How can Boxrec Rankings change to actually work?

Pretty much everyone agreed boxrecs rankings are shocking. There only taken seriously as a refrence when look at an unknown opponent rankings people outside the top 100 etc.

Examples of crazy ranks;

Derek Chisora No.9 at Heavyweight
Danny Green No.3 at Cruiserweight
Lucian Bute No.1 at Super Middleweight
Kelly Pavlik No.6 at Super Middleweight
Daniel Geale No.3 at Middleweight
Andy Lee No.5 at Middleweight
Garth Wood No.8 at Middleweight
Darren Barker No.9 at Middleweight
Saul Alvarez No.6 at Light Middleweight
Joan Guzman No.6 at Light Welterweight
Ajose Olusegun No.11 at Light Welterweight
Ricky Burns No.8 at Super Featherweight
Guillermo Rigondeaux at No.11 Super Bantamweight



How can they be improved to actually work?

http://boxrec.com/media/index.php/BoxRec_Ratings_Description

 

They can't be improved, a maths formula will never work.

They need to rip them up and start again.

 

I dont understand the maths behind the formula but its places far too much emphasis on winning/losing. Winners get too large a boost & losers to great a deduction, Paul Williams is a good example. Other factors need to have a bigger impact.

Any formula calculating rankings is never going to be perfect but Boxrec's isnt very good. The Duckworth-Lewis in cricket is a good example of how calculations can work in sport. Other sports like tennis & snooker seem to have much better systems than boxing.

 

sorry this might be ignorant but can you find the boxrec formula on the internet?

 

its on the link i posted.

 

Sorry Rob, didnt see that...

I have just looked and it looks quite complicated, in a lot of cases the ratings are quite good but sometimes as already said you get quite random high entries

 

The ones that stick out a mile are Lee at 5, Barker at 9 Bute at 1 and rigondeaux at 11

I mean, before his last fight Lee was ranked number 15 or 16. One laboured win against Craig McEwan and now he's top 5!!!

 

Thats the bottom line. A maths formula, no matter how involved, is the wrong tool for the job.

 

The good thing about using a maths formula is it removes biased opinions, which as we all know mess up the 4 organisations

 

that's a complicated question. Boxrec is based on assigning points for each win/loss. The number of points added (or subtracted) is dependent on the difference in the number of points the two fighters had before the fight.

So if a fighter with a lot of points beats a fighter with very few points, the former doesn't gain anything. If the situation is reversed, the fighter scoring the upset gets a lot of points.

The system can be unstable (see Garth Wood) because it can assign many points to a fighter who scored an unlikely upset.

When Garth Wood beat Mundine he shot from 141 to 339, while Mundine feel from 341 to 205. So Wood essentially took Mundine's points. This is too drastic a reversal, obviously.

 

Picking the names out of a hat removes biased opinions.

 

But then how does Bute get to No.1 When he has been fighting much weaker opposition than Froch for example??

 

Because.....

1. Froch lost to Kessler
2. Froch has won three of his last four by decision, one split.
3. Bute has won all his recent fights by stoppage

The problem is that you can shut an opponent out, yet gain less points than another fighter would by losing 11 rounds and winning by stoppage with a second to go.

 

Because he lost to Kessler and Kessler took all his points. Kessler had lost all his points to Ward, so the point disparity between Froch Kessler was large.

Meanwhile Bute is accumulating a small number of points every time he wins.

So the issue is that the scoring system is unstable when there are large points disparities and the underdog wins. This instability then propagates (Ward -> Kessler -> Froch).

 

They need to employ someone like me to do them (I'm available for £100k a year).