think it's still the right thing to do (sorry to JCS) Boxrec by its very nature doesn't need to follow other rankings procedures though I accept they can't look ridiculous. I don't think they do look ridiculous now.

Even if they don't change/improve they're still the best around

You can't win with rankings all you can ever do is explain the rationale.

...

Seems that WHR has it

Fury
Wilder
Joshua
Usyk
Povetkin???
Ruiz

I could be wrong but I think that’s a snapshot in time ie it won’t have updated with subsequent fights.

To have Fury at #2 P4P.. it had better be after this weekend's fights....

BoxRec Whole History Ratings are live now ...

[computerrank]

I'm sure you can explain how a model you have designed acts. But at the moment as an outside observer it seems like a black box of code that only you as the designer understand. We can only assess it on its outputs. And it doesn't make sense to most of us on this thread.

Your results may well be internally valid, but you are doing a very bad job of making a case for its external validity.

i.e. at the moment your argument as to why these results better reflect reality is "because the results"

And external validity is surely the whole point of an unbiased ranking site.



Is there a page where you publish:
The goals of your model;
Along with the methodology;
And also what benchmarks you use to assess its accuracy?

I would very much like to read it.

Here you will find all: goals of the model, methodology and benchmarks used.

https://www.remi-coulom.fr/WHR/WHR.pdf

[Finkel]

I'm sure you can explain how a model you have designed acts. But at the moment as an outside observer it seems like a black box of code that only you as the designer understand. We can only assess it on its outputs. And it doesn't make sense to most of us on this thread.

Your results may well be internally valid, but you are doing a very bad job of making a case for its external validity.

i.e. at the moment your argument as to why these results better reflect reality is "because the results"

And external validity is surely the whole point of an unbiased ranking site.



Is there a page where you publish:
The goals of your model;
Along with the methodology;
And also what benchmarks you use to assess its accuracy?

I would very much like to read it.

Here you will find all: goals of the model, methodology and benchmarks used.

https://www.remi-coulom.fr/WHR/WHR.pdf

Thank you.


The conference paper you linked used the Japanese game of Go (igo) to runs its models. Go is a game confined by its rules. There is a winner and a loser, and possibly a draw as one of three outcomes. Boxing however as we know is not confined by such a strict set of rules. It is unfortunately also a sport dogged by corruption, where outcomes are not always decided by what happens in the ring.

As I spoke about earlier in this thread, how does your model account for A-side advantage? Or advantage given to the defending champion?

Once a Champion loses their belt and is no longer the A-side then they have lost something that was calculated as intrinsic to their rating. For some champions this advantage played a more overt part in their climb to the top.

This isn’t like a fighter’s ability to cheat in ring/use ring guile; that in a sense would always be with them. I’m talking about judges and referees bending or breaking the rules to show preference to the A-side/Champion, and how your model accounts for the inclusion and the loss of that.

Because if you can't factor that in, I would question if this is an appropriate model to use for assessing the relative ability of the different fighters, and its ability to predict future outcomes based on current rank.

Ignoring the above point for a moment, what are your thoughts on network models? This for example seems like something that would fit better to boxing, it was modelled with data from ATP (Tennis):
https://www.nature.com/articles/srep00904

[marcianofan]

I'm still hoping someone will tell me what the old prediction rate was in order to compare to the 83-84% of the new system. I think it's pretty relevant, considering we've completely abandoned the idea of ratings reacting to individual fight results logically in order to achieve whatever the difference is. If it's 15%....okay. But if it's 3 or 4%? I don't see how that could be worth it. And how do draws factor in?

Also...how does the predicting model take into account (without hindsight) an Olympic Gold medalist making his debut against a mediocre opponent with some point value, but not much? Or are those not included in the figures?

[Finkel]

According to the current ratings, peak Tyson Fury, was the one who fought in Wilder v Fury 1, which ended up a draw.

In what way does that reflect reality?

Tyson Fury 653.5 Dereck Chisora 103.8
Tyson Fury 704.4 Christian Hammer 39.44
Tyson Fury 879.2 Wladimir Klitschko 537.9
Tyson Fury 990.3 Sefer Seferi 4.047
Tyson Fury 1042 Francesco Pianeta 2.408
Tyson Fury 1136 Deontay Wilder 959.3
Tyson Fury 1118 Tom Schwartz 19.66
Tyson Fury 1108 Otto Wallin 3.798
Tyson Fury 1094 Deontay Wilder 659.1

According to the ratings, when Joshua beat Ruiz Jr. in the rematch, they both got worse.

Anthony Joshua 431.9 Charles Martin 66.11
Anthony Joshua 488.7 Dominic Breazeale 61.83
Anthony Joshua 603.1 Eric Molina 13.75
Anthony Joshua 738.2 Wladimir Klitschko 236.9
Anthony Joshua 771.4 Carlos Takam 36.27
Anthony Joshua 806.8 Joseph Parker 147.2
Anthony Joshua 791.7 Alexander Povetkin 306.5
Anthony Joshua 599.2 Andy Ruiz Jr. 267.4
Anthony Joshua 552.5 Andy Ruiz Jr. 202.3

According to the ratings, the Klitschko that lost to Joshua was a shadow of his former self:

Wladimir Klitchsko 1146 Tony Thompson 79.64
Wladimir Klitchsko 1139 Mariusz Wach 28.59
Wladimir Klitchsko 1139 Francesco Pianeta 15.6
Wladimir Klitchsko 1171 Alexander Povetkin 452.9
Wladimir Klitchsko 987.2 Alex Lepai 15.96
Wladimir Klitchsko 873.7 Kubrat Pulev 171.1
Wladimir Klitchsko 714 Bryant Jennings 52.64
Wladimir Klitchsko 537.9 Tyson Fury 879.2
Wladimir Klitchsko 236.9 Anthony Joshua 738.2


I'm curious what cascade effect it would have on the rankings if the DRAW in Wilder v Fury 1, was changed to a NWS in favour of Fury (given its fair to say 70-30 of newspapers gave it to Fury if not more)

Also I am curious what would happen, if Fury and Wilder continued to fight each other exclusively for the next 5 years, and Joshua went on to beat every other challenger in the division. Given that according to the above statistics all three fighters are on the decline, so would Joshua always be at #3?

[computerrank]

I'm still hoping someone will tell me what the old prediction rate was in order to compare to the 83-84% of the new system. I think it's pretty relevant, considering we've completely abandoned the idea of ratings reacting to individual fight results logically in order to achieve whatever the difference is. If it's 15%....okay. But if it's 3 or 4%? I don't see how that could be worth it. And how do draws factor in?

Also...how does the predicting model take into account (without hindsight) an Olympic Gold medalist making his debut against a mediocre opponent with some point value, but not much? Or are those not included in the figures?

For the old Boxrec r_41 release:
- raw winner prediction ratio based on the old rating only = 0.7482
- winner prediction ratio for r_41 with 16 additional parameters = 0.8345

For WHR:
- raw winner prediction ratio based on the WHR rating only = 0.8308
- winner prediction ratio for r_41 with 16 additional parameters = 0.8402

So the raw winner prediction ratio of the old rating was much lower by more than 8 percent points (better the error was 25 percent against 17 percent, so the error ratio of the old rating was 50 percent higher).

The raw prediction power of the old rating was much worse than that of the new WHR.

16 additional parameters could nearly compensate that, but WHR is nearly as good without the 16 additional parameters.

[computerrank]

According to the current ratings, peak Tyson Fury, was the one who fought in Wilder v Fury 1, which ended up a draw.

In what way does that reflect reality?

--> WHR uncertainty increases with the last bouts. So the conservatively estimated rating values will be down at the end of the list.

Tyson Fury 653.5 Dereck Chisora 103.8
Tyson Fury 704.4 Christian Hammer 39.44
Tyson Fury 879.2 Wladimir Klitschko 537.9
Tyson Fury 990.3 Sefer Seferi 4.047
Tyson Fury 1042 Francesco Pianeta 2.408
Tyson Fury 1136 Deontay Wilder 959.3
Tyson Fury 1118 Tom Schwartz 19.66
Tyson Fury 1108 Otto Wallin 3.798
Tyson Fury 1094 Deontay Wilder 659.1

According to the ratings, when Joshua beat Ruiz Jr. in the rematch, they both got worse.

--> Same with Joshua

Anthony Joshua 431.9 Charles Martin 66.11
Anthony Joshua 488.7 Dominic Breazeale 61.83
Anthony Joshua 603.1 Eric Molina 13.75
Anthony Joshua 738.2 Wladimir Klitschko 236.9
Anthony Joshua 771.4 Carlos Takam 36.27
Anthony Joshua 806.8 Joseph Parker 147.2
Anthony Joshua 791.7 Alexander Povetkin 306.5
Anthony Joshua 599.2 Andy Ruiz Jr. 267.4
Anthony Joshua 552.5 Andy Ruiz Jr. 202.3

According to the ratings, the Klitschko that lost to Joshua was a shadow of his former self:

--> Klitschko was done

Wladimir Klitchsko 1146 Tony Thompson 79.64
Wladimir Klitchsko 1139 Mariusz Wach 28.59
Wladimir Klitchsko 1139 Francesco Pianeta 15.6
Wladimir Klitchsko 1171 Alexander Povetkin 452.9
Wladimir Klitchsko 987.2 Alex Lepai 15.96
Wladimir Klitchsko 873.7 Kubrat Pulev 171.1
Wladimir Klitchsko 714 Bryant Jennings 52.64
Wladimir Klitchsko 537.9 Tyson Fury 879.2
Wladimir Klitchsko 236.9 Anthony Joshua 738.2


I'm curious what cascade effect it would have on the rankings if the DRAW in Wilder v Fury 1, was changed to a NWS in favour of Fury (given its fair to say 70-30 of newspapers gave it to Fury if not more)

--> Fury 1078, Joshua 557, Wilder 437

Also I am curious what would happen, if Fury and Wilder continued to fight each other exclusively for the next 5 years, and Joshua went on to beat every other challenger in the division. Given that according to the above statistics all three fighters are on the decline, so would Joshua always be at #3?

-> after Joshua would have defeated Usyk and Povetkin - the ratings would be Joshua 1382, Fury 1103, Wilder 674

Answers inserted -->

[Finkel]

According to the current ratings, peak Tyson Fury, was the one who fought in Wilder v Fury 1, which ended up a draw.

In what way does that reflect reality?

--> WHR uncertainty increases with the last bouts. So the conservatively estimated rating values will be down at the end of the list.

Tyson Fury 653.5 Dereck Chisora 103.8
Tyson Fury 704.4 Christian Hammer 39.44
Tyson Fury 879.2 Wladimir Klitschko 537.9
Tyson Fury 990.3 Sefer Seferi 4.047
Tyson Fury 1042 Francesco Pianeta 2.408
Tyson Fury 1136 Deontay Wilder 959.3
Tyson Fury 1118 Tom Schwartz 19.66
Tyson Fury 1108 Otto Wallin 3.798
Tyson Fury 1094 Deontay Wilder 659.1

According to the ratings, when Joshua beat Ruiz Jr. in the rematch, they both got worse.

--> Same with Joshua

Anthony Joshua 431.9 Charles Martin 66.11
Anthony Joshua 488.7 Dominic Breazeale 61.83
Anthony Joshua 603.1 Eric Molina 13.75
Anthony Joshua 738.2 Wladimir Klitschko 236.9
Anthony Joshua 771.4 Carlos Takam 36.27
Anthony Joshua 806.8 Joseph Parker 147.2
Anthony Joshua 791.7 Alexander Povetkin 306.5
Anthony Joshua 599.2 Andy Ruiz Jr. 267.4
Anthony Joshua 552.5 Andy Ruiz Jr. 202.3

According to the ratings, the Klitschko that lost to Joshua was a shadow of his former self:

--> Klitschko was done

Wladimir Klitchsko 1146 Tony Thompson 79.64
Wladimir Klitchsko 1139 Mariusz Wach 28.59
Wladimir Klitchsko 1139 Francesco Pianeta 15.6
Wladimir Klitchsko 1171 Alexander Povetkin 452.9
Wladimir Klitchsko 987.2 Alex Lepai 15.96
Wladimir Klitchsko 873.7 Kubrat Pulev 171.1
Wladimir Klitchsko 714 Bryant Jennings 52.64
Wladimir Klitchsko 537.9 Tyson Fury 879.2
Wladimir Klitchsko 236.9 Anthony Joshua 738.2


I'm curious what cascade effect it would have on the rankings if the DRAW in Wilder v Fury 1, was changed to a NWS in favour of Fury (given its fair to say 70-30 of newspapers gave it to Fury if not more)

--> Fury 1078, Joshua 557, Wilder 437

Also I am curious what would happen, if Fury and Wilder continued to fight each other exclusively for the next 5 years, and Joshua went on to beat every other challenger in the division. Given that according to the above statistics all three fighters are on the decline, so would Joshua always be at #3?

-> after Joshua would have defeated Usyk and Povetkin - the ratings would be Joshua 1382, Fury 1103, Wilder 674

Answers inserted -->

Thanks for the response

[computerrank]

One thing I’m still unclear about is that @Computerrank says the bout rating will not reflect the winner always on top scenario but that would mean a fighter having two different ratings on the same rating system but in different places. Unless he means a fighter can be ranked higher with a lower rating on the main ranking pages...which I doubt!

I’m a little confused by this hybrid to be honest.

Looking at the bout rankings in the record, you will see the undisturbed WHR ratings. In case of an upset, the losers WHR rating may be higher than the winner's.
Looking at the division ratings, the current ratings will reflect the rule winner above loser.

[jerd16]

Both already is posted:
- Q1: yes, better likelihood for all results and better prediction power
- Q2: a toggle with multiple options is possible and on my list, but there is no decision about priorities

My last question: is there any definable scaling between the previous set of ratings and these (apologies if my statistical terminology is not correct). I.e If the average fighter had 5 points in the previous ratings, under the new formula what would the average fighter's rating be? And does this scale/ratio apply consistently?

Do you have any update on this? I know you said the star rating structure is the same. But I was most interested in, say, what would be the rough equivalent in the WHR system of a 5 point fighter in the old system. Also, whether there was a formula to convert old ratings to new WHR ones?

[computerrank]

My last question: is there any definable scaling between the previous set of ratings and these (apologies if my statistical terminology is not correct). I.e If the average fighter had 5 points in the previous ratings, under the new formula what would the average fighter's rating be? And does this scale/ratio apply consistently?

Do you have any update on this? I know you said the star rating structure is the same. But I was most interested in, say, what would be the rough equivalent in the WHR system of a 5 point fighter in the old system. Also, whether there was a formula to convert old ratings to new WHR ones?

There is no conversion formula, but the old and new tables might help:

Code: Select all

Bout Star Ratings OLD

All bouts are rated with 0 to 5 stars.

Men

    5 stars = both opponents have 279 rating points at least ~ best 100 boxers
    4 stars = both opponents have 122 rating points at least ~ best 300 boxers
    3 stars = both opponents have 40.5 rating points at least ~ best 900 boxers
    2 stars = both opponents have 5.92 rating points at least ~ best 2700 boxers
    1 stars = both opponents have 0.428 rating points at least = best 8100 boxers
    0 stars = one opponents has not even 0.428 rating points at least = not rated boxers

Women

    5 stars = both opponents have 12.2 rating points at least ~ best 50 boxers
    4 stars = both opponents have 6.15 rating points at least ~ best 100 boxers
    3 stars = both opponents have 2.20 rating points at least ~ best 200 boxers
    2 stars = both opponents have 0.554 rating points at least ~ best 400 boxers
    1 stars = both opponents have 0.023 rating points at least = best 800 boxer
    0 stars = one opponents has not even 0.023 rating points at least = not rated boxers

Code: Select all

Bout Star Ratings NEW

All bouts are rated with 0 to 5 stars.

Men

    5 stars = both opponents have 68.9 rating points at least ~ best 100 boxers
    4 stars = both opponents have 21.1 rating points at least ~ best 300 boxers
    3 stars = both opponents have 6.58 rating points at least ~ best 900 boxers
    2 stars = both opponents have 1.31 rating points at least ~ best 2700 boxers
    1 stars = both opponents have 0.0529 rating points at least = best 8100 boxers
    0 stars = one opponents has not even 0.0529 rating points at least = not rated boxers

Women

    5 stars = both opponents have 34.7 rating points at least ~ best 50 boxers
    4 stars = both opponents have 15.7 rating points at least ~ best 100 boxers
    3 stars = both opponents have 5.02 rating points at least ~ best 200 boxers
    2 stars = both opponents have 0.787 rating points at least ~ best 400 boxers
    1 stars = both opponents have 0.0128 rating points at least = best 800 boxer
    0 stars = one opponents has not even 0.0128 rating points at least = not rated boxers

[jerd16]

Do you have any update on this? I know you said the star rating structure is the same. But I was most interested in, say, what would be the rough equivalent in the WHR system of a 5 point fighter in the old system. Also, whether there was a formula to convert old ratings to new WHR ones?

There is no conversion formula, but the old and new tables might help:

Code: Select all

Bout Star Ratings OLD

All bouts are rated with 0 to 5 stars.

Men

    5 stars = both opponents have 279 rating points at least ~ best 100 boxers
    4 stars = both opponents have 122 rating points at least ~ best 300 boxers
    3 stars = both opponents have 40.5 rating points at least ~ best 900 boxers
    2 stars = both opponents have 5.92 rating points at least ~ best 2700 boxers
    1 stars = both opponents have 0.428 rating points at least = best 8100 boxers
    0 stars = one opponents has not even 0.428 rating points at least = not rated boxers

Women

    5 stars = both opponents have 12.2 rating points at least ~ best 50 boxers
    4 stars = both opponents have 6.15 rating points at least ~ best 100 boxers
    3 stars = both opponents have 2.20 rating points at least ~ best 200 boxers
    2 stars = both opponents have 0.554 rating points at least ~ best 400 boxers
    1 stars = both opponents have 0.023 rating points at least = best 800 boxer
    0 stars = one opponents has not even 0.023 rating points at least = not rated boxers

Code: Select all

Bout Star Ratings NEW

All bouts are rated with 0 to 5 stars.

Men

    5 stars = both opponents have 68.9 rating points at least ~ best 100 boxers
    4 stars = both opponents have 21.1 rating points at least ~ best 300 boxers
    3 stars = both opponents have 6.58 rating points at least ~ best 900 boxers
    2 stars = both opponents have 1.31 rating points at least ~ best 2700 boxers
    1 stars = both opponents have 0.0529 rating points at least = best 8100 boxers
    0 stars = one opponents has not even 0.0529 rating points at least = not rated boxers

Women

    5 stars = both opponents have 34.7 rating points at least ~ best 50 boxers
    4 stars = both opponents have 15.7 rating points at least ~ best 100 boxers
    3 stars = both opponents have 5.02 rating points at least ~ best 200 boxers
    2 stars = both opponents have 0.787 rating points at least ~ best 400 boxers
    1 stars = both opponents have 0.0128 rating points at least = best 800 boxer
    0 stars = one opponents has not even 0.0128 rating points at least = not rated boxers

Thanks thats helpful. As a slight aside, I noticed there are ~20990 active fighters now. For how many years has the number been this high? Is there demonstrably more active fighters now than there was, say, in the 70s?

[Finkel]

So I went back through this thread to see what discussions led up to the change to this new system of WHR. As I wanted to have a better understanding of the arguments for and against the change.

JCS was strongly against the idea of the change, citing multiple issues the rankings were throwing up.

Cobwebcat is the main supporter who I think is fair to say was encouraging the change.

Btw I just want to say I can see both sides of the argument. I personally side with JCS, but I think it's a problem when the main supporter of the change Cobwebcat acknowledges an obvious flaw just before the new system goes live:



...



...

Finkel: just to say that I liked the former ratings too and Computerrank was already looking to change to the WHR to speed things up; I just supported the idea. The two points that I liked about WHR was the increased predictability and, most importantly, not placing the winner over the loser in all circumstances. The second of these two reasons has been taken away by the site owners which I understand but disagree with. It seems like the winner of the fight will always jump in front of the loser but that, all else being equal, the winner will start to fall away, after a year, at a quicker rate than the previously higher ranked fighter...I’m not sure how that will work in practice.

My preference then and now was to have both old and new formats. The old would always show winner above the loser while WHR would be free to decide. It seems like we might get both versions in time but that neither system will allow the winner to be ranked lower than the loser. That’s a pity to my mind as it would be something different to all the other systems out there and I liked it’s logic. It’s what I think Computerrank would have done had a block not been put on it.

One thing I’m still unclear about is that @Computerrank says the bout rating will not reflect the winner always on top scenario but that would mean a fighter having two different ratings on the same rating system but in different places. Unless he means a fighter can be ranked higher with a lower rating on the main ranking pages...which I doubt!

I’m a little confused by this hybrid to be honest.

Thanks for the reply. I completely agree we should have access to both formats. My issue is that boxrec markets itself as the official source of boxing record keeping.

They in effect are a database of fighter resumes.

So it is intuitive that the rankings would be reflective of who had the best resume at any point in time.

I understand your position, that you favor a rankings system based on being able to predict more accurately who wins a head-to-head. However, whilst this new ranking system is clearly designed to show who the algorithm thinks would win a head to head, something that keeps being avoided is the question of how it takes into account A-side advantage.

Also whilst this is something I wasn't initially going to mention, it feels heavily like Boxrec is showing favoritism toward the American market (again) with this change in how it calculates its rankings.

Look at the change in the rankings of the Heavyweight top 10 based both on nationality and which market they are based in.

Previously it was
1. UK
2. UK
3. Ukraine
4. USA
5. USA
6. UK
7. Russia
8. Cuba
9. Bulgaria
10. UK

Now
1. UK
2. USA
3. UK
4. Ukraine
5. Russia
6. USA
7. UK
8. USA
9. UK
10. Cuba

Red = fighters based in the USA
Blue = fighters based in the UK

One of my issues with boxrecs rankings was it generally favoured fighters in the American circuit who were racking up wins against journeymen. They always seemed to rate fighters like Gerald Washington a little too highly. But no system is perfect.

But this change has gone even further to show favouritism toward the USA market.

So I ask is this just a happy coincidence at a time when people are openly discussing the dearth of quality heavyweights coming out of America? Looking at the new rankings the American system looks alive and well.

Personally I go back to what you suggest there should be an option for both. But it is misleading to recalculate the rankings on hypothetical head-to-heads as opposed to strength of resume, specifically on a website dedicated to recording boxer's resumes accurately.
Especially when there is no explanation for the casual observer that this is the desired function of the boxrec rankings.

[margaret thatcher]

Well looking there, it seems that Fury and Povetkin both got substantial improvements in this system, so it follows that Wilder and Hunter would since they drew with them. Not sure it's some systematically pro USA thing

Japanese tend to have been the most overrated by BoxRec imo, at least before. Something about the ratings led to them getting too many points for their domestic Japanese level fights. You'd get Japanese guys up at 160 and 168 getting it too, I think the guy Murata debuted against was ranked top 20 at the time. at least under the old rankings, but in reality he was nowhere near that