Yes, 10% of max career rating converted to division. Before I had 10 points fix.
Ratings - please read before commenting - Archived
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Interesting.. do you think this is the primary reason for the prediction rate improvement or something else?computerrank wrote: ↑09 Jan 2019, 16:14Yes, 10% of max career rating converted to division. Before I had 10 points fix.
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
No, this is only a side effect.JCS wrote: ↑09 Jan 2019, 17:04Interesting.. do you think this is the primary reason for the prediction rate improvement or something else?computerrank wrote: ↑09 Jan 2019, 16:14Yes, 10% of max career rating converted to division. Before I had 10 points fix.
The main reason is to have eliminated all factors depending on the Boxrec ranking at bout time. In earlier ratings the Boxrec ranking dependent factors improved the rating.
Now with the improved concept of an individual time period of strong results for every boxer, depending on strong results between the members, these additional rankings points are no longer needed and even counter-productive.
Re: Boxrec Ratings - Read first before commenting on the ratings
Hi Martin and what has changed in the current rating and all-time rating ???
Re: Boxrec Ratings - Read first before commenting on the ratings
Why does Entony Joshua So High in an all time ????
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Look here in this thread:
viewtopic.php?f=22&t=28051&p=5080046#p5079858
The algorithm for the all-time ratings didn't change.
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Johua didn't get to much extra points with the new release - from 380 to 442 points. But others are rated lower now:. Compare old and new points:
Code: Select all
1 Heavyweight 9027 Joe Louis US 2027.64 1388.09
2 Heavyweight 180 Muhammad Ali US 1824.35 1708.38
3 Heavyweight 1187 Jack Johnson US 1670.57 1192.39
4 Heavyweight 499 Evander Holyfield US 1560.26 1383.53
5 Heavyweight 11023 Sam Langford CA 1486.23 1100.59
6 Heavyweight 9032 Rocky Marciano US 1440.54 1061.94
7 Heavyweight 9012 Ezzard Charles US 1410.87 1256.88
8 Heavyweight 9046 Gene Tunney US 1269.03 980.57
9 Heavyweight 9038 Floyd Patterson US 1240.85 1176.56
10 Heavyweight 9041 Max Schmeling DE 1198.12 966.04
11 Heavyweight 17615 Harry Wills US 1194.52 934.93
12 Heavyweight 12077 Max Baer US 1119.18 849.57
13 Heavyweight 10616 Jack Sharkey US 1062.32 768.25
14 Heavyweight 7035 Wladimir Klitschko UA 1010.07 1029.18
15 Heavyweight 150 Larry Holmes US 942.13 941.55
16 Heavyweight 11622 Sam McVea US 909.52 656.39
17 Heavyweight 10744 Tommy Burns CA 908.19 654.23
18 Heavyweight 9022 James J. Jeffries US 895.40 156.48
19 Heavyweight 147 Joe Frazier US 894.78 779.80
20 Heavyweight 10817 Philadelphia Jack O'Brien US 829.89 557.39
21 Heavyweight 474 Mike Tyson US 755.81 622.42
22 Heavyweight 9031 Sonny Liston US 742.16 544.16
23 Heavyweight 90 George Foreman US 718.76 768.17
24 Heavyweight 11627 Marvin Hart US 714.03 465.52
25 Heavyweight 10799 Peter Maher IE 709.64 70.98
26 Heavyweight 11641 Fireman Jim Flynn US 670.88 403.57
27 Heavyweight 11028 Jersey Joe Walcott US 643.28 528.84
28 Heavyweight 10784 Frank Slavin AU 627.02 6.69
29 Heavyweight 1853 Lennox Lewis UK 623.32 697.45
30 Heavyweight 9009 Jack Dempsey US 621.31 497.69
31 Heavyweight 11631 Joe Jeannette US 597.94 530.08
32 Heavyweight 10543 Peter Jackson AU 593.86 85.88
33 Heavyweight 168 Ken Norton US 551.09 549.44
34 Heavyweight 12711 Ingemar Johansson SE 532.68 482.88
35 Heavyweight 1640 Riddick Bowe US 521.89 441.69
36 Heavyweight 7033 Vitali Klitschko UA 517.85 418.72
37 Heavyweight 32590 Patsy Cardiff US 491.99 4.17
38 Heavyweight 12126 Arturo Godoy CL 476.99 316.07
39 Heavyweight 40331 Pat Killen US 468.41 4.72
40 Heavyweight 12072 Jim Braddock US 452.31 412.84
41 Heavyweight 10526 Melio Bettina US 452.12 415.19
42 Heavyweight 12087 Walter Neusel DE 450.45 430.69
43 Heavyweight 12086 Primo Carnera IT 426.28 357.42
44 Heavyweight 405 Mike Weaver US 418.90 342.71
45 Heavyweight 251 Michael Dokes US 405.89 308.91
46 Heavyweight 155774 David Haye UK 395.60 485.30
47 Heavyweight 479205 Tyson Fury UK 395.48 425.01
48 Heavyweight 40145 Joe McAuliffe US 395.47 5.38
49 Heavyweight 10551 Tom Sharkey IE 389.98 72.11
50 Heavyweight 4655 John Ruiz US 382.54 266.36
51 Heavyweight 659461 Anthony Joshua UK 380.60 442.37
Re: Boxrec Ratings - Read first before commenting on the ratings
It turns out that Entony Joshua for Povetkin and Klitschko received more all-time rating than Usik and Lomachenko ??? Why ??? Can you give an example ???computerrank wrote: ↑09 Jan 2019, 19:07Johua didn't get to much extra points with the new release - from 380 to 442 points. But others are rated lower now:. Compare old and new points:
Code: Select all
1 Heavyweight 9027 Joe Louis US 2027.64 1388.09 2 Heavyweight 180 Muhammad Ali US 1824.35 1708.38 3 Heavyweight 1187 Jack Johnson US 1670.57 1192.39 4 Heavyweight 499 Evander Holyfield US 1560.26 1383.53 5 Heavyweight 11023 Sam Langford CA 1486.23 1100.59 6 Heavyweight 9032 Rocky Marciano US 1440.54 1061.94 7 Heavyweight 9012 Ezzard Charles US 1410.87 1256.88 8 Heavyweight 9046 Gene Tunney US 1269.03 980.57 9 Heavyweight 9038 Floyd Patterson US 1240.85 1176.56 10 Heavyweight 9041 Max Schmeling DE 1198.12 966.04 11 Heavyweight 17615 Harry Wills US 1194.52 934.93 12 Heavyweight 12077 Max Baer US 1119.18 849.57 13 Heavyweight 10616 Jack Sharkey US 1062.32 768.25 14 Heavyweight 7035 Wladimir Klitschko UA 1010.07 1029.18 15 Heavyweight 150 Larry Holmes US 942.13 941.55 16 Heavyweight 11622 Sam McVea US 909.52 656.39 17 Heavyweight 10744 Tommy Burns CA 908.19 654.23 18 Heavyweight 9022 James J. Jeffries US 895.40 156.48 19 Heavyweight 147 Joe Frazier US 894.78 779.80 20 Heavyweight 10817 Philadelphia Jack O'Brien US 829.89 557.39 21 Heavyweight 474 Mike Tyson US 755.81 622.42 22 Heavyweight 9031 Sonny Liston US 742.16 544.16 23 Heavyweight 90 George Foreman US 718.76 768.17 24 Heavyweight 11627 Marvin Hart US 714.03 465.52 25 Heavyweight 10799 Peter Maher IE 709.64 70.98 26 Heavyweight 11641 Fireman Jim Flynn US 670.88 403.57 27 Heavyweight 11028 Jersey Joe Walcott US 643.28 528.84 28 Heavyweight 10784 Frank Slavin AU 627.02 6.69 29 Heavyweight 1853 Lennox Lewis UK 623.32 697.45 30 Heavyweight 9009 Jack Dempsey US 621.31 497.69 31 Heavyweight 11631 Joe Jeannette US 597.94 530.08 32 Heavyweight 10543 Peter Jackson AU 593.86 85.88 33 Heavyweight 168 Ken Norton US 551.09 549.44 34 Heavyweight 12711 Ingemar Johansson SE 532.68 482.88 35 Heavyweight 1640 Riddick Bowe US 521.89 441.69 36 Heavyweight 7033 Vitali Klitschko UA 517.85 418.72 37 Heavyweight 32590 Patsy Cardiff US 491.99 4.17 38 Heavyweight 12126 Arturo Godoy CL 476.99 316.07 39 Heavyweight 40331 Pat Killen US 468.41 4.72 40 Heavyweight 12072 Jim Braddock US 452.31 412.84 41 Heavyweight 10526 Melio Bettina US 452.12 415.19 42 Heavyweight 12087 Walter Neusel DE 450.45 430.69 43 Heavyweight 12086 Primo Carnera IT 426.28 357.42 44 Heavyweight 405 Mike Weaver US 418.90 342.71 45 Heavyweight 251 Michael Dokes US 405.89 308.91 46 Heavyweight 155774 David Haye UK 395.60 485.30 47 Heavyweight 479205 Tyson Fury UK 395.48 425.01 48 Heavyweight 40145 Joe McAuliffe US 395.47 5.38 49 Heavyweight 10551 Tom Sharkey IE 389.98 72.11 50 Heavyweight 4655 John Ruiz US 382.54 266.36 51 Heavyweight 659461 Anthony Joshua UK 380.60 442.37
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
-
pugilisticspecialist
- Heavyweight

- Posts: 252
- Joined: 19 May 2009, 16:23
Re: Boxrec Ratings - Read first before commenting on the ratings
With the new ratings system:
How many points would Usyk have if he moved up to heavyweight?
How many points would Joshua have if he (T)KO'd Usyk on 2019-04-13?
How many points would Usyk have if he moved up to heavyweight?
How many points would Joshua have if he (T)KO'd Usyk on 2019-04-13?
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
- 129pugilisticspecialist wrote: ↑11 Jan 2019, 15:00 With the new ratings system:
How many points would Usyk have if he moved up to heavyweight?
How many points would Joshua have if he (T)KO'd Usyk on 2019-04-13?
- 267
Re: Boxrec Ratings - Read first before commenting on the ratings
The highest rating among all active boxers is 328. Is this expected behavior? Because it's a big drop from past numbers. Is this new change to inactivity penalty stable? Or are ratings going to steadily decline over time?
Related observation: has the star rating system for fights been appropriately adjusted? Because it is currently displaying Pacquiao - Broner, Eleider - Kovalev, and Badou - Browne as 3-star fights. My expectation, based on the star values of the individual fighters, was that these were all 4.5+ star fights.
Related observation: has the star rating system for fights been appropriately adjusted? Because it is currently displaying Pacquiao - Broner, Eleider - Kovalev, and Badou - Browne as 3-star fights. My expectation, based on the star values of the individual fighters, was that these were all 4.5+ star fights.
Re: Boxrec Ratings - Read first before commenting on the ratings
I agree, it is kinda stupid to lower so much the rating that was already good and stable IMO. All January is a new rating form (in the female it was changed 3x+), this way matchmakers go crazy without a decent north, you guys from boxrec can't play with the numbers, ppl may depend on it, with so tight ratings, many boxers would be considered as the same level, when most times it doesn't .Lackeos wrote: ↑12 Jan 2019, 02:04 The highest rating among all active boxers is 328. Is this expected behavior? Because it's a big drop from past numbers. Is this new change to inactivity penalty stable? Or are ratings going to steadily decline over time?
Related observation: has the star rating system for fights been appropriately adjusted? Because it is currently displaying Pacquiao - Broner, Eleider - Kovalev, and Badou - Browne as 3-star fights. My expectation, based on the star values of the individual fighters, was that these were all 4.5+ star fights.
e.g in a (now old star class system) a 'c' class fighter (3 star), now is a 'd' one (2 star) and in this new form, there are thousands of fighters whom are put in the same basket as if all were on the same level when they aren't, you could slightly distinguish who are 'better' then who, and now with so many equal ratings you can't, I don't agree with the new form at all, this is not socialism.
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
- the ratings level dropped - as expectedLackeos wrote: ↑12 Jan 2019, 02:04 The highest rating among all active boxers is 328. Is this expected behavior? Because it's a big drop from past numbers. Is this new change to inactivity penalty stable? Or are ratings going to steadily decline over time?
Related observation: has the star rating system for fights been appropriately adjusted? Because it is currently displaying Pacquiao - Broner, Eleider - Kovalev, and Badou - Browne as 3-star fights. My expectation, based on the star values of the individual fighters, was that these were all 4.5+ star fights.
- it is stable and will not decline over time
- the star ratings levels are adjusted to the new ratings level
In my test production I get 4.5+ star bouts - but Boxrec shows only 3 stars.
I will go into it ...
Re: Boxrec Ratings - Read first before commenting on the ratings
Hey buddy, great work with the new ratings. Any chance the annual ratings get updated to reflect the new algorithm? Also, perhaps 2018 added?
http://boxrec.com/media/index.php/BoxRe ... al_Ratings

http://boxrec.com/media/index.php/BoxRe ... al_Ratings
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
I will wait until end of January. Then the 2018 results will be consolidated.Manrae wrote: ↑12 Jan 2019, 20:18 Hey buddy, great work with the new ratings. Any chance the annual ratings get updated to reflect the new algorithm? Also, perhaps 2018 added?
http://boxrec.com/media/index.php/BoxRe ... al_Ratings
![]()
Re: Boxrec Ratings - Read first before commenting on the ratings
Awesome , looking forward to the updated listscomputerrank wrote: ↑13 Jan 2019, 02:20I will wait until end of January. Then the 2018 results will be consolidated.Manrae wrote: ↑12 Jan 2019, 20:18 Hey buddy, great work with the new ratings. Any chance the annual ratings get updated to reflect the new algorithm? Also, perhaps 2018 added?
http://boxrec.com/media/index.php/BoxRe ... al_Ratings
![]()
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
bug fix with release r_37a
I had to fix a bug in the ratings - now release r_37a launched.
No functional changes.
No functional changes.
Re: Boxrec Ratings - Read first before commenting on the ratings
Martin,
Have you explored boxers like Caleb Plant? This boxer never lost or by the numbers, has not struggled in any of his previous bouts. Always dominant victories, hardly losing rounds. I believe in cases like him and a match like the one against Uzcategui, it seems fair to increase his pre-bout rating a bit as Uzcategui takes a bit of an unnecessary hit, against a boxer that has been rated too low. Both of their current ratings now seem too low.
I'm curious what something like this would do for prediction, but understand it'd be tough to place rules on a situation like this.
Have you explored boxers like Caleb Plant? This boxer never lost or by the numbers, has not struggled in any of his previous bouts. Always dominant victories, hardly losing rounds. I believe in cases like him and a match like the one against Uzcategui, it seems fair to increase his pre-bout rating a bit as Uzcategui takes a bit of an unnecessary hit, against a boxer that has been rated too low. Both of their current ratings now seem too low.
I'm curious what something like this would do for prediction, but understand it'd be tough to place rules on a situation like this.
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Jason, can you specify your suggestion in numbers?JCS wrote: ↑14 Jan 2019, 14:55 Martin,
Have you explored boxers like Caleb Plant? This boxer never lost or by the numbers, has not struggled in any of his previous bouts. Always dominant victories, hardly losing rounds. I believe in cases like him and a match like the one against Uzcategui, it seems fair to increase his pre-bout rating a bit as Uzcategui takes a bit of an unnecessary hit, against a boxer that has been rated too low. Both of their current ratings now seem too low.
I'm curious what something like this would do for prediction, but understand it'd be tough to place rules on a situation like this.
Re: Boxrec Ratings - Read first before commenting on the ratings
Plant isn't the greatest example because his opposition hasn't been outstanding, but he's a very recent example, which is why I picked him.computerrank wrote: ↑14 Jan 2019, 16:23Jason, can you specify your suggestion in numbers?JCS wrote: ↑14 Jan 2019, 14:55 Martin,
Have you explored boxers like Caleb Plant? This boxer never lost or by the numbers, has not struggled in any of his previous bouts. Always dominant victories, hardly losing rounds. I believe in cases like him and a match like the one against Uzcategui, it seems fair to increase his pre-bout rating a bit as Uzcategui takes a bit of an unnecessary hit, against a boxer that has been rated too low. Both of their current ratings now seem too low.
I'm curious what something like this would do for prediction, but understand it'd be tough to place rules on a situation like this.
Tough to cite specifics for a theory.. but..
The following is true.
1 - Boxer has won all his matches with dominant decisions. Maybe you allow exemptions in certain cases.. like debuts or against boxers in a similar position. In cases like this, we can use none of his prior wins as ceilings... If he had close wins, we could and this is void.
2 - Boxer beats higher-rated opponent.
I'm unsure if Hernandez was rated higher than Plant... so let's ignore that one. With the Uzcategui fight, since Plant fits #1 and #2, we play a little pre-bout catchup as it is a system limitation... the system only "sees" the result, not the fight.. right? A UD 118-110 could be competitive, or it could be dominant but on away turf. The system doesn't know either way.
For the Uzcategui fight now..
Plant 32 (W) Uzcategui 71
This, I think.. is where the tinkering enters. You could suggest "levels" to this.. in other words, how many times has the early career dominant boxer proved the system wrong?
What if we gave Plant half the difference.. and Plant enters the fight at ~51.5 points instead of 32?
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Half the difference - we will see ...JCS wrote: ↑14 Jan 2019, 17:15Plant isn't the greatest example because his opposition hasn't been outstanding, but he's a very recent example, which is why I picked him.computerrank wrote: ↑14 Jan 2019, 16:23Jason, can you specify your suggestion in numbers?JCS wrote: ↑14 Jan 2019, 14:55 Martin,
Have you explored boxers like Caleb Plant? This boxer never lost or by the numbers, has not struggled in any of his previous bouts. Always dominant victories, hardly losing rounds. I believe in cases like him and a match like the one against Uzcategui, it seems fair to increase his pre-bout rating a bit as Uzcategui takes a bit of an unnecessary hit, against a boxer that has been rated too low. Both of their current ratings now seem too low.
I'm curious what something like this would do for prediction, but understand it'd be tough to place rules on a situation like this.
Tough to cite specifics for a theory.. but..
The following is true.
1 - Boxer has won all his matches with dominant decisions. Maybe you allow exemptions in certain cases.. like debuts or against boxers in a similar position. In cases like this, we can use none of his prior wins as ceilings... If he had close wins, we could and this is void.
2 - Boxer beats higher-rated opponent.
I'm unsure if Hernandez was rated higher than Plant... so let's ignore that one. With the Uzcategui fight, since Plant fits #1 and #2, we play a little pre-bout catchup as it is a system limitation... the system only "sees" the result, not the fight.. right? A UD 118-110 could be competitive, or it could be dominant but on away turf. The system doesn't know either way.
For the Uzcategui fight now..
Plant 32 (W) Uzcategui 71
This, I think.. is where the tinkering enters. You could suggest "levels" to this.. in other words, how many times has the early career dominant boxer proved the system wrong?
What if we gave Plant half the difference.. and Plant enters the fight at ~51.5 points instead of 32?
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
The ratings simply grow to infinity very quickly ...
Re: Boxrec Ratings - Read first before commenting on the ratings
Interesting.. seems like it'd work its way out, but w/ some minor-moderate level of inflation.computerrank wrote: ↑14 Jan 2019, 18:46The ratings simply grow to infinity very quickly ...![]()
You restricted it for those who only have dominant wins and only in bouts vs. higher rated opposition (where they also won)? Interesting..
-
computerrank
- Editor

- Posts: 2492
- Joined: 04 Jan 2003, 18:59
Re: Boxrec Ratings - Read first before commenting on the ratings
Yes, restricted to stopped bouts and UD with at least 10 rounds and vs higher rated opposition.JCS wrote: ↑14 Jan 2019, 20:22Interesting.. seems like it'd work its way out, but w/ some minor-moderate level of inflation.
You restricted it for those who only have dominant wins and only in bouts vs. higher rated opposition (where they also won)? Interesting..
Only, when I restrict to stopped bouts, the ratings are at least stable. But the prediction success is 2 percent points lower.