JCS83MD wrote:I just took a wack at a ranking-type system, although I'm not strictly following any ABC rules. I'll just post heavy.
This is as of 8/3 (Before Mora vs. Johnson was changed to a NC, and yes, I know its still kinda messed up)
Its definitely a work in progress, but I bet some organization could pass this off as their ratings.
Heavy
1 Fres Oquendo
2 Wlad Klitschko
3 Javier Mora
4 Samuel Peter
5 Sergei Lyakhovich
6 Kirk Johnson
7 Lamon Brewster
8 Matt Skelton
9 Oleg Maskaev
10 Luan Krasniqi
11 Vladimir Virchis
12 Danny Williams
13 Chris Byrd
14 Ruslan Chagaev
15 Sinan Samil Sam
16 Paolo Vidoz
17 Michael Sprott
18 Tony Thompson
19 Eddie Chambers
20 Hasim Rahman
Good on you for trying WCS. That could've been the WBO. I cant for the life of think why Fres ended up on top, but hey, the rest aint that bad.
conan
Well the Kirk Johnson debacle w/ Mora, leapfrogged Mora over Johnson, than leapfrogged Oquendo over Mora.
There's an issue w/ Johnson holding on too long there...
Revision
1. Wlad Klitschko
2. Chris Byrd
3. Hasim Rahman
4. Fres Oquendo
5. Javier Mora
6. Sergei Lyakhovich
7. Sam Peter
8. Matt Skelton
9. Kirk Johnson
10. Tony Thompson
11. Lamon Brewster
12. Calvin Brock
13. Nicolay Valuev
14. Eddie Chambers
15. Sinan Samil Sam
16. Kevin Johnson
17. Vladimir Virchis
18. Luan Krasniqi
19. Danny Williams
20. Dominick Guinn
I'm very interested in the rankings, and I apologize if I'm setting the discussion back any. I occasionally get into a snit about them, but then I remember to come here and read how they are derived and I appreciate the work you guys have put in on this. I understand and appreciate statistics, but not good enough so to be able to contribute mathematical input to the ideas.
But I do have a few issues that I don't like or find odd.
One thing is how the end of the 'launched' fighter creates a big block of veteran fighters with a lot of losses in the rankings, followed by the sudden inclusion of a bunch of young fighters. I understand the rationale for the launching idea, but it creates an odd ranking tier which does not, IMO, reflect reality.
And I guess I question whether or not predictive success is a good metric for evaluating the rankings, although I guess I can't really think of a better method. However, it seems to me that to have a high level of predictive success, you have to tweak the ranking mechanism to account for upsets. Essentially, it seems to me, you are saying that an optimal ranking system would have had Omar Nino ranked above Brian Viloria because he won the fight. Even if Nino is the better fighter, his fight history could not justify being placed ahead of Viloria, so it seems strange to me to try and come up with statistical reasoning to describe why he won.
It would surprise me if you could find statistically significant factors in upsets, but perhaps thats what this thread is about - you have, and I honestly haven't reviewed the data or formulas very thoroughly.
I think what it comes down to, for me, is that I would rather have a ranking of accomplishment over a ranking of predictability. I'm not sure there's a lot of difference between those two ideas in the end.
Also, I think a daily deduction of points earned from fights is not right. Why start deducting points immediately after a fight - boxers cannot fight right away. Why not just make the victory worth less points and not start deducting from it for a certain period of time? Since BoxRec turns fighters inactive after a year, it seems to me like a fighter's points from any fight maybe should last that long before they start to be deducted. Too long?
Thanks again for all your work with this. I really enjoy the ratings and think they are essential to contextualize the database, so I hope to see you guys continue to fine tune.
OK, I learned my lesson - thanks. As simple as possible is the requirement, I am sure now after all discussions.
The traditional rankings work for their simplicity - no goal on predicition - simply take the rank of the better ranked by a win. Transparent and basta.
All chasing for the details and performance ratio is dedicated to the performance ratings.
5 simple rules
separate for every division - and the result is great - this will be published as robust, simple, traditional rankings:
R1) winner was better ranked than loser
-> no change
R2) winner was worse ranked than loser
-> winner gets loser's rank
-> loser gets geometric mean rank of winner and loser
R3) draw
-> no change
R4) inactivity
-> after 18 months of inactivity the boxer is removed from the division
-> his background ranking is multilpied with factor 3 for every complete inactivity period of 18 months
-> in case of come back, the background ranking is activated and 1 reank added
R5) missing quality of opponents
-> after 18 months of missing an opponent within factor 3 of own rank, quality is missing
-> for top 15 the tolerance area is within rank 45
-> the boxer's rank is multiplied with 2 and 1 rank added for missing quality of opponents
Result:
M Heavyweight 1 Nikolay Valuev 69
M Heavyweight 2 Hasim Rahman 146
M Heavyweight 3 Wladimir Klitschko 111
M Heavyweight 4 John Ruiz 237
M Heavyweight 5 Calvin Brock 48
M Heavyweight 6 Chris Byrd 111
M Heavyweight 7 Monte Barrett 363
M Heavyweight 8 Matt Skelton 34
M Heavyweight 9 Serguei Lyakhovich 132
M Heavyweight 10 James Toney 146
M Heavyweight 11 Lamon Brewster 132
M Heavyweight 12 Zuri Lawrence 167
M Heavyweight 13 DaVarryl Williamson 97
M Heavyweight 14 Kevin McBride 132
M Heavyweight 15 Samuel Peter 105
M Heavyweight 16 Danny Williams 34
M Heavyweight 17 Damian Wills 64
M Heavyweight 18 Tony Thompson 44
M Heavyweight 19 Ruslan Chagaev 27
M Heavyweight 20 Larry Donald 314
M Cruiserweight 1 O'Neil Bell 216
M Cruiserweight 2 Johnny Nelson 258
M Cruiserweight 3 Jean Marc Mormeck 216
M Cruiserweight 4 Guillermo Jones 342
M Cruiserweight 5 Wayne Braithwaite 342
M Cruiserweight 6 Enzo Maccarinelli 34
M Cruiserweight 7 Vincenzo Cantatore 83
M Cruiserweight 8 Steve Cunningham 216
M Cruiserweight 9 David Haye 21
M Cruiserweight 10 Alexander Gurov 238
M Cruiserweight 11 Aloryi Moyoyo Mensah 91
M Cruiserweight 12 Krzysztof Wlodarczyk 41
M Cruiserweight 13 Dale Brown 239
M Cruiserweight 14 Antonio Mercado 181
M Cruiserweight 15 Marco Huck 69
M Cruiserweight 16 Mark Hobson 71
M Cruiserweight 17 Vincenzo Rossitto 27
M Cruiserweight 18 Patrick Nwamu 239
M Cruiserweight 19 Luis Andres Pineda 244
M Cruiserweight 20 Grigory Drozd 13
M Light Heavyweight 1 Zsolt Erdei 13
M Light Heavyweight 2 Glen Johnson 168
M Light Heavyweight 3 Roy Jones Jr 13
M Light Heavyweight 4 Clinton Woods 90
M Light Heavyweight 5 Julio Gonzalez 22
M Light Heavyweight 6 George Khalid Jones 315
M Light Heavyweight 7 Tomasz Adamek 300
M Light Heavyweight 8 Chad Dawson 70
M Light Heavyweight 9 Gabriel Campillo 251
M Light Heavyweight 10 Stipe Drews 76
M Light Heavyweight 11 Paul Briggs 300
M Light Heavyweight 12 Thomas Ulrich 13
M Light Heavyweight 13 Mehdi Sahnoune 293
M Light Heavyweight 14 Hugo Hernan Garay 125
M Light Heavyweight 15 Montell Griffin 133
M Light Heavyweight 16 Kai Kurzawa 76
M Light Heavyweight 17 Julio Cesar Dominguez 23
M Light Heavyweight 18 Adrian Diaconu 87
M Light Heavyweight 19 Jason DeLisle 90
M Light Heavyweight 20 Yuri Barashian 301
M Super Middleweight 1 Markus Beyer 195
M Super Middleweight 2 Joe Calzaghe 160
M Super Middleweight 3 Mikkel Kessler 209
M Super Middleweight 4 Anthony Mundine 86
M Super Middleweight 5 Jeff Lacy 160
M Super Middleweight 6 Peter Manfredo Jr 179
M Super Middleweight 7 Danny Green 86
M Super Middleweight 8 Cristian Sanavia 15
M Super Middleweight 9 Robin Reid 370
M Super Middleweight 10 Omar Sheika 342
M Super Middleweight 11 Mger Mkrtchian 119
M Super Middleweight 12 Carl Froch 77
M Super Middleweight 13 Lucian Bute 87
M Super Middleweight 14 Brian Magee 77
M Super Middleweight 15 Scott Pemberton 179
M Super Middleweight 16 Mario Veit 76
M Super Middleweight 17 Alejandro Berrio 84
M Super Middleweight 18 Vitali Tsypko 266
M Super Middleweight 19 Librado Andrade 125
M Super Middleweight 20 Robert Stieglitz 251
Thanks for your discussion and inspiration
Martin
Hi Martin,
the first four divisions look great, but I'm worried that your five rules dont cover the lower divisions where division jumps are more common. Can you publish the rest as well.
What is your solution to the division jump problem?
... now the results with division changes regarded - 1 additional rule R6:
6 simple rules
R1) winner was better ranked than loser
-> no change
R2) winner was worse ranked than loser
-> winner gets loser's rank
-> loser gets geometric mean rank of winner and loser
R3) draw
-> no change
R4) inactivity
-> after 18 months of inactivity the boxer is removed from the division
-> his background ranking is multilpied with factor 3 for every complete inactivity period of 18 months
-> in case of come back, the background ranking is activated and 1 reank added
R5) missing quality of opponents
-> after 18 months of missing an opponent within factor 3 of own rank, quality is missing
-> for top 15 the tolerance area is within rank 45
-> the boxer's rank is multiplied with 2 and 1 rank added for missing quality of opponents
R6) division change
-> for move up the rank is multiplied by 3 per divsion step
-> add another 10 ranks for move to heavyweight
-> rank kept for moving down
Results until August 3, 2006 - prediction rate 67%:
M Heavyweight 1 Nikolay Valuev
M Heavyweight 2 Hasim Rahman
M Heavyweight 3 John Ruiz
M Heavyweight 4 Wladimir Klitschko
M Heavyweight 5 Calvin Brock
M Heavyweight 6 Monte Barrett
M Heavyweight 7 James Toney
M Heavyweight 8 Chris Byrd
M Heavyweight 9 Serguei Lyakhovich
M Heavyweight 10 Matt Skelton
M Heavyweight 11 Lamon Brewster
M Heavyweight 12 Zuri Lawrence
M Heavyweight 13 Luan Krasniqi
M Heavyweight 14 Larry Donald
M Heavyweight 15 Sultan Ibragimov
M Heavyweight 16 Samuel Peter
M Heavyweight 17 DaVarryl Williamson
M Heavyweight 18 Danny Williams
M Heavyweight 19 Ruslan Chagaev
M Heavyweight 20 Damian Wills
M Cruiserweight 1 Johnny Nelson
M Cruiserweight 2 O'Neil Bell
M Cruiserweight 3 Guillermo Jones
M Cruiserweight 4 Jean Marc Mormeck
M Cruiserweight 5 Wayne Braithwaite
M Cruiserweight 7 Steve Cunningham
M Cruiserweight 8 Enzo Maccarinelli
M Cruiserweight 9 Vincenzo Cantatore
M Cruiserweight 10 Aloryi Moyoyo Mensah
M Cruiserweight 11 Antonio Mercado
M Cruiserweight 12 David Haye
M Cruiserweight 13 Alexander Gurov
M Cruiserweight 15 Luis Andres Pineda
M Cruiserweight 17 Virgil Hill
M Cruiserweight 18 Jorge Fernando Castro
M Cruiserweight 18 Grigory Drozd
M Cruiserweight 18 Jose Luis Herrera
M Cruiserweight 19 Mark Hobson
M Cruiserweight 20 Hastings Rasani
M Light Heavyweight 1 Zsolt Erdei
M Light Heavyweight 3 Glen Johnson
M Light Heavyweight 4 Roy Jones Jr
M Light Heavyweight 7 Antonio Tarver
M Light Heavyweight 9 Clinton Woods
M Light Heavyweight 10 Julio Gonzalez
M Light Heavyweight 11 Tomasz Adamek
M Light Heavyweight 12 Thomas Ulrich
M Light Heavyweight 15 Mehdi Sahnoune
M Light Heavyweight 16 Paul Briggs
M Light Heavyweight 17 Stipe Drews
M Light Heavyweight 18 George Khalid Jones
M Light Heavyweight 18 Danilo Haussler
M Light Heavyweight 20 Chad Dawson
M Super Middleweight 1 Joe Calzaghe
M Super Middleweight 2 Mikkel Kessler
M Super Middleweight 3 Markus Beyer
M Super Middleweight 4 Jeff Lacy
M Super Middleweight 5 Anthony Mundine
M Super Middleweight 7 Danny Green
M Super Middleweight 10 Robin Reid
M Super Middleweight 11 Mger Mkrtchian
M Super Middleweight 12 Cristian Sanavia
M Super Middleweight 13 Lucian Bute
M Super Middleweight 14 Carl Froch
M Super Middleweight 15 Brian Magee
M Super Middleweight 16 Vitali Tsypko
M Super Middleweight 17 Peter Manfredo Jr
M Super Middleweight 18 Alejandro Berrio
M Middleweight 1 Jermain Taylor
M Middleweight 3 Ronald Wright
M Middleweight 4 Arthur Abraham
M Middleweight 6 Edison Miranda
M Middleweight 7 Amin Asikainen
M Middleweight 8 Franck Mezaache
M Middleweight 10 Kingsley Ikeke
M Middleweight 12 Khoren Gevor
M Middleweight 13 Kelly Pavlik
M Middleweight 15 Sam Soliman
M Middleweight 19 Bronco McKart
M Middleweight 20 Yoshihiro Araki
M Light Middleweight 1 Shane Mosley
M Light Middleweight 2 Sergio Gabriel Martinez
M Light Middleweight 3 Cory Spinks
M Light Middleweight 4 Roman Karmazin
M Light Middleweight 5 Fernando Vargas
M Light Middleweight 6 Verno Phillips
M Light Middleweight 7 Oscar De La Hoya
M Light Middleweight 8 Ricardo Mayorga
M Light Middleweight 10 Sergiy Dzinziruk
M Light Middleweight 11 Attila Kovacs
M Light Middleweight 12 Kassim Ouma
M Light Middleweight 17 Steve Conway
M Light Middleweight 19 Michele Piccirillo
M Welterweight 1 Carlos Manuel Baldomir
M Welterweight 3 Ricky Hatton
M Welterweight 4 Luis Collazo
M Welterweight 5 Antonio Margarito
M Welterweight 6 Floyd Mayweather Jr
M Welterweight 8 Zab Judah
M Welterweight 9 Kermit Cintron
M Welterweight 10 Arturo Gatti
M Welterweight 11 Oktay Urkal
M Welterweight 13 Dondon Sultan
M Welterweight 14 Sharmba Mitchell
M Welterweight 15 Frederic Klose
M Welterweight 16 Carlos Quintana
M Welterweight 17 Juan Lazcano
M Welterweight 18 Thomas Damgaard
M Welterweight 19 Kevin Anderson
M Welterweight 20 Antonio Pitalua
M Light Welterweight 2 Jose Luis Castillo
M Light Welterweight 6 Miguel Angel Cotto
M Light Welterweight 7 Carlos Maussa
M Light Welterweight 8 DeMarcus Corley
M Light Welterweight 9 Demetrius Hopkins
M Light Welterweight 12 Cesar Rene Cuenca
M Light Welterweight 12 Lamont Peterson
M Light Welterweight 13 Juan Urango
M Light Welterweight 14 Jonathan Thaxton
M Light Welterweight 15 Samuel Malinga
M Light Welterweight 16 Junior Witter
M Light Welterweight 18 Arturo Morua
M Light Welterweight 19 Andreas Kotelnik
M Light Welterweight 20 Herman Ngoudjo
M Lightweight 1 Diego Corrales
M Lightweight 3 Acelino Freitas
M Lightweight 4 Julio Diaz
M Lightweight 5 Jesus Chavez
M Lightweight 6 Zahir Raheem
M Lightweight 7 Juan Carlos Diaz Melero
M Lightweight 9 Graham Earl
M Lightweight 10 Juan Diaz
M Lightweight 11 Ricky Quiles
M Lightweight 14 Fernando David Saucedo
M Lightweight 15 Yuri Romanov
M Lightweight 16 Aldo Nazareno Rios
M Lightweight 17 Diego Martin Alzugaray
M Lightweight 19 Joan Guzman
M Lightweight 20 Joel Casamayor
M Super Featherweight 1 Gairy St Clair
M Super Featherweight 2 Manny Pacquiao
M Super Featherweight 3 Erik Morales
M Super Featherweight 4 Marco Antonio Barrera
M Super Featherweight 5 Alex Arthur
M Super Featherweight 6 Cassius Baloyi
M Super Featherweight 8 Manuel Medina
M Super Featherweight 9 Craig Docherty
M Super Featherweight 10 Roberto David Arrieta
M Super Featherweight 11 Jorge Rodrigo Barrios
M Super Featherweight 12 Janos Nagy
M Super Featherweight 13 Vicente Mosquera
M Super Featherweight 15 Robbie Peden
M Super Featherweight 16 Bobby Pacquiao
M Super Featherweight 19 Mzonke Fana
M Featherweight 1 Sandro Marcos
M Featherweight 2 Chris John
M Featherweight 3 Humberto Soto
M Featherweight 4 Juan Manuel Marquez
M Featherweight 5 Oscar Leon
M Featherweight 7 Scott Harrison
M Featherweight 8 Orlando Salido
M Featherweight 9 Thomas Mashaba
M Featherweight 12 Robert Guerrero
M Featherweight 13 Tomas Villa
M Featherweight 14 Johnny Tapia
M Featherweight 17 Jorge Lacierva
M Featherweight 18 Eric Aiken
M Featherweight 19 Gamaliel Diaz
M Featherweight 20 Spend Abazi
M Super Bantamweight 1 Israel Vazquez
M Super Bantamweight 2 Somsak Sithchatchawal
M Super Bantamweight 4 Wethya Sakmuangklang
M Super Bantamweight 9 Celestino Caballero
M Super Bantamweight 10 Daniel Ponce De Leon
M Super Bantamweight 13 Zolani Marali
M Super Bantamweight 15 Sod Looknongyangtoy
M Super Bantamweight 16 Sergio Manuel Medina
M Super Bantamweight 17 Ricardo Castillo
M Super Bantamweight 20 Al Seeger
M Bantamweight 1 Hozumi Hasegawa
M Bantamweight 2 Rafael Marquez
M Bantamweight 3 Veeraphol Sahaprom
M Bantamweight 4 Jhonny Gonzalez
M Bantamweight 5 Simone Maludrottu
M Bantamweight 6 Johnny Bredahl
M Bantamweight 7 Damaen Kelly
M Bantamweight 9 Wladimir Sidorenko
M Bantamweight 10 Silence Mabuza
M Bantamweight 12 Samuel Lopez
M Bantamweight 13 Irene Pacheco
M Bantamweight 14 Carmelo Ballone
M Bantamweight 15 Alejandro Valdez
M Bantamweight 16 Sasha Bakhtin
M Bantamweight 17 Mauricio Martinez
M Bantamweight 18 Fernando Montiel
M Super Flyweight 1 Masamori Tokuyama
M Super Flyweight 3 Nobuo Nashiro
M Super Flyweight 5 Simon Ramoni
M Super Flyweight 7 Martin Castillo
M Super Flyweight 10 Theo Modise
M Super Flyweight 11 Luis Alberto Perez
M Super Flyweight 13 Cristian Mijares
M Super Flyweight 18 Jair Jimenez
M Super Flyweight 19 Alexander Munoz
M Flyweight 1 Vic Darchinyan
M Flyweight 2 Pongsaklek Wonjongkam
M Flyweight 3 Jorge Armando Arce
M Flyweight 4 Lorenzo Parra
M Flyweight 5 Brahim Asloum
M Flyweight 7 Jose Lopez Bueno
M Flyweight 8 Omar Andres Narvaez
M Flyweight 9 Jose Lopez
M Flyweight 10 Juan Alberto Rosas
M Flyweight 12 Charitt Mukondelela
M Flyweight 13 Alejandro Hernandez
M Flyweight 15 Denkaosan Kaovichit
M Flyweight 17 Daisuke Naito
M Flyweight 18 Monelisi Mhikiza Myekeni
M Light Flyweight 1 Roberto Vasquez
M Light Flyweight 2 Munetsugu Kayo
M Light Flyweight 3 Muvhuso Nedzanani
M Light Flyweight 5 Ulises Solis
M Light Flyweight 7 Will Grigsby
M Light Flyweight 8 Samora Msopi
M Light Flyweight 9 Wandee Singwangcha
M Light Flyweight 10 Mfundo Gwayana
M Light Flyweight 11 Brian Viloria
M Light Flyweight 12 Hugo Fidel Cazares
M Light Flyweight 13 Juanito Rubillar
M Light Flyweight 14 Takahisa Masuda
M Light Flyweight 15 Vuyani Kheswa
M Light Flyweight 20 Giuseppe Lagana
M Minimumweight 1 Muhammad Rachman
M Minimumweight 2 Eagle Kyowa
M Minimumweight 3 Yutaka Niida
M Minimumweight 4 Katsunari Takayama
M Minimumweight 5 Nkosinathi Joyi
M Minimumweight 6 Mawanda Sineko
M Minimumweight 7 Ivan Calderon
M Minimumweight 9 Isaac Bustos
M Minimumweight 11 Thobani Mbangeni
M Minimumweight 12 Lulama Ngqaka
M Minimumweight 13 Omar Soto
M Minimumweight 14 Satoshi Kogumazaka
M Minimumweight 15 Florante Condes
M Minimumweight 16 Pornsawan Kratingdaenggym
M Minimumweight 17 Teruo Misawa
M Minimumweight 18 Omar Soto
M Minimumweight 20 Toshiki Ogawa
You: ... One thing is how the end of the 'launched' fighter creates a big block of veteran fighters with a lot of losses in the rankings, followed by the sudden inclusion of a bunch of young fighters. I understand the rationale for the launching idea, but it creates an odd ranking tier which does not, IMO, reflect reality.
-> Yes there is this separation of boxers into 2 blocks
- you state, the ratings are not correct now, but the ratings are only valid for the boxers launched - so boxers no yet launched or no more launched don't have a valid rating and are not rated
- we only show their launching points (not their rating, which isn't valid)
- the launching process is plausible, you step up grade by grade until perhaps hgetting launched - but the process is not optimized by an evaluation process
- so please look at the top launched boxers for a rating, and please don't talk about a rating for the rest
- the optimization and evaluation process for the ratings (launched boxers) only uses bouts between launched boxers strictly - so there is no interference of effects.
You: And I guess I question whether or not predictive success is a good metric for evaluating the rankings, although I guess I can't really think of a better method. However, it seems to me that to have a high level of predictive success, you have to tweak the ranking mechanism to account for upsets. Essentially, it seems to me, you are saying that an optimal ranking system would have had Omar Nino ranked above Brian Viloria because he won the fight. Even if Nino is the better fighter, his fight history could not justify being placed ahead of Viloria, so it seems strange to me to try and come up with statistical reasoning to describe why he won.
It would surprise me if you could find statistically significant factors in upsets, but perhaps thats what this thread is about - you have, and I honestly haven't reviewed the data or formulas very thoroughly.
I think what it comes down to, for me, is that I would rather have a ranking of accomplishment over a ranking of predictability. I'm not sure there's a lot of difference between those two ideas in the end.
-> There is a lot of difference in the two methods, we check for this.
- performance ratings show about 80% prediction rate: complex and statisically based
- traditional ratings show about 67%: simple rules, looks more intuitive just
But at this moment we try to implement both flavours of rating/ranking - and maybe you will soon see both tables on the web site ... and you can study then both or only your favourite.
You: Also, I think a daily deduction of points earned from fights is not right. Why start deducting points immediately after a fight - boxers cannot fight right away. Why not just make the victory worth less points and not start deducting from it for a certain period of time? Since BoxRec turns fighters inactive after a year, it seems to me like a fighter's points from any fight maybe should last that long before they start to be deducted. Too long?
-> we found that all manipulations and corrections hurt hte predicition rate - statistics say let them get the whole values ...
computerrank wrote:... now the results with division changes regarded - 1 additional rule R6:
6 simple rules
R1) winner was better ranked than loser
-> no change
R2) winner was worse ranked than loser
-> winner gets loser's rank
-> loser gets geometric mean rank of winner and loser
R3) draw
-> no change
R4) inactivity
-> after 18 months of inactivity the boxer is removed from the division
-> his background ranking is multilpied with factor 3 for every complete inactivity period of 18 months
-> in case of come back, the background ranking is activated and 1 reank added
R5) missing quality of opponents
-> after 18 months of missing an opponent within factor 3 of own rank, quality is missing
-> for top 15 the tolerance area is within rank 45
-> the boxer's rank is multiplied with 2 and 1 rank added for missing quality of opponents
R6) division change
-> for move up the rank is multiplied by 3 per divsion step
-> add another 10 ranks for move to heavyweight
-> rank kept for moving down
Results until August 3, 2006 - prediction rate 67%:
M Heavyweight 1 Nikolay Valuev
M Heavyweight 2 Hasim Rahman
M Heavyweight 3 John Ruiz
M Heavyweight 4 Wladimir Klitschko
M Heavyweight 5 Calvin Brock
M Heavyweight 6 Monte Barrett
M Heavyweight 7 James Toney
M Heavyweight 8 Chris Byrd
M Heavyweight 9 Serguei Lyakhovich
M Heavyweight 10 Matt Skelton
M Heavyweight 11 Lamon Brewster
M Heavyweight 12 Zuri Lawrence
M Heavyweight 13 Luan Krasniqi
M Heavyweight 14 Larry Donald
M Heavyweight 15 Sultan Ibragimov
M Heavyweight 16 Samuel Peter
M Heavyweight 17 DaVarryl Williamson
M Heavyweight 18 Danny Williams
M Heavyweight 19 Ruslan Chagaev
M Heavyweight 20 Damian Wills
M Cruiserweight 1 Johnny Nelson
M Cruiserweight 2 O'Neil Bell
M Cruiserweight 3 Guillermo Jones
M Cruiserweight 4 Jean Marc Mormeck
M Cruiserweight 5 Wayne Braithwaite
M Cruiserweight 7 Steve Cunningham
M Cruiserweight 8 Enzo Maccarinelli
M Cruiserweight 9 Vincenzo Cantatore
M Cruiserweight 10 Aloryi Moyoyo Mensah
M Cruiserweight 11 Antonio Mercado
M Cruiserweight 12 David Haye
M Cruiserweight 13 Alexander Gurov
M Cruiserweight 15 Luis Andres Pineda
M Cruiserweight 17 Virgil Hill
M Cruiserweight 18 Jorge Fernando Castro
M Cruiserweight 18 Grigory Drozd
M Cruiserweight 18 Jose Luis Herrera
M Cruiserweight 19 Mark Hobson
M Cruiserweight 20 Hastings Rasani
M Light Heavyweight 1 Zsolt Erdei
M Light Heavyweight 3 Glen Johnson
M Light Heavyweight 4 Roy Jones Jr
M Light Heavyweight 7 Antonio Tarver
M Light Heavyweight 9 Clinton Woods
M Light Heavyweight 10 Julio Gonzalez
M Light Heavyweight 11 Tomasz Adamek
M Light Heavyweight 12 Thomas Ulrich
M Light Heavyweight 15 Mehdi Sahnoune
M Light Heavyweight 16 Paul Briggs
M Light Heavyweight 17 Stipe Drews
M Light Heavyweight 18 George Khalid Jones
M Light Heavyweight 18 Danilo Haussler
M Light Heavyweight 20 Chad Dawson
M Super Middleweight 1 Joe Calzaghe
M Super Middleweight 2 Mikkel Kessler
M Super Middleweight 3 Markus Beyer
M Super Middleweight 4 Jeff Lacy
M Super Middleweight 5 Anthony Mundine
M Super Middleweight 7 Danny Green
M Super Middleweight 10 Robin Reid
M Super Middleweight 11 Mger Mkrtchian
M Super Middleweight 12 Cristian Sanavia
M Super Middleweight 13 Lucian Bute
M Super Middleweight 14 Carl Froch
M Super Middleweight 15 Brian Magee
M Super Middleweight 16 Vitali Tsypko
M Super Middleweight 17 Peter Manfredo Jr
M Super Middleweight 18 Alejandro Berrio
M Middleweight 1 Jermain Taylor
M Middleweight 3 Ronald Wright
M Middleweight 4 Arthur Abraham
M Middleweight 6 Edison Miranda
M Middleweight 7 Amin Asikainen
M Middleweight 8 Franck Mezaache
M Middleweight 10 Kingsley Ikeke
M Middleweight 12 Khoren Gevor
M Middleweight 13 Kelly Pavlik
M Middleweight 15 Sam Soliman
M Middleweight 19 Bronco McKart
M Middleweight 20 Yoshihiro Araki
M Light Middleweight 1 Shane Mosley
M Light Middleweight 2 Sergio Gabriel Martinez
M Light Middleweight 3 Cory Spinks
M Light Middleweight 4 Roman Karmazin
M Light Middleweight 5 Fernando Vargas
M Light Middleweight 6 Verno Phillips
M Light Middleweight 7 Oscar De La Hoya
M Light Middleweight 8 Ricardo Mayorga
M Light Middleweight 10 Sergiy Dzinziruk
M Light Middleweight 11 Attila Kovacs
M Light Middleweight 12 Kassim Ouma
M Light Middleweight 17 Steve Conway
M Light Middleweight 19 Michele Piccirillo
M Welterweight 1 Carlos Manuel Baldomir
M Welterweight 3 Ricky Hatton
M Welterweight 4 Luis Collazo
M Welterweight 5 Antonio Margarito
M Welterweight 6 Floyd Mayweather Jr
M Welterweight 8 Zab Judah
M Welterweight 9 Kermit Cintron
M Welterweight 10 Arturo Gatti
M Welterweight 11 Oktay Urkal
M Welterweight 13 Dondon Sultan
M Welterweight 14 Sharmba Mitchell
M Welterweight 15 Frederic Klose
M Welterweight 16 Carlos Quintana
M Welterweight 17 Juan Lazcano
M Welterweight 18 Thomas Damgaard
M Welterweight 19 Kevin Anderson
M Welterweight 20 Antonio Pitalua
M Light Welterweight 2 Jose Luis Castillo
M Light Welterweight 6 Miguel Angel Cotto
M Light Welterweight 7 Carlos Maussa
M Light Welterweight 8 DeMarcus Corley
M Light Welterweight 9 Demetrius Hopkins
M Light Welterweight 12 Cesar Rene Cuenca
M Light Welterweight 12 Lamont Peterson
M Light Welterweight 13 Juan Urango
M Light Welterweight 14 Jonathan Thaxton
M Light Welterweight 15 Samuel Malinga
M Light Welterweight 16 Junior Witter
M Light Welterweight 18 Arturo Morua
M Light Welterweight 19 Andreas Kotelnik
M Light Welterweight 20 Herman Ngoudjo
M Lightweight 1 Diego Corrales
M Lightweight 3 Acelino Freitas
M Lightweight 4 Julio Diaz
M Lightweight 5 Jesus Chavez
M Lightweight 6 Zahir Raheem
M Lightweight 7 Juan Carlos Diaz Melero
M Lightweight 9 Graham Earl
M Lightweight 10 Juan Diaz
M Lightweight 11 Ricky Quiles
M Lightweight 14 Fernando David Saucedo
M Lightweight 15 Yuri Romanov
M Lightweight 16 Aldo Nazareno Rios
M Lightweight 17 Diego Martin Alzugaray
M Lightweight 19 Joan Guzman
M Lightweight 20 Joel Casamayor
M Super Featherweight 1 Gairy St Clair
M Super Featherweight 2 Manny Pacquiao
M Super Featherweight 3 Erik Morales
M Super Featherweight 4 Marco Antonio Barrera
M Super Featherweight 5 Alex Arthur
M Super Featherweight 6 Cassius Baloyi
M Super Featherweight 8 Manuel Medina
M Super Featherweight 9 Craig Docherty
M Super Featherweight 10 Roberto David Arrieta
M Super Featherweight 11 Jorge Rodrigo Barrios
M Super Featherweight 12 Janos Nagy
M Super Featherweight 13 Vicente Mosquera
M Super Featherweight 15 Robbie Peden
M Super Featherweight 16 Bobby Pacquiao
M Super Featherweight 19 Mzonke Fana
M Featherweight 1 Sandro Marcos
M Featherweight 2 Chris John
M Featherweight 3 Humberto Soto
M Featherweight 4 Juan Manuel Marquez
M Featherweight 5 Oscar Leon
M Featherweight 7 Scott Harrison
M Featherweight 8 Orlando Salido
M Featherweight 9 Thomas Mashaba
M Featherweight 12 Robert Guerrero
M Featherweight 13 Tomas Villa
M Featherweight 14 Johnny Tapia
M Featherweight 17 Jorge Lacierva
M Featherweight 18 Eric Aiken
M Featherweight 19 Gamaliel Diaz
M Featherweight 20 Spend Abazi
M Super Bantamweight 1 Israel Vazquez
M Super Bantamweight 2 Somsak Sithchatchawal
M Super Bantamweight 4 Wethya Sakmuangklang
M Super Bantamweight 9 Celestino Caballero
M Super Bantamweight 10 Daniel Ponce De Leon
M Super Bantamweight 13 Zolani Marali
M Super Bantamweight 15 Sod Looknongyangtoy
M Super Bantamweight 16 Sergio Manuel Medina
M Super Bantamweight 17 Ricardo Castillo
M Super Bantamweight 20 Al Seeger
M Bantamweight 1 Hozumi Hasegawa
M Bantamweight 2 Rafael Marquez
M Bantamweight 3 Veeraphol Sahaprom
M Bantamweight 4 Jhonny Gonzalez
M Bantamweight 5 Simone Maludrottu
M Bantamweight 6 Johnny Bredahl
M Bantamweight 7 Damaen Kelly
M Bantamweight 9 Wladimir Sidorenko
M Bantamweight 10 Silence Mabuza
M Bantamweight 12 Samuel Lopez
M Bantamweight 13 Irene Pacheco
M Bantamweight 14 Carmelo Ballone
M Bantamweight 15 Alejandro Valdez
M Bantamweight 16 Sasha Bakhtin
M Bantamweight 17 Mauricio Martinez
M Bantamweight 18 Fernando Montiel
M Super Flyweight 1 Masamori Tokuyama
M Super Flyweight 3 Nobuo Nashiro
M Super Flyweight 5 Simon Ramoni
M Super Flyweight 7 Martin Castillo
M Super Flyweight 10 Theo Modise
M Super Flyweight 11 Luis Alberto Perez
M Super Flyweight 13 Cristian Mijares
M Super Flyweight 18 Jair Jimenez
M Super Flyweight 19 Alexander Munoz
M Flyweight 1 Vic Darchinyan
M Flyweight 2 Pongsaklek Wonjongkam
M Flyweight 3 Jorge Armando Arce
M Flyweight 4 Lorenzo Parra
M Flyweight 5 Brahim Asloum
M Flyweight 7 Jose Lopez Bueno
M Flyweight 8 Omar Andres Narvaez
M Flyweight 9 Jose Lopez
M Flyweight 10 Juan Alberto Rosas
M Flyweight 12 Charitt Mukondelela
M Flyweight 13 Alejandro Hernandez
M Flyweight 15 Denkaosan Kaovichit
M Flyweight 17 Daisuke Naito
M Flyweight 18 Monelisi Mhikiza Myekeni
M Light Flyweight 1 Roberto Vasquez
M Light Flyweight 2 Munetsugu Kayo
M Light Flyweight 3 Muvhuso Nedzanani
M Light Flyweight 5 Ulises Solis
M Light Flyweight 7 Will Grigsby
M Light Flyweight 8 Samora Msopi
M Light Flyweight 9 Wandee Singwangcha
M Light Flyweight 10 Mfundo Gwayana
M Light Flyweight 11 Brian Viloria
M Light Flyweight 12 Hugo Fidel Cazares
M Light Flyweight 13 Juanito Rubillar
M Light Flyweight 14 Takahisa Masuda
M Light Flyweight 15 Vuyani Kheswa
M Light Flyweight 20 Giuseppe Lagana
M Minimumweight 1 Muhammad Rachman
M Minimumweight 2 Eagle Kyowa
M Minimumweight 3 Yutaka Niida
M Minimumweight 4 Katsunari Takayama
M Minimumweight 5 Nkosinathi Joyi
M Minimumweight 6 Mawanda Sineko
M Minimumweight 7 Ivan Calderon
M Minimumweight 9 Isaac Bustos
M Minimumweight 11 Thobani Mbangeni
M Minimumweight 12 Lulama Ngqaka
M Minimumweight 13 Omar Soto
M Minimumweight 14 Satoshi Kogumazaka
M Minimumweight 15 Florante Condes
M Minimumweight 16 Pornsawan Kratingdaenggym
M Minimumweight 17 Teruo Misawa
M Minimumweight 18 Omar Soto
M Minimumweight 20 Toshiki Ogawa
Hi Martin,
getting there, getting there. I think you might get some flack about Gary St. Clair being ahead of pacman, barrera and morales and 25-15 Sandro marcos being ahead of Chris John but to quote Stacy Goodman "Those guys are just hatters". (hope you understood that).
Seriously, I think R6 is a bitch and that it is artificial to look for simplicity there. I already proposed an exception for the heavyweight division, because the jump is notoriously difficult. I think a generic times "multiply by three" is probably too simple for this, especially for all the weights. I would propose something like the following.
Some "real life" observations. I think you can throw all the weight classes including and under 140 in a bag and say, that a single jump from one weight class to the next is no big deal. The difference between fighters is in some cases less than two kilos. Hell, I take a decent crap and I come out 1kilo lighter. Broadly I'd say it's harder in the weight classes from 140 to 175 to jump up in weight and stay competitive. 175 to 200 has now become more difficiult and of course the hardest is the jump from cuiser to heavy.
I think that the weight jump R6 might have to be tuned according to which divisions are being jumped. It may be necessary, as with heavy, to have slightly different variations on the R6) algorithm for them e.g. multiply by 2 for the lower weights.
Any thoughts?
conan
p.s. I'd also like to see a trace of Sergio Martinez. It's a bit surprising that he's number two at light middle.
p.p.s I assume the missing ranks are due to fighters being in multiple divisions and in the end being eliminated from one of the lists.
M 11 1999-07-17 014429 014665 W UD 29<- 78 48<- 29
M 11 1999-08-28 014429 017081 W UD 29<- 29 78<- 78
M 11 1999-09-11 014429 007664 W TKO 29<- 29 30<- 30
M 11 2000-04-15 014429 039648 W PTS 24<- 24 536<- 536
M 11 2000-06-16 014429 007827 W KO 24<- 24 48<- 48
M 11 2001-05-19 014429 044531 W TKO 18<- 18 96<- 96
M 11 2001-09-08 014429 038066 W UD 16<- 19 18<- 16
M 11 2001-10-27 014429 007837 W TKO 14<- 16 15<- 14
M 12 2002-02-02 014429 018118 W UD 21<- 27 24<- 21
M 11 2002-04-26 014429 021791 W PTS 9<- 9 555<- 555
M 12 2002-07-12 014429 016953 W PTS 23<- 23 1787<-1787
M 12 2003-02-07 014429 004460 W KO 17<- 17 1153<-1153
M 12 2003-05-09 014429 071690 W UD 19<- 19 1958<-1958
M 12 2003-06-21 014429 014264 W UD 5<- 19 10<- 5
M 12 2003-10-09 014429 014696 W KO 6<- 6 144<- 144
M 12 2004-04-17 014429 014264 W TKO 5<- 5 10<- 10
M 12 2005-01-07 014429 247259 W TKO 4<- 4 1273<-1273
M 12 2005-03-04 014429 014515 W TKO 4<- 4 33<- 33
M 12 2005-10-05 014429 021791 W KO 4<- 4 1358<-1358
M 12 2005-11-04 014429 014514 W TKO 4<- 4 1975<-1975
M 12 2006-04-01 014429 081070 W TKO 3<- 3 650<- 650
M 13 2006-05-26 014429 016953 W UD 9<- 9 234<- 234
- yes, missing ranks belong to boxers assigned to other divisions by Boxrec editors
- boxers are only presented in the disivision assigned to by Boxrec editors
You: ... I think you might get some flack about Gary St. Clair being ahead of pacman, barrera and morales and 25-15 Sandro marcos being ahead of Chris John but to quote Stacy Goodman "Those guys are just hatters". (hope you understood that).
-> St. Clair is very acceptable from the traditional ranking perspective - he simple defeated #1: he got Baloyi, who got Medina, who got Harrison ...
-> Sandro is also clear (but hurts a little bit more, but same reson and consequent) - but he simply defeated Tapia.
No manipulations allowed - see ABC orgs ...
So I have no real problem with the divsion changes at this moment ...
computerrank, to continue the discussion I was having on the other thread, do you really think Shannon Briggs should be favored to beat most top heavyweights? Do you really think Alonzo Butler should be favored to beat guys like Danny Williams and JC Gomez?
When you said that most of these "overrated contenders" beat their lower ranked opponents more than 50% of the time, what are you considering as an "overrated contender"? What opponents are you considering? If you're considering all opponents ranked lower than them, that is basically meaningless because they're fighting BUMS. The system should not get credit for picking an established guy over a BUM. If an up and coming prospect takes 10 fights, and 9 are against bums, which the system correctly predicts that he defeats, and 1 is against a good fighter, who the system incorrectly predicts he defeats, do you consider that a resounding success because the prediction rate was 90%?
You: computerrank, to continue the discussion I was having on the other thread, do you really think Shannon Briggs should be favored to beat most top heavyweights? Do you really think Alonzo Butler should be favored to beat guys like Danny Williams and JC Gomez?
-> I do not feel that. I just explained to you in the other tread.
When you said that most of these "overrated contenders" beat their lower ranked opponents more than 50% of the time, what are you considering as an "overrated contender"?
-> with "overrated" contender I exactly target to your's an others's feelings. Everybody feels, these guys seem to be overrated.
You: What opponents are you considering? If you're considering all opponents ranked lower than them, that is basically meaningless because they're fighting BUMS.
- The system should not get credit for picking an established guy over a BUM. If an up and coming prospect takes 10 fights, and 9 are against bums, which the system correctly predicts that he defeats, and 1 is against a good fighter, who the system incorrectly predicts he defeats, do you consider that a resounding success because the prediction rate was 90%?
-> The evaluation only goes over the launched boxers, as I told you. And these boxers are quite high, most over 1000 points. We only evaluate the best 10 to 15%.
What you also moan is, we shouldn't consider fights, where ratings differ very much. This is no problem as:
- if everything is clear, every rating or ranking would see it on the same side - or it would be a common upset - so it makes no difference
- if it goes to small differences, different ratings and rankings will be on different sides . so it will make the points
- but you cannot, before testing it, decide, what should be in the evaluation and what should not
- outcomes, which are equal for all ratings and rankings only give a higher predicition rate for all the ratings tested - but don't have any effect to the differences between them.
Going to the methods, I used:
I sorted the opponents in rating difference areas of every 50 points. So I looked at opponents rated
- 951 to 1000 points higher
- 901 to 950 points higher
....
- 51 to 100 points higher
- 1 to 50 points higher
- equal to 49 points lower
- 50 to 99 points lower
...
You see the system - and I looked at what result I found for the next opponent - wins and losses.
On this base I found, that the probability to defeat boxers in any area with lower ratings was more than 50%.
I was some nearly 90 % for big differences, but it also was more than 50% for the area - 1 to 50 points lower opponents.
And it got more and more negative, if the opponents were rated higher and higher.
So the message is, the ratings seem to be quite exact in position statistically.
jujigatame wrote:computerrank, to continue the discussion I was having on the other thread, do you really think Shannon Briggs should be favored to beat most top heavyweights? Do you really think Alonzo Butler should be favored to beat guys like Danny Williams and JC Gomez?
When you said that most of these "overrated contenders" beat their lower ranked opponents more than 50% of the time, what are you considering as an "overrated contender"? What opponents are you considering? If you're considering all opponents ranked lower than them, that is basically meaningless because they're fighting BUMS. The system should not get credit for picking an established guy over a BUM. If an up and coming prospect takes 10 fights, and 9 are against bums, which the system correctly predicts that he defeats, and 1 is against a good fighter, who the system incorrectly predicts he defeats, do you consider that a resounding success because the prediction rate was 90%?
I don't think you understand the complexity of the system. In order to be objective, we use the system's success through time rating fighters based on historical bouts. So in essence, we are using patterns through time in order to mold the ratings of today, and going forward.
ALL we have to rate boxers is results... and that is it. Your and everyone else's individual opinions on Alonzo Butler really do not matter. The system cannot tell that he looks like shit in some of his bouts. All the system sees are blow-out UD wins vs. solid journeymen. He could've easily looked dominant in those performances and you would be singing a different tune.
What is bugging everyone, is that the system is so volatile and ratings are affected by almost all bouts, no matter their significance. Even bouts considered as "tune-ups" affect ratings. Should this be so? Statistics agree with that. That's a factor everyone chooses to ignore, or does not know, because "tune-ups" DO matter to fighters, or else they would not take them.
All in all, you can't really get a near-perfect computerized system for traditional ratings OR predictive ratings.. but I think the decision was made that we could get CLOSER to a solid system by only having a goal of prediction.
Last edited by JCS on 14 Aug 2006, 11:49, edited 1 time in total.
You: ... I think you might get some flack about Gary St. Clair being ahead of pacman, barrera and morales and 25-15 Sandro marcos being ahead of Chris John but to quote Stacy Goodman "Those guys are just hatters". (hope you understood that).
-> St. Clair is very acceptable from the traditional ranking perspective - he simple defeated #1: he got Baloyi, who got Medina, who got Harrison ...
-> Sandro is also clear (but hurts a little bit more, but same reson and consequent) - but he simply defeated Tapia.
No manipulations allowed - see ABC orgs ...
So I have no real problem with the divsion changes at this moment ...
Sorry doesn't work. Harrison beat Medina before Baloyi did, so Baloyi must be behind Harrison. Harrison must have been penalised for lack of competition, which is harsh with Victor Polo and Michael Brodie on the record. Or something else is wrong.
Sandro is not clear. Tapia must have taken a ranking dive when he got beaten by Frankie Archuleta. Surely he was still not somewhere near the top?
M 11 1999-07-17 014429 014665 W UD 29<- 78 48<- 29
M 11 1999-08-28 014429 017081 W UD 29<- 29 78<- 78
M 11 1999-09-11 014429 007664 W TKO 29<- 29 30<- 30
M 11 2000-04-15 014429 039648 W PTS 24<- 24 536<- 536
M 11 2000-06-16 014429 007827 W KO 24<- 24 48<- 48
M 11 2001-05-19 014429 044531 W TKO 18<- 18 96<- 96
M 11 2001-09-08 014429 038066 W UD 16<- 19 18<- 16
M 11 2001-10-27 014429 007837 W TKO 14<- 16 15<- 14
M 12 2002-02-02 014429 018118 W UD 21<- 27 24<- 21
M 11 2002-04-26 014429 021791 W PTS 9<- 9 555<- 555
M 12 2002-07-12 014429 016953 W PTS 23<- 23 1787<-1787
M 12 2003-02-07 014429 004460 W KO 17<- 17 1153<-1153
M 12 2003-05-09 014429 071690 W UD 19<- 19 1958<-1958
M 12 2003-06-21 014429 014264 W UD 5<- 19 10<- 5
M 12 2003-10-09 014429 014696 W KO 6<- 6 144<- 144
M 12 2004-04-17 014429 014264 W TKO 5<- 5 10<- 10
M 12 2005-01-07 014429 247259 W TKO 4<- 4 1273<-1273
M 12 2005-03-04 014429 014515 W TKO 4<- 4 33<- 33
M 12 2005-10-05 014429 021791 W KO 4<- 4 1358<-1358
M 12 2005-11-04 014429 014514 W TKO 4<- 4 1975<-1975
M 12 2006-04-01 014429 081070 W TKO 3<- 3 650<- 650
M 13 2006-05-26 014429 016953 W UD 9<- 9 234<- 234
- yes, missing ranks belong to boxers assigned to other divisions by Boxrec editors
- boxers are only presented in the disivision assigned to by Boxrec editors
Sorry missed that post. No need to re-post it. how come Martinez ends at rank 9 and you printed him at position 2?
M 11 1999-07-17 014429 014665 W UD 29<- 78 48<- 29
M 11 1999-08-28 014429 017081 W UD 29<- 29 78<- 78
M 11 1999-09-11 014429 007664 W TKO 29<- 29 30<- 30
M 11 2000-04-15 014429 039648 W PTS 24<- 24 536<- 536
M 11 2000-06-16 014429 007827 W KO 24<- 24 48<- 48
M 11 2001-05-19 014429 044531 W TKO 18<- 18 96<- 96
M 11 2001-09-08 014429 038066 W UD 16<- 19 18<- 16
M 11 2001-10-27 014429 007837 W TKO 14<- 16 15<- 14
M 12 2002-02-02 014429 018118 W UD 21<- 27 24<- 21
M 11 2002-04-26 014429 021791 W PTS 9<- 9 555<- 555
M 12 2002-07-12 014429 016953 W PTS 23<- 23 1787<-1787
M 12 2003-02-07 014429 004460 W KO 17<- 17 1153<-1153
M 12 2003-05-09 014429 071690 W UD 19<- 19 1958<-1958
M 12 2003-06-21 014429 014264 W UD 5<- 19 10<- 5
M 12 2003-10-09 014429 014696 W KO 6<- 6 144<- 144
M 12 2004-04-17 014429 014264 W TKO 5<- 5 10<- 10
M 12 2005-01-07 014429 247259 W TKO 4<- 4 1273<-1273
M 12 2005-03-04 014429 014515 W TKO 4<- 4 33<- 33
M 12 2005-10-05 014429 021791 W KO 4<- 4 1358<-1358
M 12 2005-11-04 014429 014514 W TKO 4<- 4 1975<-1975
M 12 2006-04-01 014429 081070 W TKO 3<- 3 650<- 650
M 13 2006-05-26 014429 016953 W UD 9<- 9 234<- 234
- yes, missing ranks belong to boxers assigned to other divisions by Boxrec editors
- boxers are only presented in the disivision assigned to by Boxrec editors
Sorry missed that post. No need to re-post it. how come Martinez ends at rank 9 and you printed him at position 2?
conan
... as stated, this is a trace over all division for him ... including the changes...
M17 = men heavy
...
M13 = men middle
M12 = men light middle
ALL we have to rate boxers is results... and that is it. Your and everyone else's individual opinions on Alonzo Butler really do not matter. The system cannot tell that he looks like shit in some of his bouts. All the system sees are blow-out UD wins vs. solid journeymen. He could've easily looked dominant in those performances and you would be singing a different tune.
What is bugging everyone, is that the system is so volatile and ratings are affected by almost all bouts, no matter their significance. Even bouts considered as "tune-ups" affect ratings. Should this be so? Statistics agree with that. That's a factor everyone chooses to ignore, or does not know, because "tune-ups" DO matter to fighters, or else they would not take them.
I am well aware that the computer can't tell what Alonzo Butler looks like in his fights. What I'm saying is that a fighter at his level should not be getting any more credit for beating journeymen who are ranked at #200+. Shannon Briggs is the poster child for this syndrome. You have him ranked #6 because he's taken 1000 "tune-ups" despite the fact that he has been defeated handily by all the solid opponents he's faced. All logic points to him getting blown out if he takes on a top fighter, but the system favors him because he's beaten a large volume of bums.
This notion of "objective statistical accuracy" that you've created is a myth, a pipe dream. Once you start defining and tweaking the definition of "predictive rate" and "established fighter", objectivity has gone out the window. You deciding that an established fighter is a guy with more than 7 wins or a guy ranked #300 or higher is as arbitrary as tweaking any of the other numerical constants that the system is based on.
Cobwebcat wrote:I still think the traditional ranking you guys are coming up with is just a form guide if you always promote winner over loser.
If Tiger Woods gets beaten by Nick Faldo in a play-off or matchplay does this make him the better player? NO WAY! He just played better on that given day.
Keep with the prediction formlae if only to stop this nonsense of A beats B therefore A must be ranked higher....that's a form guide nothing more.
You'll have more elephants in your living room than you can cope with if you carry on! :P
... this is why the serious BoxRec Ratings always will be performance ratings...
... this flavour of traditional ratings is only for fun ...
Thanks for the earlier response Martin. I understand that at the lower portions of any weight classes, the results are less statistically meaningful because you have fighters with few fights. But I still would like to see those fighters ranked, because it provides a clue as to what has been accomplished. And you do have them rated, since the "launch" rating is based on results, and their certainly seems to be a differentiation in the way fighters are listed even with the same launch rating (ie fighters at the top of zero rating have better records than fighters at the bottom). If you do not consider them to be rated, then they should probably be not given any rating at all and just listed alphabetically - I don't want that to happen but if they are not rated, then they shouldn't be shown as such.
As to your last post, I think you are far too confident about what makes a "serious" rating and what makes a fun rating. I have to agree with Conan - the entire basis of your methodology is drawn from an arbitrary goal, since the predictive rate guidelines are something you have constructed. It seems to me that you are too rigid in your views - you have two types of ratings and you have constructed formulas for them, but there are many possibilities in between and altogther different. For example, a "traditional" rating system does not have to follow a beats b beats c, therefore a>c. All ELO-based ratings don't work this way, do they?
I agree with you that chasing popular opinion is a hopeless exercise, but at the same time, if the ratings don't look realistic, then the credibility of the ratings is lost, even if they meet your predictive standard. Can you tell me if the variables in the formula are actually statistically significant? Or are you only measuring their signifigance based on the overall predictive rate? If they are not statitsically significant, it seems that the variables will continue to produce outliers which damage the functionality of the ratings, and will not really create something that is truly predictive for future bouts, only for past data.
Thanks for the earlier response Martin. I understand that at the lower portions of any weight classes, the results are less statistically meaningful because you have fighters with few fights. But I still would like to see those fighters ranked, because it provides a clue as to what has been accomplished. And you do have them rated, since the "launch" rating is based on results, and their certainly seems to be a differentiation in the way fighters are listed even with the same launch rating (ie fighters at the top of zero rating have better records than fighters at the bottom). If you do not consider them to be rated, then they should probably be not given any rating at all and just listed alphabetically - I don't want that to happen but if they are not rated, then they shouldn't be shown as such.
I described the separation in the the ratings description - and the 2 parts are clearly separated:
- valid ratings with mostly from just below 1000 points and to the tops
- grades 0 to 15 with a micro differentiation (better records above worse) in the not launched party - this is not statistically valid, but no so bad, I think, for a rough ranking
We want to show every boxer - launched and not launched
- As to your last post, I think you are far too confident about what makes a "serious" rating and what makes a fun rating. I have to agree with Conan - the entire basis of your methodology is drawn from an arbitrary goal, since the predictive rate guidelines are something you have constructed. It seems to me that you are too rigid in your views - you have two types of ratings and you have constructed formulas for them, but there are many possibilities in between and altogther different. For example, a "traditional" rating system does not have to follow a beats b beats c, therefore a>c. All ELO-based ratings don't work this way, do they?
- performance ratings are not my construct - too much honour for me - they are standard method in statistical performance measurement
- prediction rate is the heart of this class of methods, not an arbitrary goal
- ELO method is one of the representations of a performance rating
- ELO rating and Boxrec rating have the same background, they optimize the prediction rate
- by the way, I never could understand, why anybody obeying to logic can moan, that a rating with best correlation with actual outcomes isn't the best; either the rating is constructed to correlate with real outcomes, then it is ok - or it does not, then it is artifical, although it might match the majority feelings. Otherwise you better would organize a poll and use these results.
- yes there are countless possibilities in between, but there is only one criteria you can really trust - and this is matching with the real outcomes, which is nothing else as the prediction rate. Anything else would be compromises between having a quite good correlation with the real predictition rate, but make soma corrections for some "faulty evidences", based on feelings and look - this on cost of prediction rate and correlation with reality.
I agree with you that chasing popular opinion is a hopeless exercise, but at the same time, if the ratings don't look realistic, then the credibility of the ratings is lost, even if they meet your predictive standard. Can you tell me if the variables in the formula are actually statistically significant? Or are you only measuring their signifigance based on the overall predictive rate? If they are not statitsically significant, it seems that the variables will continue to produce outliers which damage the functionality of the ratings, and will not really create something that is truly predictive for future bouts, only for past data.
At the point - credibility - as you say - doesn't mean anything - as this means looking realistic - based on feeling well with it.
The parameterts of the ratings are only implemented, when statisically significant. Basic guidance obeyed is - keep things as simple as possible - but implement all, that improves the prediction rate significantly.
Is is absolute not astonishing for me, that not all probelms (maybe outliers) are solved.
But to end - I very well know the way to integrate most of all critics would be, to cut all not approved contenders to the level they have allready defeated.
But it is evaluated, that this only produces prediction rates and so correleation with real outcomes of up to 70%, wheras the performance method produces up to 80%.
And the prediction rate is improving for recent bouts, as the base (registrated records) seems to be better - so definitely not only valid for the past.
The enlightened background is:
- most people are anxious about no boxer to be rated above some proved level
- but people never seem to see the injustice, to have boxers cut down in ratings and then see them defeating much higher rated boxers - although it could have been regarded by performance ratings in time in most of the cases.
Best regards
Martin
Last edited by computerrank on 14 Aug 2006, 17:11, edited 1 time in total.
- prediction rate is the heart of this class of methods, not an arbitrary goal
What you keep failing to recognize is that "prediction rate" is an artificial construct that YOU defined!!
jujigatame,
I think, it is quite hopeless with us...
- you think, I invented things and use them to my choice
- I think, you cannot or want not really follow the background of the performance ratings - as a generalized method based on prediciton rate
Ok, look, why don't you then explain to me how you decided on the current criteria for what fights are counted in the predictive rate. I want to know how you came to define the criteria.
If you have already done this in a prior post, I apologize, but there is a bit of a language barrier here and some of your posts have been very confusing to me.
