Page 1 of 1

New approach to ranking heavyweights

Posted: 17 Feb 2016, 09:19
by Simon Lock
I've come up with a way of ranking heavyweights based on their last 5 fight. A worked example is below.


To get a score for Tyson Fury, we look at his last 5 fights.

Wins against: Steve Cunningham, Joey Abell, Dereck Chisora, Christian Hammer, Wladimir Klitschko


We assign a score out of 125 to each opponent. Let's look at Christian Hammer as an example.


In his last 5 fight before losing to Fury, Hammer's record was as follows.

Wins against: Oleksiy Mazikin, Leif Larsen, Kevin Johnson, Konstantin Airich, Irineu Beato Costa Junior


We assign a score out of 25 to each of Hammer's opponents. Let's look at Konstantin Airich as an example.

In his last 5 fight before losing to Hammer, Airich's had 3 losses and 2 wins. For each win, we give a score out of 5 based on the last 5 fights, 1 for a win and 0.5 for a draw.

Varol Vekiloglu (WWWWW) - 5 points
Ondrej Pala (WWWWW) - 5 points


So that's a total of 10/25 for Hammer's win over Airich.


Doing the same for the rest of Hammer's opponents gives the following:

Oleksiy Mazikin (W) - 2 points
Leif Larsen (W) - 6 points
Kevin Johnson (W) - 11 points
Konstantin Airich (W) - 10 points
Irineu Beato Costa Junior (W) - 14 points

Added up, that's a score of 43/125 for Fury's win over Hammer.


Doing the same for the rest of Fury's opponents gives the following:

Steve Cunningham - 18 points
Joey Abell - 10 points
Dereck Chisora - 42 points
Christian Hammer - 43 points
Wladimir Klitschko - 81 points

Added up, that's a total score of 193 points for Tyson Fury



I've done this for every single one of the top 30 heavyweights on Boxrec, giving the following Top 10 rankings.

Tyson Fury - 193 points
Wladimir Klitschko - 189 points
Deontay Wilder - 135 points
Luis Ortiz - 107 points
Alexander Povetkin - 101 points
Antonio Tarver - 91 points
Erkan Teper - 81 points
Bermane Stiverne - 81 points
Bryant Jennings - 79 points
Charles Martin - 67 points

Re: New approach to ranking heavyweights

Posted: 17 Feb 2016, 09:21
by Ricky_
You've got too much time on your hands.

Re: New approach to ranking heavyweights

Posted: 17 Feb 2016, 10:12
by jezzamundo
Ricky_ wrote:You've got too much time on your hands.
Yep, although this is an interesting way of doing rankings, it won't be as good as the regular BoxRec ratings as it doesn't take margin of victory into account.

Re: New approach to ranking heavyweights

Posted: 17 Feb 2016, 10:19
by Ricky_
jezzamundo wrote:
Ricky_ wrote:You've got too much time on your hands.
Yep, although this is an interesting way of doing rankings, it won't be as good as the regular BoxRec ratings as it doesn't take margin of victory into account.

The best form of rankings is human judgement. It's just a matter of opinion anyway until 2 guys actually fight, and mechanical scoring systems are left open to abuse.

Re: New approach to ranking heavyweights

Posted: 17 Feb 2016, 11:02
by cold187
good effort lad
but reality is there is no 100% system due to number of belts and ducking

the best way is just keep depth short to 5-7 maximum and that separated bets from rest

fury, wladimir, povetkin, wilder, Ortiz, and pulev.
the rest can suck it

Re: New approach to ranking heavyweights

Posted: 17 Feb 2016, 15:35
by TBEwasLangford
Simon Lock wrote:I've come up with a way of ranking heavyweights based on their last 5 fight. A worked example is below.


To get a score for Tyson Fury, we look at his last 5 fights.

Wins against: Steve Cunningham, Joey Abell, Dereck Chisora, Christian Hammer, Wladimir Klitschko


We assign a score out of 125 to each opponent. Let's look at Christian Hammer as an example.


In his last 5 fight before losing to Fury, Hammer's record was as follows.

Wins against: Oleksiy Mazikin, Leif Larsen, Kevin Johnson, Konstantin Airich, Irineu Beato Costa Junior


We assign a score out of 25 to each of Hammer's opponents. Let's look at Konstantin Airich as an example.

In his last 5 fight before losing to Hammer, Airich's had 3 losses and 2 wins. For each win, we give a score out of 5 based on the last 5 fights, 1 for a win and 0.5 for a draw.

Varol Vekiloglu (WWWWW) - 5 points
Ondrej Pala (WWWWW) - 5 points


So that's a total of 10/25 for Hammer's win over Airich.


Doing the same for the rest of Hammer's opponents gives the following:

Oleksiy Mazikin (W) - 2 points
Leif Larsen (W) - 6 points
Kevin Johnson (W) - 11 points
Konstantin Airich (W) - 10 points
Irineu Beato Costa Junior (W) - 14 points

Added up, that's a score of 43/125 for Fury's win over Hammer.


Doing the same for the rest of Fury's opponents gives the following:

Steve Cunningham - 18 points
Joey Abell - 10 points
Dereck Chisora - 42 points
Christian Hammer - 43 points
Wladimir Klitschko - 81 points

Added up, that's a total score of 193 points for Tyson Fury



I've done this for every single one of the top 30 heavyweights on Boxrec, giving the following Top 10 rankings.

Tyson Fury - 193 points
Wladimir Klitschko - 189 points
Deontay Wilder - 135 points
Luis Ortiz - 107 points
Alexander Povetkin - 101 points
Antonio Tarver - 91 points
Erkan Teper - 81 points
Bermane Stiverne - 81 points
Bryant Jennings - 79 points
Charles Martin - 67 points
Interesting idea. Obviously all measuring tools have there flaws but good system. Could you publish the full list of 30 please?

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 05:20
by Lennox
Why that does not work is that there are plenty of average fighters with lots of wins against not great opposition.

A win against a Kevin Johnson would mean less than over some 5-0 fighter from Slovakia.

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 09:14
by Simon Lock
Lennox wrote:Why that does not work is that there are plenty of average fighters with lots of wins against not great opposition.

A win against a Kevin Johnson would mean less than over some 5-0 fighter from Slovakia.
A win against Kevin Johnson (WLLLL)would be worth 15 points, as he has lost 4 of his last 5, and the win over Solomon Haumono was worth 15/25.

A win against Alexander Dimitrenko (WWWWW) would be worth 8 points, as his last 5 wins break down as follows:

Ivica Perkovic (0/25)
Patryk Kowoll (0/25)
Zoltan Csala (8/25)
Milos Dovedan (0/25)
Drazan Janjanin (0/25)

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 09:36
by gp.
Ricky_ wrote:

The best form of rankings is human judgement.

If the human in question is me, it certainly isn't.

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 09:48
by Oiky
clever, different way to rank, but you can't beat seeing two men fight, thats the best way to rank a mans levels

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 12:35
by Lennox
Simon Lock wrote:
Lennox wrote:Why that does not work is that there are plenty of average fighters with lots of wins against not great opposition.

A win against a Kevin Johnson would mean less than over some 5-0 fighter from Slovakia.
A win against Kevin Johnson (WLLLL)would be worth 15 points, as he has lost 4 of his last 5, and the win over Solomon Haumono was worth 15/25.

A win against Alexander Dimitrenko (WWWWW) would be worth 8 points, as his last 5 wins break down as follows:

Ivica Perkovic (0/25)
Patryk Kowoll (0/25)
Zoltan Csala (8/25)
Milos Dovedan (0/25)
Drazan Janjanin (0/25)
That highlights the problem Simon, going back 5 fights, its not really enough. A win against Dimitrenko is better than Johnson, but with the straight 5 fight criteria is not reflective in a fair assessment of Dimitrenko, go back 10 fights and his good wins are shown. Equally you can make a case it is a long time ago, which in many ways a better way to factor. With the PBO that is how we rate fighters, recent (12 month) compliant (18 month-24-30-36 month) and then all time. Most rankings show up a silly, yours is Tarver at 8 or whatever he is. When you list 20 30 40 or 100 there will be more, but the principle is not so far away from our rankings of each fighter having a value. Ours have bars and tags so sometimes 5 or 6 fighters will move along together all being tagged (if you beat a fighter you are ranked above him for a year). Our fighters values also have age decay, multiple loss accelaration, division change reduction. www.premierboxingorganisation.com they are updated monthly. I do agree that a subjective top 10 is going to be better than a computer based or formala based ranking, you can get nearly there but there is always something not quite right.

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 13:15
by greg
@Lennox
..just checked out your site, didn't find P4P rankings..any reason for it?

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 13:33
by Lennox
greg wrote:@Lennox
..just checked out your site, didn't find P4P rankings..any reason for it?
Found that they don't really work with our system as a fighter is penalized by moving up in weight class. Therefore they favour fighters that stay in the same division, so it all got a bit skewed. The only thing you could do using our ratings is to count the last 20 fights and the lowest total number is the best fighter. A win would have to count as 100, not sure how to count a loss or a draw though or even a NC.
Floyd Mayweather absolutely won that btw but I never really looked deeply into it, perhaps I should.

Re: New approach to ranking heavyweights

Posted: 18 Feb 2016, 14:09
by Simon Lock
Lennox wrote:
Simon Lock wrote:
Lennox wrote:Why that does not work is that there are plenty of average fighters with lots of wins against not great opposition.

A win against a Kevin Johnson would mean less than over some 5-0 fighter from Slovakia.
A win against Kevin Johnson (WLLLL)would be worth 15 points, as he has lost 4 of his last 5, and the win over Solomon Haumono was worth 15/25.

A win against Alexander Dimitrenko (WWWWW) would be worth 8 points, as his last 5 wins break down as follows:

Ivica Perkovic (0/25)
Patryk Kowoll (0/25)
Zoltan Csala (8/25)
Milos Dovedan (0/25)
Drazan Janjanin (0/25)
That highlights the problem Simon, going back 5 fights, its not really enough. A win against Dimitrenko is better than Johnson, but with the straight 5 fight criteria is not reflective in a fair assessment of Dimitrenko, go back 10 fights and his good wins are shown. Equally you can make a case it is a long time ago, which in many ways a better way to factor. With the PBO that is how we rate fighters, recent (12 month) compliant (18 month-24-30-36 month) and then all time. Most rankings show up a silly, yours is Tarver at 8 or whatever he is. When you list 20 30 40 or 100 there will be more, but the principle is not so far away from our rankings of each fighter having a value. Ours have bars and tags so sometimes 5 or 6 fighters will move along together all being tagged (if you beat a fighter you are ranked above him for a year). Our fighters values also have age decay, multiple loss accelaration, division change reduction. http://www.premierboxingorganisation.com they are updated monthly. I do agree that a subjective top 10 is going to be better than a computer based or formala based ranking, you can get nearly there but there is always something not quite right.
Agree about Tarver, I think the problem with him is that his score includes some Cruiserweight fights which should probably be excluded.

Re: New approach to ranking heavyweights

Posted: 20 Feb 2016, 18:36
by Lackeos
This pyramid rating system is more simplified and more flawed than an elo-style system, or boxrec's nearly elo-style. The more that you treat every fighter as a 1 or a 0, the less nuanced the system is, and the more information you're going to lose. Similarly, the 5-fight cutoff arbitrarily cuts-off all information beyond a certain point. It's not an utterly terrible pencil-and-paper system, but it's much more flawed than what computers are capable of.

Re: New approach to ranking heavyweights

Posted: 21 Feb 2016, 02:57
by Mexi-Box
Number rankings NEVER work. Just look at the messed up BoxRec system. No doubt, I love BoxRec, but the ranking system is so off.

Re: New approach to ranking heavyweights

Posted: 24 Feb 2016, 01:55
by Like a Boss
Simon Lock wrote:I've come up with a way of ranking heavyweights based on their last 5 fight. A worked example is below.


To get a score for Tyson Fury, we look at his last 5 fights.

Wins against: Steve Cunningham, Joey Abell, Dereck Chisora, Christian Hammer, Wladimir Klitschko


We assign a score out of 125 to each opponent. Let's look at Christian Hammer as an example.


In his last 5 fight before losing to Fury, Hammer's record was as follows.

Wins against: Oleksiy Mazikin, Leif Larsen, Kevin Johnson, Konstantin Airich, Irineu Beato Costa Junior


We assign a score out of 25 to each of Hammer's opponents. Let's look at Konstantin Airich as an example.

In his last 5 fight before losing to Hammer, Airich's had 3 losses and 2 wins. For each win, we give a score out of 5 based on the last 5 fights, 1 for a win and 0.5 for a draw.

Varol Vekiloglu (WWWWW) - 5 points
Ondrej Pala (WWWWW) - 5 points


So that's a total of 10/25 for Hammer's win over Airich.


Doing the same for the rest of Hammer's opponents gives the following:

Oleksiy Mazikin (W) - 2 points
Leif Larsen (W) - 6 points
Kevin Johnson (W) - 11 points
Konstantin Airich (W) - 10 points
Irineu Beato Costa Junior (W) - 14 points

Added up, that's a score of 43/125 for Fury's win over Hammer.


Doing the same for the rest of Fury's opponents gives the following:

Steve Cunningham - 18 points
Joey Abell - 10 points
Dereck Chisora - 42 points
Christian Hammer - 43 points
Wladimir Klitschko - 81 points

Added up, that's a total score of 193 points for Tyson Fury



I've done this for every single one of the top 30 heavyweights on Boxrec, giving the following Top 10 rankings.

Tyson Fury - 193 points
Wladimir Klitschko - 189 points
Deontay Wilder - 135 points
Luis Ortiz - 107 points
Alexander Povetkin - 101 points
Antonio Tarver - 91 points
Erkan Teper - 81 points
Bermane Stiverne - 81 points
Bryant Jennings - 79 points
Charles Martin - 67 points
How would Mark de Mori have rated points-wise prior to David Haye obliterating him inside one round?