Ratings - please read before commenting - Archived

[computerrank]

Wladimir Klitschko in 2012 in annual not rank No. 2 and No. 1 the downgrade it happened 13.01.02. not 12.12.31. from 11.07.02. 1 year 6 months after the fight with David Hay

For computational reasons all dates are converted to Universal Time = days since Universal Time on January 1, 4713 BCE

julian date Haye bout = julian date (2011-07-02) = 2455745

julian date end of 2012 = julian date (2012-12-31) = 2456293

difference = 548 days

1.5 years = 365.24 days per year * 1.5 year = 547.86 days

So the Haye bout lost validity exactly on 2012-12-31 - and Wladimir Klitschko's rating dropped.

The second of the year has more days than the first one and in addition 2012 was a leap year.

[computerrank]

I see HWs took a hit in their absolute current rating points, which made it worse for them in the p4p ratings. But the numbers were already very low if we take a look at the past.

Does this influence the all time rankings in the part of 14% of career top rating? I never got how this all-time works, ie. #1 boxer beat #2 in the year and got additional (200/1)*33% + (200/2)*33% for the year? It means 99 points maximum for the year? I remember maximum of 200 points per year mentioned by Computerrank. How is that?
And if 14% of career top rating means 14% * highest number of active rating point, then Marchiano gets 14%* 5705 = 800 points just for that, plus 500 for his entire carreer (5 years at the top * 99 maximum per year). And Wlad gets more like 1000 for his carrear plus only 133 for 14% of 953 (carreer top) to arrive at lower than Rocky in total?

example for a boxer:

2014: #5 at his division cluster at end of the year, his best not lost to opponent was #9 at time of bout
2015: #2 at his division cluster at end of the year, his best not lost to opponent was #6 at time of bout

His highest career rating was 300.

- ranking points: 200/5 = 40 for 2014, 200/2 = 100 for 2015 - sum is 40+100 = 140 - 33% go into the all time rating = 0.33 *140 = 47 points

- opponent rank points: 200/9 = 22 for 2014, 200/6 = 33 for 2015 - sum is 22+33 = 55 - 33% go into the all time rating = 0.33 *55 = 18 points

- career top rating: 300 - 14% go into the all time rating = 0.14*300 = 42 points


All time rating = 47 + 18 + 42 = 107 points.

-

[computerrank]

all right, when you update the rankings of all-time ???

I will first have a look into some other issues. As soon as the corrected version is completed and launched, I will post a message here ... thanks a lot

New release is active now ...

[Leonid]

I see HWs took a hit in their absolute current rating points, which made it worse for them in the p4p ratings. But the numbers were already very low if we take a look at the past.

Does this influence the all time rankings in the part of 14% of career top rating? I never got how this all-time works, ie. #1 boxer beat #2 in the year and got additional (200/1)*33% + (200/2)*33% for the year? It means 99 points maximum for the year? I remember maximum of 200 points per year mentioned by Computerrank. How is that?
And if 14% of career top rating means 14% * highest number of active rating point, then Marchiano gets 14%* 5705 = 800 points just for that, plus 500 for his entire carreer (5 years at the top * 99 maximum per year). And Wlad gets more like 1000 for his carrear plus only 133 for 14% of 953 (carreer top) to arrive at lower than Rocky in total?

example for a boxer:

2014: #5 at his division cluster at end of the year, his best not lost to opponent was #9 at time of bout
2015: #2 at his division cluster at end of the year, his best not lost to opponent was #6 at time of bout

His highest career rating was 300.

- ranking points: 200/5 = 40 for 2014, 200/2 = 100 for 2015 - sum is 40+100 = 140 - 33% go into the all time rating = 0.33 *140 = 47 points

- opponent rank points: 200/9 = 22 for 2014, 200/6 = 33 for 2015 - sum is 22+33 = 55 - 33% go into the all time rating = 0.33 *55 = 18 points

- career top rating: 300 - 14% go into the all time rating = 0.14*300 = 42 points


All time rating = 47 + 18 + 42 = 107 points.

-

How do you get 300 (career top rating) in the example and what's the maximum here? Is it the highest of any current ratings at any point or of any year's highest total of own and opponent ranking points? I'd much rather see it on the example of Marciano's carreer, since I don't get how he got to 1310 all-time if the most one can add to his all-time in 1 year seems to be 100 points (200/1*33%+200/2*33%) and his carreer was quite short.
And thanks for taking time to make explainations in this thread))

[computerrank]

How do you get 300 (career top rating) in the example and what's the maximum here? Is it the highest of any current ratings at any point or of any year's highest total of own and opponent ranking points? I'd much rather see it on the example of Marciano's carreer, since I don't get how he got to 1310 all-time if the most one can add to his all-time in 1 year seems to be 100 points (200/1*33%+200/2*33%) and his carreer was quite short.
And thanks for taking time to make explainations in this thread))

- career top rating is the highest value of the current rating at any time of the career

- Marciano had a top rating of 5706 points after the last bout of his career (other boxers show a career high earlier in their career and later drop)

So he gets 14% of 5706 = 799 points for this

- for every year's end at top of his division cluster he gets 200/1*0.33 = 67 points

- for every year with a not lost bout against #1 he gets 200/1*0.33 = 67 points. (To be precise here. The divisor is 1 for a #1 AND ALSO for a #2 opponent - and the divisior is n-1 for a #n opponent - this compensates for the situation, when the boxer is #1 and so cannot fight #1)

- so the rest of his 511 points come from these up to 133 points per year. I wll add a table with his annual standings.

[computerrank]

... and here is Rocky Marciano:

Code: Select all

Heavyweight          1951  2082   352    28    33   291   9032 Rocky Marciano
Heavyweight          1952  3512   684    94    99   492   9032 Rocky Marciano
Heavyweight          1953  3999   841   160   121   560   9032 Rocky Marciano
Heavyweight          1954  4027   943   226   154   564   9032 Rocky Marciano
Heavyweight          1955  5706  1311   292   220   799   9032 Rocky Marciano

- the column with 799 at the end - career top rating
- the column with 220 at the end - accumulated points for top opponent ranks
- the column with 292 at the end - accumulated points for top year's end ranks

the numbers are rounded here a bit different in this excerpt

[Leonid]

... and here is Rocky Marciano:

Code: Select all

Heavyweight          1951  2082   352    28    33   291   9032 Rocky Marciano
Heavyweight          1952  3512   684    94    99   492   9032 Rocky Marciano
Heavyweight          1953  3999   841   160   121   560   9032 Rocky Marciano
Heavyweight          1954  4027   943   226   154   564   9032 Rocky Marciano
Heavyweight          1955  5706  1311   292   220   799   9032 Rocky Marciano

- the column with 799 at the end - career top rating
- the column with 220 at the end - accumulated points for top opponent ranks
- the column with 292 at the end - accumulated points for top year's end ranks

the numbers are rounded here a bit different in this excerpt

Thanx. This is actually how I got it from the beginning, but I though I must have misinterpreted smthn, cause this seems really unfair. Rocky got more than 60% of his all-time rating just for being #1 in a division, where everyone at HW had sky-high current ratings. HW boxers had almost 10! times more points int their "active" ratings than they have today (#1 Marciano had 5706, while #1 Fury has 650), which means that division must have been 10 times stronger than the current scene. Wlad never had a chance to get anywhere near those kind of points for his career top rating no matter who he beat. Imagine Wlad had a chance to beat old middleweight Archie Moore and get his points. That missmatch would be just awfull, but the result would bring Wlad hundreds of additional all-time points and lift him up many places in the computerised all-time standings (both HW and p4p).

Anyways, "active" ratings of past and modern divisions differ way too much to be factored in all-time ratings, 50s were not the best HW era as it appears to be according to such ratings, but one of the weekest (maybe due to being post-war), therefore I'd suggest to remove this 14% of carreer top ranking from the formula at all.

[Leonid]

... and here is Rocky Marciano:

Code: Select all

Heavyweight          1951  2082   352    28    33   291   9032 Rocky Marciano
Heavyweight          1952  3512   684    94    99   492   9032 Rocky Marciano
Heavyweight          1953  3999   841   160   121   560   9032 Rocky Marciano
Heavyweight          1954  4027   943   226   154   564   9032 Rocky Marciano
Heavyweight          1955  5706  1311   292   220   799   9032 Rocky Marciano

- the column with 799 at the end - career top rating
- the column with 220 at the end - accumulated points for top opponent ranks
- the column with 292 at the end - accumulated points for top year's end ranks

the numbers are rounded here a bit different in this excerpt

Thanx. This is actually how I got it from the beginning, but I though I must have misinterpreted smthn, cause this seems really unfair. Rocky got more than 60% of his all-time rating just for being #1 in a division, where everyone at HW had sky-high current "active" ratings. HW boxers had almost 10! times more points int their "active" ratings than they have today (#1 Marciano had 5706, while #1 Fury has 650), which means that division must have been 10 times stronger than the current scene. Wlad never had a chance to get anywhere near those kind of points for his career top rating no matter who he beat. Imagine Wlad had a chance to beat old middleweight Archie Moore and get his points. That missmatch would be just awfull, but the result would bring Wlad hundreds of additional all-time points and lift him up many places in the computerised all-time standings (both HW and p4p).

Anyways, "active" ratings of past and modern divisions differ way too much to be factored in all-time ratings, 50s were not the best HW era as it appears to be according to such ratings, but one of the weekest (maybe due to being post-war), therefore I'd suggest to remove this 14% of carreer top ranking from the formula at all.

[Leonid]

If you want an alternative factor to reflect carreer-top achievement, maybe use smthn like 14% of boxer's best year's all-time points earned; or divide carreer top rating by top-10 average rating at that moment or for that year (though it would be a challenge to impement that formula technically).

In order to factor in strenth of a division simply a number of active fighters would suffice (eg. instead of 200/annual rank, use 10% of # of fighters in the division or cluster in certain year / annual rank).

[mike1989]

New release is active now ...[/quote]

you can reset updated rating Holyfield and Wladimir Klitschko???
the best anual the rating of the opposition is assessed 1 best rival a year ?

[mike1989]

Was Bryant Jennings in the top 20 at the time of his fight with Klitschko ) :
annual rating for 2014 No. 15 Deontay Wilder 191 point, Jennings 190 point
annual the Cruiserweight rating for 2014 + 4 fighter above Jennings
before the fight with Klitschko-Jennings in 2015: Christian Hammer 141 point; Odlanier Solis 135 points; Steve Cunningham 182 point; Tony Thompson 242 points, and -3 +1 ; Cruiserweight changes have occurred influencing the rating of Jennings, and a rating of Jennings top 18-20 at the time of his fight with Klitschko ??? )

[computerrank]

Was Bryant Jennings in the top 20 at the time of his fight with Klitschko ) :
annual rating for 2014 No. 15 Deontay Wilder 191 point, Jennings 190 point
annual the Cruiserweight rating for 2014 + 4 fighter above Jennings
before the fight with Klitschko-Jennings in 2015: Christian Hammer 141 point; Odlanier Solis 135 points; Steve Cunningham 182 point; Tony Thompson 242 points, and -3 +1 ; Cruiserweight changes have occurred influencing the rating of Jennings, and a rating of Jennings top 18-20 at the time of his fight with Klitschko ??? )

at time of bout - where is Jennings ???

Jenning's 190 was outside top 20 at time of his last bout - but now he should be in ... this widens the problem ...

Code: Select all

1  7035               Wladimir Klitschko   807
2  318081             Alexander Povetkin   388
3  280342                     Marco Huck   382
4  479205                     Tyson Fury   374
5  511850                   Kubrat Pulev   347
6  496178             Vyacheslav Glazkov   327
7  324252           Yoan Pablo Hernandez   320
8  38558                   Grigory Drozd   301
9  468841                 Deontay Wilder   293
10 438266                 Thabiso Mchunu   292
11 34020                   Denis Lebedev   245
12 21633                   Tony Thompson   242
13 425328                    Tony Bellew   218
14 14043                  Antonio Tarver   211
15 178679                 Ovill McKenzie   189
16 36642                Steve Cunningham   183
17 508480               Rakhim Chakhkiev   180
18 355273                   Carlos Takam   179
19 536325 William Fernando Souza Bezerra   177
20 489762                   Andy Ruiz Jr   173

[computerrank]

When I take the top 40, Jennings is in at #15 at time of the bout -

I will take in all above a minimum value ...

Code: Select all

1  7035               Wladimir Klitschko   807
2  318081             Alexander Povetkin   388
3  280342                     Marco Huck   382
4  479205                     Tyson Fury   374
5  511850                   Kubrat Pulev   347
6  496178             Vyacheslav Glazkov   327
7  324252           Yoan Pablo Hernandez   320
8  38558                   Grigory Drozd   301
9  468841                 Deontay Wilder   293
10 438266                 Thabiso Mchunu   292
11 34020                   Denis Lebedev   245
12 21633                   Tony Thompson   242
13 425328                    Tony Bellew   218
14 14043                  Antonio Tarver   211
15 529630                Bryant Jennings   190
16 178679                 Ovill McKenzie   189
17 36642                Steve Cunningham   183
18 14587                    Firat Arslan   182
19 508480               Rakhim Chakhkiev   180
20 355273                   Carlos Takam   179
21 536325 William Fernando Souza Bezerra   177
22 17752            Krzysztof Wlodarczyk   175
23 324239                Nathan Cleverly   175
24 489762                   Andy Ruiz Jr   173
25 579033              Dmitry Kudryashov   172
26 438253                     Mike Perez   166
27 473857                  Artur Szpilka   166
28 381739                 Dereck Chisora   165
29 659772                 Oleksandr Usyk   163
30 545749                    Erkan Teper   161
31 324806               Bermane Stiverne   158
32 14883                   Tomasz Adamek   152
33 489046                   Lucas Browne   151
34 24358                  Ruslan Chagaev   151
35 613846                  Joseph Parker   143
36 475159               Christian Hammer   141
37 507154                Yunier Dorticos   141
38 352748          Victor Emilio Ramirez   138
39 530410         Youri Kayembre Kalenga   132
40 583767                  Murat Gassiev   132

[mike1989]

at time of bout - where is Jennings ???

Jenning's 190 was outside top 20 at time of his last bout - but now he should be in ... this widens the problem ...

Code: Select all

1  7035               Wladimir Klitschko   807
2  318081             Alexander Povetkin   388
3  280342                     Marco Huck   382
4  479205                     Tyson Fury   374
5  511850                   Kubrat Pulev   347
6  496178             Vyacheslav Glazkov   327
7  324252           Yoan Pablo Hernandez   320
8  38558                   Grigory Drozd   301
9  468841                 Deontay Wilder   293
10 438266                 Thabiso Mchunu   292
11 34020                   Denis Lebedev   245
12 21633                   Tony Thompson   242
13 425328                    Tony Bellew   218
14 14043                  Antonio Tarver   211
15 178679                 Ovill McKenzie   189
16 36642                Steve Cunningham   183
17 508480               Rakhim Chakhkiev   180
18 355273                   Carlos Takam   179
19 536325 William Fernando Souza Bezerra   177
20 489762                   Andy Ruiz Jr   173

[/quote]

where coefficient decrease for cruisers? according to the table above in the top 20 fighters rated below 190 points at the time of his fight with Klitschko Jennings was 190 points ????
http://boxrec.com/media/index.php?title=Fight:1942392

[mike1989]

[quote="computerrank"]When I take the top 40, Jennings is in at #15 at time of the bout -

I will take in all above a minimum value ...

I do not quite understand ??? ))))

[computerrank]

When I take the top 40, Jennings is in at #15 at time of the bout -

I will take in all above a minimum value ...

I do not quite understand ??? ))))

I will include all boxers with a post-bout rating above 100 independet of the length of the list.

In the current release the list is limited to the top 20, and the check is with the post-bout rating of a boxer at time of this bout.

When Jennings got his 190 points, he was outside the top 20.

At time of his Klitschko bout he would have be in with these 190 points ...

[mike1989]

When I take the top 40, Jennings is in at #15 at time of the bout -

I will include all boxers with a post-bout rating above 100 independet of the length of the list.

In the current release the list is limited to the top 20, and the check is with the post-bout rating of a boxer at time of this bout.

When Jennings got his 190 points, he was outside the top 20.

At time of his Klitschko bout he would have be in with these 190 points ...

so Klitschko + points for Jennings opponent rank points accumulated ???

[computerrank]

...so Klitschko + points for Jennings opponent rank points accumulated ???

Yes, Jennings will be included ... and now looks like this - 5 points added for Jennings - but lost some points due to further corrections before ...

Code: Select all

Heavyweight          1999   176    30     5     0    25   7035 Wladimir Klitschko
Heavyweight          2000   405   108    38    13    57   7035 Wladimir Klitschko
Heavyweight          2001   431   145    71    13    60   7035 Wladimir Klitschko
Heavyweight          2002   467   191   104    21    65   7035 Wladimir Klitschko
Heavyweight          2003   303   199   112    21    65   7035 Wladimir Klitschko
Heavyweight          2005   313   231   127    39    65   7035 Wladimir Klitschko
Heavyweight          2006   543   374   193   105    76   7035 Wladimir Klitschko
Heavyweight          2007   612   462   259   112    90   7035 Wladimir Klitschko
Heavyweight          2008   720   549   325   123   101   7035 Wladimir Klitschko
Heavyweight          2009   872   702   391   189   122   7035 Wladimir Klitschko
Heavyweight          2010   793   788   457   197   133   7035 Wladimir Klitschko
Heavyweight          2011   855   887   523   230   133   7035 Wladimir Klitschko
Heavyweight          2012   690   925   556   235   133   7035 Wladimir Klitschko
Heavyweight          2013   720  1024   622   268   133   7035 Wladimir Klitschko
Heavyweight          2014   807  1156   688   334   133   7035 Wladimir Klitschko
Heavyweight          2015   590  1194   721   339   133   7035 Wladimir Klitschko

[mike1989]

[quote="computerrank"]

Yes, Jennings will be included ... and now looks like this - 5 points added for Jennings - but lost some points due to further corrections before ...


and beyond that, Klitschko took the points ???

[computerrank]

Yes, Jennings will be included ... and now looks like this - 5 points added for Jennings - but lost some points due to further corrections before ...


and beyond that, Klitschko took the points ???

These are values from the test system - on the official server you will see the changes after a new release will be activated.

[computerrank]

If you want an alternative factor to reflect carreer-top achievement, maybe use smthn like 14% of boxer's best year's all-time points earned; or divide carreer top rating by top-10 average rating at that moment or for that year (though it would be a challenge to impement that formula technically).

In order to factor in strenth of a division simply a number of active fighters would suffice (eg. instead of 200/annual rank, use 10% of # of fighters in the division or cluster in certain year / annual rank).

Just to show, that dependencies are NOT as simple as expected - here a diagram from 1900 to 2015 (2 years time period each)

- maximum rating of a boxer in time period
- boxers winning in time period
- bouts in time period

diagram

[Leonid]

If you want an alternative factor to reflect carreer-top achievement, maybe use smthn like 14% of boxer's best year's all-time points earned; or divide carreer top rating by top-10 average rating at that moment or for that year (though it would be a challenge to impement that formula technically).

In order to factor in strenth of a division simply a number of active fighters would suffice (eg. instead of 200/annual rank, use 10% of # of fighters in the division or cluster in certain year / annual rank).

Just to show, that dependencies are NOT as simple as expected - here a diagram from 1900 to 2015 (2 years time period each)

- maximum rating of a boxer in time period
- boxers winning in time period
- bouts in time period

diagram

So this means that maximum rating of a boxer in a time period does not correlate well with the number of boxers/bouts in time period, right? That's my point - it doesn't and IMO number of boxers is a better factor of a strenth of division.

But its not my main point, main point is that the strength of the division should not be factored in all-time ratings as heavily as it is now (+ 14% of carreer high rating). Here's a supporting diagram:

[mike1989]

and the rating Holyfield can lose ???

[computerrank]

... So this means that maximum rating of a boxer in a time period does not correlate well with the number of boxers/bouts in time period, right? That's my point - it doesn't and IMO number of boxers is a better factor of a strenth of division.

But its not my main point, main point is that the strength of the division should not be factored in all-time ratings as heavily as it is now (+ 14% of carreer high rating). ...

(1) The number of boxers in a time period is NOT a good measure for the top ratings in a time period.

(2) It looks like the top rating in a time period correlates with the number of boxers in the time periods before ... with some lag of 10 - 15 years. The model might be . if a lot of boxers start their career and peak their career 10 to 15 year later, the top rating will peak too ...

I will go into this - perhaps understanding this process will help ...

[computerrank]

and the rating Holyfield can lose ???

Holyfield looses points - from 1312 down to 1304