Re: Ratings - please read before commenting
Posted: 25 Oct 2020, 22:08
shots fired
@Manrae, @Offthehook, stop that talking in this thread, please. Otherwise, I would have to take you out of this thread.
BoxRec
Ratings - please read before commenting - Archived
Page 165 of 180
Re: Ratings - please read before commenting
Posted: 26 Oct 2020, 03:34
Re: Ratings - please read before commenting
Posted: 28 Oct 2020, 19:36
Computerrank, I think that the annual ratings were better before when you were using the highest point exchange for the year instead of the last weight that they fought at.computerrank wrote: ↑26 Oct 2020, 03:34@Manrae, @Offthehook, stop that talking in this thread, please. Otherwise, I would have to take you out of this thread.
The problem now that I see it, is boxrec's handling with weight classes. The erroneously put fights in the wrong weight classes when they see fighters 1/2lb to 1lb over the weight limit, not realizing that it is actually ALLOWED in non-title fights by all sanctioning bodies.
For example, if I weighed in at 148, it's still a welterweight bout, as long as it's non-title. But boxrec will but the fight at Junior Middleweight.
Who's in charge of that?
Re: Ratings - please read before commenting
Posted: 29 Oct 2020, 08:02
The editors enter the bout division.Manrae wrote: ↑28 Oct 2020, 19:36 Computerrank, I think that the annual ratings were better before when you were using the highest point exchange for the year instead of the last weight that they fought at.
The problem now that I see it, is boxrec's handling with weight classes. The erroneously put fights in the wrong weight classes when they see fighters 1/2lb to 1lb over the weight limit, not realizing that it is actually ALLOWED in non-title fights by all sanctioning bodies.
For example, if I weighed in at 148, it's still a welterweight bout, as long as it's non-title. But boxrec will but the fight at Junior Middleweight.
Who's in charge of that?
You discuss a quite simple case. But how to handle a case, where one boxer's weight is lot more over or under that limits. And it may depend on whether a boxer won or lost with a much lower or higher weight ...
Re: Ratings - please read before commenting
Posted: 29 Oct 2020, 16:42
No, I didn't mean to fire any shots. I was being sincere, dammit!
Re: Ratings - please read before commenting
Posted: 29 Oct 2020, 21:26
Nah, even his biographer admits "he is rarely mentioned among Australia's greatest ever boxers". Carroll was the biggest regional star in his day, so it's not like he was overlooked by historians. It's just somehow staying local still made him WHR's #7 welterweight of all time.
Sorry, I forgot you're not really a boxing guy.
Re: Ratings - please read before commenting
Posted: 30 Oct 2020, 00:18
Assuming the editors do not receive official bout sheets with results, weight class information, etccomputerrank wrote: ↑29 Oct 2020, 08:02The editors enter the bout division.Manrae wrote: ↑28 Oct 2020, 19:36 Computerrank, I think that the annual ratings were better before when you were using the highest point exchange for the year instead of the last weight that they fought at.
The problem now that I see it, is boxrec's handling with weight classes. The erroneously put fights in the wrong weight classes when they see fighters 1/2lb to 1lb over the weight limit, not realizing that it is actually ALLOWED in non-title fights by all sanctioning bodies.
For example, if I weighed in at 148, it's still a welterweight bout, as long as it's non-title. But boxrec will but the fight at Junior Middleweight.
Who's in charge of that?
You discuss a quite simple case. But how to handle a case, where one boxer's weight is lot more over or under that limits. And it may depend on whether a boxer won or lost with a much lower or higher weight ...
Assuming there's no video evidence of which weight class they fought in
Assuming there's no online news article/coverage of the fight for clarification of which weight they fought in
One would have to simply leave it blank
Also, concerning p4p... is there ANY other algorithm that can be used because the heavyweight division is over represented. Alexander Povetkin is a top 10 heavyweight, no question, but p4p?
Re: Ratings - please read before commenting
Posted: 30 Oct 2020, 11:25
The current top 100 men distribute like that:Manrae wrote: ↑30 Oct 2020, 00:18
Assuming the editors do not receive official bout sheets with results, weight class information, etc
Assuming there's no video evidence of which weight class they fought in
Assuming there's no online news article/coverage of the fight for clarification of which weight they fought in
One would have to simply leave it blank
Also, concerning p4p... is there ANY other algorithm that can be used because the heavyweight division is over represented. Alexander Povetkin is a top 10 heavyweight, no question, but p4p?
Code: Select all
Heavyweight 10
Cruiserweight 7
Light Heavyweight 11
Super Middleweight 5
Middleweight 9
Light Middleweight 5
Welterweight 15
Light Welterweight 8
Lightweight 8
Super Featherweight 7
Featherweight 4
Super Bantamweight 1
Bantamweight 4
Super Flyweight 5
Flyweight 1Re: Ratings - please read before commenting
Posted: 30 Oct 2020, 18:57
Alexander Povetkin p4p top 10, should raise major concern...computerrank wrote: ↑30 Oct 2020, 11:25The current top 100 men distribute like that:Manrae wrote: ↑30 Oct 2020, 00:18
Assuming the editors do not receive official bout sheets with results, weight class information, etc
Assuming there's no video evidence of which weight class they fought in
Assuming there's no online news article/coverage of the fight for clarification of which weight they fought in
One would have to simply leave it blank
Also, concerning p4p... is there ANY other algorithm that can be used because the heavyweight division is over represented. Alexander Povetkin is a top 10 heavyweight, no question, but p4p?
I cannot see, what should be wrong with that.Code: Select all
Heavyweight 10 Cruiserweight 7 Light Heavyweight 11 Super Middleweight 5 Middleweight 9 Light Middleweight 5 Welterweight 15 Light Welterweight 8 Lightweight 8 Super Featherweight 7 Featherweight 4 Super Bantamweight 1 Bantamweight 4 Super Flyweight 5 Flyweight 1
Re: Ratings - please read before commenting
Posted: 30 Oct 2020, 19:28
P4P on Boxrec is more of an "Absolute" than a P4P that the magazines have/had.
For curiosity... I ran a calculation using the BoxRec ratings...
I took the boxers' ratings and compared it to the boxer ranked in the spot that was equal (or closest) to the amount of total fighters in the division divided by 100.
For example.. in Tyson Fury's case, he has 1025 points. There are 1408 boxers in the division at the moment.. so I compared his rating to the #14 ranked boxer. His rating is 20.7x greater than that boxer. I did this for all top boxers relative to their current division and this is what I came up with..
1 Saul Alvarez
2 Gennadiy Golovkin
3 Tyson Fury
4 Errol Spence Jr
5 Teofimo Lopez
6 Naoya Inoue
7 Terence Crawford
8 Jermall Charlo
9 Manny Pacquiao
10 Deontay Wilder
11 Josh Warrington
12 Callum Smith
13 Mairis Briedis
14 Vasyl Lomachenko
15 Anthony Joshua
16 Miguel Berchelt
17 Daniel Jacobs
18 Chris Eubank Jr
19 Shawn Porter
20 Gervonta Davis
21 Rey Vargas
22 Josh Taylor
23 Keith Thurman
24 Jose Carlos Ramirez
25 Jermell Charlo
Personally, I think that is a pretty good list. It isn't perfect, but I think it paints a better picture.
Granted, it is somewhat artificial.... once a boxer moves divisions, his "P4P" rating will shift and could shift dramatically..... I have something like this for my MMA site, but call it Division Dominance... I have a separate P4P rating which is very comparable to the actual rating, but there's no divisional adjustments.
For curiosity... I ran a calculation using the BoxRec ratings...
I took the boxers' ratings and compared it to the boxer ranked in the spot that was equal (or closest) to the amount of total fighters in the division divided by 100.
For example.. in Tyson Fury's case, he has 1025 points. There are 1408 boxers in the division at the moment.. so I compared his rating to the #14 ranked boxer. His rating is 20.7x greater than that boxer. I did this for all top boxers relative to their current division and this is what I came up with..
1 Saul Alvarez
2 Gennadiy Golovkin
3 Tyson Fury
4 Errol Spence Jr
5 Teofimo Lopez
6 Naoya Inoue
7 Terence Crawford
8 Jermall Charlo
9 Manny Pacquiao
10 Deontay Wilder
11 Josh Warrington
12 Callum Smith
13 Mairis Briedis
14 Vasyl Lomachenko
15 Anthony Joshua
16 Miguel Berchelt
17 Daniel Jacobs
18 Chris Eubank Jr
19 Shawn Porter
20 Gervonta Davis
21 Rey Vargas
22 Josh Taylor
23 Keith Thurman
24 Jose Carlos Ramirez
25 Jermell Charlo
Personally, I think that is a pretty good list. It isn't perfect, but I think it paints a better picture.
Granted, it is somewhat artificial.... once a boxer moves divisions, his "P4P" rating will shift and could shift dramatically..... I have something like this for my MMA site, but call it Division Dominance... I have a separate P4P rating which is very comparable to the actual rating, but there's no divisional adjustments.
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 00:18
Relative rating ratio within a division used as placement in P4P. Impressive work!JCS wrote: ↑30 Oct 2020, 19:28 For example.. in Tyson Fury's case, he has 1025 points. There are 1408 boxers in the division at the moment.. so I compared his rating to the #14 ranked boxer. His rating is 20.7x greater than that boxer. I did this for all top boxers relative to their current division and this is what I came up with..
I wonder if this can be applied to all-time rankings...Yup, happened to Usyk. Or, if Loma slims back down to a more natural 130lbs, he'll rise in your list.
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 04:52
It is for divisional dominance, yes.JCS wrote: ↑30 Oct 2020, 19:28 ...
Personally, I think that is a pretty good list. It isn't perfect, but I think it paints a better picture.
Granted, it is somewhat artificial.... once a boxer moves divisions, his "P4P" rating will shift and could shift dramatically..... I have something like this for my MMA site, but call it Division Dominance... I have a separate P4P rating which is very comparable to the actual rating, but there's no divisional adjustments.
But for example Light Heavyweight is not represented there, although it's level is quite high. But there are no dominating boxers.
So it more about what your personal preferences are: absolute level like WHR or divisional dominance or defeating top boxers or staying undefeated or major titles ...
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 10:47
Boxers mainly box within their divisions.. especially for top ranked boxers... so this speaks to there being no Top P4P boxers within that division.computerrank wrote: ↑31 Oct 2020, 04:52 But for example Light Heavyweight is not represented there, although it's level is quite high. But there are no dominating boxers.
In my DD setup, I take a little bit of info from surrounding divisions.. to help smooth things out.
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 13:39
If by high level, you mean the top 10-20 LHWs have on average more points than guys in other weights, maybe someone can check that. However, LHWs aren't on JCS' list because unlike other sites, Boxrec still has Kovalev on top and WHR thinks none of the other elites have a good enough career to rise above him.computerrank wrote: ↑31 Oct 2020, 04:52
But for example Light Heavyweight is not represented there, although it's level is quite high. But there are no dominating boxers.
Why don't you run a hypothetical Bivol-Beterbiev fight and see if the winner gets in? If not, then let's just chalk it up to parity within the division.
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 14:18
My example was simple and quick. If I had query access to the DB, I'd experiment a bit more....
At some point, the depth of a division becomes less of a factor..... I think it is a bit unfair to compare the #1 @ 105 to the #3 boxer... while the top boxers at popular weights were compared to guys outside the Top 20, even though 105 has 1/8th the amount of boxers.... there are still 300+ boxers in the division. This is probably something that could benefit from a bit more logic.
There's still the issue of the sudden division change damaging a boxer's ranking... so perhaps a mechanism that'd still compare Usyk's rating to the Cruiserweight landscape until he spends sufficient time at Heavyweight is probably another thing to think about.
"P4P" is the kind of ranking where you just mess w/ it until it feels right.
At some point, the depth of a division becomes less of a factor..... I think it is a bit unfair to compare the #1 @ 105 to the #3 boxer... while the top boxers at popular weights were compared to guys outside the Top 20, even though 105 has 1/8th the amount of boxers.... there are still 300+ boxers in the division. This is probably something that could benefit from a bit more logic.
There's still the issue of the sudden division change damaging a boxer's ranking... so perhaps a mechanism that'd still compare Usyk's rating to the Cruiserweight landscape until he spends sufficient time at Heavyweight is probably another thing to think about.
"P4P" is the kind of ranking where you just mess w/ it until it feels right.
Re: Ratings - please read before commenting
Posted: 31 Oct 2020, 20:00
Why not combine 105 with 108 boxers? What's 3 lbs compared the 40 lb difference bet HW and cruiser?JCS wrote: ↑31 Oct 2020, 14:18 At some point, the depth of a division becomes less of a factor..... I think it is a bit unfair to compare the #1 @ 105 to the #3 boxer... while the top boxers at popular weights were compared to guys outside the Top 20, even though 105 has 1/8th the amount of boxers.... there are still 300+ boxers in the division. This is probably something that could benefit from a bit more logic.
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:17
What happens to the ratings? I wrote earlier that the current rating formula gives a negative result for points. Almost all fights end with a negative point value of the fighters.
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:21
This fight was Jared Anderson's debut:
https://boxrec.com/en/event/795355/2400541
his opponent's current rating was 0.01 points, and Anderson's rating after the fight was 8,520 points ??? The question is how ?????
https://boxrec.com/en/event/795355/2400541
his opponent's current rating was 0.01 points, and Anderson's rating after the fight was 8,520 points ??? The question is how ?????
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:25
Don't know if that happened, but it's 8.52 pts now. Why he got so much, I don't know either.mike1989 wrote: ↑01 Nov 2020, 12:21 This fight was Jared Anderson's debut:
https://boxrec.com/en/event/795355/2400541
his opponent's current rating was 0.01 points, and Anderson's rating after the fight was 8,520 points ??? The question is how ?????
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:29
https://boxrec.com/en/event/817046/2558094
in this fight, Kamposos's point before the fight was 64.69 points after 60.19 points; point Selby before the fight 120.9 points after the fight 77.76 points. Kamposos won the fight and his current rating is below the defeated Selby. How ????
in this fight, Kamposos's point before the fight was 64.69 points after 60.19 points; point Selby before the fight 120.9 points after the fight 77.76 points. Kamposos won the fight and his current rating is below the defeated Selby. How ????
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:30
Did you look at the rankings?mike1989 wrote: ↑01 Nov 2020, 12:29 https://boxrec.com/en/event/817046/2558094
in this fight, Kamposos's point before the fight was 64.69 points after 60.19 points; point Selby before the fight 120.9 points after the fight 77.76 points. Kamposos won the fight and his current rating is below the defeated Selby. How ????
https://boxrec.com/en/ratings?role=prob ... ightweight
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:31
The fights after that? do you guys understand WHR yet?mike1989 wrote: ↑01 Nov 2020, 12:21 This fight was Jared Anderson's debut:
https://boxrec.com/en/event/795355/2400541
his opponent's current rating was 0.01 points, and Anderson's rating after the fight was 8,520 points ??? The question is how ?????
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:35
If you mean both lose points no matter the results, it's because the points get spread out over both fighters' careers. Then there's decay for previous inactivity or penalty for lack of quality opponents etc etc. The rating formula is too complex and needs a supercomputer or 2 just to show you how each bout impacted an individual boxer's rating on the day it happened.
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 12:47
Ratings and rankings are different in WHR. The formula thinks Kambosos hasn't done enough in his career to have more ratings points than Selby. However, I think BoxRec has a policy so that the winner is placed over the loser for a year in the divisional rankings. So Selby has 77.76 points at this point in his career and should be 7th behind Kambosos' 79.52 but somehow is listed 9th with 59.15 pts.mike1989 wrote: ↑01 Nov 2020, 12:29 https://boxrec.com/en/event/817046/2558094
in this fight, Kamposos's point before the fight was 64.69 points after 60.19 points; point Selby before the fight 120.9 points after the fight 77.76 points. Kamposos won the fight and his current rating is below the defeated Selby. How ????
Why WHR switches their points, I have no idea. I would've thought Kambosos would just get some bonus points. Seems very unfair to me that Selby is getting penalized twice for losing. And not only that. It's misleading the public if they try to calculate odds of future Selby bouts within the year using the rankings list.
Re: Ratings - please read before commenting
Posted: 01 Nov 2020, 15:02
https://boxrec.com/en/proboxer/646781JCS wrote: ↑01 Nov 2020, 12:30Did you look at the rankings?mike1989 wrote: ↑01 Nov 2020, 12:29 https://boxrec.com/en/event/817046/2558094
in this fight, Kamposos's point before the fight was 64.69 points after 60.19 points; point Selby before the fight 120.9 points after the fight 77.76 points. Kamposos won the fight and his current rating is below the defeated Selby. How ????
https://boxrec.com/en/ratings?role=prob ... ightweight
and then what ???