Ratings - please read before commenting - Archived

[giacomino]

I'm not sure what other people think, but I'd say the 50% or even the 0% number looks better. Nobnom at #30 is not unreasonable, and I do remember Gamboa shooting up the rankings a bit quickly as well. He was in the top 10 before he ever beat anyone with a pulse. Hell, he's #4 now and he still hasn't beaten any particularly good opponents.

I'd go with 50% or maybe even 25%.

I am really not trying to be an intellectual Neanderthal on this issue but any rating system that allows a fighter to step into the ring for the first time and beat an 11-9 journeyman, and then ranks him #30 in the world, let alone #5, is deeply flawed IMO. Many of the rankings in this system are good. This one, to me, borders on silly

[jujigatame]

I would have no problem with #30, or maybe even a bit higher. If a debuting fighter beats let say the #15 guy in the weight class, where do you think he should be ranked?

[computerrank]

I'm not sure what other people think, but I'd say the 50% or even the 0% number looks better. Nobnom at #30 is not unreasonable, and I do remember Gamboa shooting up the rankings a bit quickly as well. He was in the top 10 before he ever beat anyone with a pulse. Hell, he's #4 now and he still hasn't beaten any particularly good opponents.

I'd go with 50% or maybe even 25%.

I am really not trying to be an intellectual Neanderthal on this issue but any rating system that allows a fighter to step into the ring for the first time and beat an 11-9 journeyman, and then ranks him #30 in the world, let alone #5, is deeply flawed IMO. Many of the rankings in this system are good. This one, to me, borders on silly

giacomino,

this #30 from (boxing) scratch is inevitable, if you obay the inevitable ranking rules of professional boxing:

- the winner is rated higher than the loser
- a boxer with some achievemants is not dropped to nowhere for simply losing to any other boxer

no logical alternative in any way.

If you still think this is deeply flawed, this implicitly means, that the inevitable ranking rules of professional boxing are deeply flawed, which I mean is true ...

But this won't help for anything ...

[JCS]

The problem here is the one-trick pony syndrome. A helpful solution (that I eventually found) is to incorporate a secondary rating that basically represents how "proven" a boxer's rating is. That way, if a guy has a poor career, gets one lucky win, then loses.. upon that loss, he'd fall faster than a guy who has been at that lofty rating for years. This isn't immediate and doesn't work in all situations though -- and is difficult to implement fairly.

Although Martin, I think 50% of pre-bout rating may be too much... In my recent studies w/ MMA, I think I use something closer to 20-25%. Different sport, but same idea.

[jujigatame]

If you're using 25% for MMA, I'd think the appropriate percentage for boxing would be even smaller, as boxers tend to have more fights and rise through the ranks a bit slower than MMAers.

[computerrank]

The defeated boxer Megrino was #11 before he lost to Por Nobnom.

So I will reduce the debutant's pre-bout rating to 25% of the loser's.

In this case Por Nobnom will be #17 with 338 points.

The fix will need some time, as I am waiting for another adaption in conjunction with the implementation of an order number for one-day tournaments with multiple fights per boxer.

[jujigatame]

Sounds good to me!!

[greg]

Sorry, I had unfortunately no time to read all these pages...
My question concerns the ratings of the boxers moving up... let's say from CW to HW. I've always had a problem accepting the ratings of the former top CW boxers as the current top 10 HWs: names such as Adamek, Haye, Gomez come to mind... What's the logic behind that?

[computerrank]

Sorry, I had unfortunately no time to read all these pages...
My question concerns the ratings of the boxers moving up... let's say from CW to HW. I've always had a problem accepting the ratings of the former top CW boxers as the current top 10 HWs: names such as Adamek, Haye, Gomez come to mind... What's the logic behind that?

The ratings are changed by a factor depending on the typical weights in these divisions.

The typical weight is asumed to be the upper division weight limit, and for heavyweight it is asumed to be 240 pound.

The adaption factor is the quotient of these weights multiplied by itself.

So for the change from cruiserweight to heayyweight the factor is:

200 * 200 / 240 / 240 = 0.694

There is am additional correction for the specific competition level in the divisions - as the average rating of #8 to #12 in every division is set to a rating of 500.

[Asterix]

I think using 250 lbs (0.64) would be more accurate in terms of how the cruiserweights are rated.

[computerrank]

I think using 250 lbs (0.64) would be more accurate in terms of how the cruiserweights are rated.

The value isn't arbitrary - the 10-year mean value of 10 top rated over the years is taken.
The increase was around 5 pound per decade since 1900, starting with 190 pound.

Maybe it will be 245 next decade.

[conan_the_cribber]

Hi Martin,

just out of interest, would it be possible via the program to calculate a highest rank and date of highest rank. This is pretty common amongst other ranking programs.

cheers

conan

[computerrank]

The rating program is not tracking the ranks - and it would mean a time and ressource consuming additional load to do it.

The ranks would have to be changed for a lot of boxers (not only the opponents in a specific bout) for every bout.

[computerrank]

The defeated boxer Megrino was #11 before he lost to Por Nobnom.

So I will reduce the debutant's pre-bout rating to 25% of the loser's.

In this case Por Nobnom will be #17 with 338 points.

The fix will need some time, as I am waiting for another adaption in conjunction with the implementation of an order number for one-day tournaments with multiple fights per boxer.

The fixed release is launched ...

[computerrank]

maybe points gain would be an option:

- the boxer with highest points gain per division would be displayed
- but only, if the margin is higher than a minimum gain - say 50 points

... and maybe it should be the climber of the month

[Asterix]

maybe points gain would be an option:

- the boxer with highest points gain per division would be displayed
- but only, if the margin is higher than a minimum gain - say 50 points

... and maybe it should be the climber of the month

Climber of the month would be very interesting.

[paul_ftm]

First off - what a great site!

My first few weeks on here and I have spent hours marvelling at the detail and the functionality of the site - outstanding!

Just a small thing though (and let me say I don't understand the rating system yet) - how on God's earth is David Haye ranked so low in the P4P rankings???!!!!

Maybe it will take something like when Calzaghe whipped Lacy (i.e. beat an American "hope") that the world will pay attention.

Look at his only defeat against Thompson - he didn't have to take that fight - he fought a more experienced, more battle hardened fighter and still tried to knok him out.

Look at his record of going into other people's back yards and beating them.

Surely the ratings computer must have some room for a human "common-sense" override?!

Again - great site - love it and heres to a golden era of boxing.

[conan_the_cribber]

This is pretty odd. At first I thought it was just another Haye nuthugger, but I think he's right. Haye is ranked 46, a good 15 places behind Marco Huck who has done dick all in his career. I mean Haye did unify at cruiser and then beat the number three heavy.

It seems as if the lower weight fighters are over represented. There are four flyweights before Haye.

Tough to please everyone, but it doesn't feel right.

conan

[Freedom]

Any chance of having a "African, Asian and South America ratings?" There is a Commonwealth, Mex/Can/USA and Europe.

[computerrank]

As already discussed before, Boxrec's P4P list reflects the weight division excellence in the boxer's current weight class - it is not a representation of the collected achievements over multiple weight divisions.

[Asterix]

You cannot seriously believe that David Haye belongs any where NEAR a P4P list, do you?

He might belong a little higher, but not really, even still he should not be mentioned in the P4P talks until he actually beats someone worth mentioning, and the Valuev fight was garbage.

Are you kidding? The Boxrec P4P list goes up to 100. I'm pretty sure Haye is in the top 100.

[conan_the_cribber]

As already discussed before, Boxrec's P4P list reflects the weight division excellence in the boxer's current weight class - it is not a representation of the collected achievements over multiple weight divisions.

OK, my bad. In that case Haye is nowhere near p4p on this critieria.

I have no idea, but do the All-Time ratings incorporate mutli-division excellence? If yes, then perhaps its possible to do an All-Time ranking of active fighters in order to get a multi-division excellence for active fighters. Somewhat more like the pfp people are expecting.

However, something that occurred to me the other day. There are now 74 pages of posts on the ratings. There is so much material in there, it is not funny. I get the feeling that there needs to be a ratings FAQ or something. The link on page one of the thread only discusses the current ratings. And although the examples are good to someone who's been involved in the ratings, they do seem a bit technical for a first time reader.

conan

[computerrank]

@conan

both suggestions will be reconsidered, thanks.

[Asterix]

Are you kidding? The Boxrec P4P list goes up to 100. I'm pretty sure Haye is in the top 100.

I just meant in the top 10 P4P, sorry if I offended you oh mighty one,

You didn't offend me. :P

[computerrank]

@conan

here an p4p option based on the annual ratings:

- current rating with weight 100%
- 1 year ago with weight 80%
...
- 4 years ago with weight 20%
- 5 years ago and older not regarded

I like the new one ...

Code: Select all

+--------------+---------------+-----+---------------------------+
| first_name   | last_name     | p4p | optional p4p (current p4p)
+--------------+---------------+-----+---------------------------+
| Manny        | Pacquiao      | 367 |  1 ( 1)
| Floyd        | Mayweather Jr | 363 |  2 ( 2)
| Israel       | Vazquez       | 255 |  3 (65)
| Juan Manuel  | Marquez       | 252 |  4 (10)
| Shane        | Mosley        | 241 |  5 ( 3) 
| Wladimir     | Klitschko     | 223 |  6 ( 5)
| Chris        | John          | 204 |  7 (13)
| Rafael       | Marquez       | 190 |  8 (84)
| Celestino    | Caballero     | 188 |  9 (16)
| Chad         | Dawson        | 187 | 10 ( 4) 
| Kelly        | Pavlik        | 183 | 11 ( 8)
| Nonito       | Donaire       | 179 | 12 (26)
| Hozumi       | Hasegawa      | 176 | 13 ( 6)
| Ricky        | Hatton        | 172 | 14 (41)
| Antonio      | Margarito     | 171 | 15 (34)
| Pongsaklek   | Wonjongkam    | 169 | 16 ( 9)
| Vitali       | Klitschko     | 153 | 17 ( 7)
| Vic          | Darchinyan    | 153 | 18 (22)
| Jermain      | Taylor        | 148 | 19 ( -)
| David        | Haye          | 147 | 20 (42)
| Paul         | Williams      | 143 | 21 (11)
| Daisuke      | Naito         | 134 | 22 (18)
| Arthur       | Abraham       | 133 | 23 (14)
| Edgar        | Sosa          | 132 | 24 (15)
| Tomasz       | Adamek        | 127 | 25 ( -)
| Miguel Angel | Cotto         | 125 | 26 (85)
| Omar Andres  | Narvaez       | 120 | 27 (19)
| Ronald       | Wright        | 113 | 28 ( -)
| Ivan         | Calderon      | 112 | 29 ( -)
| Robert       | Guerrero      | 110 | 30 (20)
| Felix        | Sturm         | 110 | 31 (17)
| Juan         | Diaz          | 105 | 32 ( -)
| Denkaosan    | Kaovichit     | 103 | 33 (28)
| Nate         | Campbell      | 103 | 34 (49)
+--------------+---------------+-----+---------------------------+