BoxRec

Posted: **02 Nov 2008, 11:30**

I am surprised Raheem/Funeka were unchanged w/ this update...

Posted: **02 Nov 2008, 12:32**

JCS wrote:I am surprised Raheem/Funeka were unchanged w/ this update...

???

Posted: **02 Nov 2008, 13:07**

computerrank wrote:
JCS wrote:I am surprised Raheem/Funeka were unchanged w/ this update...
???

Zahir Raheem vs. Ali Funeka. Raheem traveled to South Africa and got KOd by Ali Funeka. I would expect the promoter handicap update to put less weight on that matchup. Maybe we need a country-based handicap as well?

Posted: **02 Nov 2008, 14:09**

The rankings seem to have taken another ... uh ... interesting turn today. I just don't get all these changes. It's like changes just for the sake of changes. They don't seem to be improving anything at all.

Posted: **03 Nov 2008, 05:13**

JCS wrote:
computerrank wrote:
JCS wrote:I am surprised Raheem/Funeka were unchanged w/ this update...
???

Zahir Raheem vs. Ali Funeka. Raheem traveled to South Africa and got KOd by Ali Funeka. I would expect the promoter handicap update to put less weight on that matchup. Maybe we need a country-based handicap as well?

This was a multiple promoter event and so it was neutral.

The country-based handicap shows a lot of difficulties, as the country is somehow connected to the boxer, but not in a very defined way. Sometimes it means nationality, sometimes it is connected to the residence, sometimes it is connected to the country, where the boxer is registrated. - And although - it would need a lot of additional computer ressources.

Posted: **03 Nov 2008, 08:58**

There are multiple promoters, but how many times had each of the boxers fought under each promoter?

Posted: **03 Nov 2008, 09:14**

JCS wrote:There are multiple promoters, but how many times had each of the boxers fought under each promoter?

Multiple promoters entries are currently assumed to be neutral by default.

But it looks like there might be an underlying problem anyway ...

- Funeka with Eyethu Boxing Promotions the 7th time effectively
- neither opponent with Branco Sports Productions before

... but, there seem to exist multiple IDs for one promototion company ...

Code: Select all

Funeka

+------------+------------+-------------------+---------------------------+
| date       | first_name | last_name         | company_name              |
+------------+------------+-------------------+---------------------------+
| 2004-10-23 | Eyethu     | Boxing Promotions | NULL                      | 
| 2006-05-27 | Eyethu     | Boxing Promotions | NULL                      | 
| 2006-07-22 | Eyethu     | Boxing Promotions | NULL                      | 
| 2006-10-07 | Eyethu     | Boxing Promotions | NULL                      | 
| 2007-04-01 | Eyethu     | Boxing Promotions | NULL                      | 
| 2007-09-23 | Mzi        | Mnguni            | NULL                      | 
| 2008-07-05 | Branco     | Milenkovic        | Branco Sports Productions | 
| 2008-07-05 | Mzi        | Mnguni            | Eyethu Boxing Promotions  | 
+------------+------------+-------------------+---------------------------+


Raheem

+------------+------------+--------------------+----------------------------+
| date       | first_name | last_name          | company_name               |
+------------+------------+--------------------+----------------------------+
| 1996-11-16 | Bob        | Arum               | Top Rank                   | 
| 1997-01-10 | Bob        | Arum               | Top Rank                   | 
| 1997-02-22 | Bob        | Arum               | Top Rank                   | 
| 1997-03-12 | Bob        | Arum               | Top Rank                   | 
| 1997-09-19 | Bob        | Arum               | Top Rank                   | 
| 1997-12-06 | Bob        | Arum               | Top Rank                   | 
| 1998-05-05 | Bob        | Arum               | Top Rank                   | 
| 1998-09-25 | Bob        | Arum               | Top Rank                   | 
| 2000-05-02 | Russell    | Peltz              | Peltz Boxing Promotions    | 
| 2000-09-23 | Art        | Pelullo            | Banner Promotions          | 
| 2001-08-24 | Terry      | Lane               | Let's Get It On Promotions | 
| 2002-03-09 | Ron        | Kern               | NULL                       | 
| 2002-03-30 | Holden     | Productions        | NULL                       | 
| 2002-10-11 | Holden     | Productions        | NULL                       | 
| 2003-02-24 | Holden     | Productions        | NULL                       | 
| 2003-03-28 | Holden     | Productions        | NULL                       | 
| 2003-07-25 | Holden     | Productions        | NULL                       | 
| 2004-04-22 | Holden     | Productions        | NULL                       | 
| 2004-07-17 | Kathy      | Duva               | Main Events                | 
| 2005-02-26 | Bob        | Arum               | Top Rank                   | 
| 2005-09-10 | Bob        | Arum               | Top Rank                   | 
| 2006-04-29 | Art        | Pelullo            | Banner Promotions          | 
| 2007-05-18 | Hall       | of Fame Promotions | NULL                       | 
| 2008-01-04 | Tony       | Holden             | Holden Productions         | 
| 2008-07-05 | Mzi        | Mnguni             | Eyethu Boxing Promotions   | 
| 2008-07-05 | Branco     | Milenkovic         | Branco Sports Productions  | 
+------------+------------+--------------------+----------------------------+

Posted: **03 Nov 2008, 09:17**

Yes, even though it is multiple promoters, I should think that Funeka should get the advantage since it was his 7th with one of them.

If Raheem had a relationship with the other.. at that point it would become neutral.

This is one of those data problems I suspected.

Posted: **03 Nov 2008, 09:22**

JCS wrote:Yes, even though it is multiple promoters, I should think that Funeka should get the advantage since it was his 7th with one of them.

If Raheem had a relationship with the other.. at that point it would become neutral.

This is one of those data problems I suspected.

... both issues will be treated ...

Posted: **03 Nov 2008, 18:03**

A release with corrected treatment of multiple promoter shows is proposed for launch.

Multiple promoter IDs will be tidied up step by step.

Posted: **03 Nov 2008, 19:35**

What will the process be when there is more than one promoter?

Are there instances where there are more than two? What then?

Posted: **04 Nov 2008, 01:12**

JCS wrote:What will the process be when there is more than one promoter?

Are there instances where there are more than two? What then?

In case of more than 1 promoter involved - as in general - for every party the promoter with most occurences on it's own side is regarded.

Posted: **06 Nov 2008, 10:51**

Did a ratings change take place? I didn't see the Funeka/Raheem movement that I expected.

Posted: **06 Nov 2008, 11:30**

JCS wrote:Did a ratings change take place? I didn't see the Funeka/Raheem movement that I expected.

The ratings algorithm is fixed for multiple promoter bouts now - and launched just these minutes.

But multiple promoter IDs must be tidied up step by step - by the editors. There was only 1 data field for the promoter initially. Meanwhile this was expanded to promoter's CEO_first_name, CEO_last_name and company_name - and the contents are quite mixed up sometimes.

Give them a break ...

Posted: **06 Nov 2008, 11:46**

Martin,

Did Juan Manuel Marquez/Chris John have the promoter fix?

Posted: **06 Nov 2008, 11:47**

This "fix" seems to have made things worse if anything. Povetkin and Ibragimov pushed farther from the top 10 while Tony Thompson is sitting pretty at #5. Hasegawa displaced from the #1 spot again. It is just baffling to me how anyone can accept this as an "improvement".

Posted: **06 Nov 2008, 12:14**

jujigatame wrote:This "fix" seems to have made things worse if anything. Povetkin and Ibragimov pushed farther from the top 10 while Tony Thompson is sitting pretty at #5. Hasegawa displaced from the #1 spot again. It is just baffling to me how anyone can accept this as an "improvement".

I personally don't have a problem with that Heavyweight situation. Hasegawa dropping from #1 is a bit troublesome... but its basically tied up as the point difference is negligible.

Posted: **06 Nov 2008, 12:16**

JCS wrote:Martin,

Did Juan Manuel Marquez/Chris John have the promoter fix?

The program is fixed.

I don't know, whether all promoters in all bouts of both boxers are normalized to 1 promoter ID each, yet.

Posted: **06 Nov 2008, 14:05**

JCS wrote:
jujigatame wrote:This "fix" seems to have made things worse if anything. Povetkin and Ibragimov pushed farther from the top 10 while Tony Thompson is sitting pretty at #5. Hasegawa displaced from the #1 spot again. It is just baffling to me how anyone can accept this as an "improvement".
I personally don't have a problem with that Heavyweight situation. Hasegawa dropping from #1 is a bit troublesome... but its basically tied up as the point difference is negligible.

Hasegawa should not tied with anyone at BW. He is a clear #1. Penalosa loses every round against Gonzalez, then wins via a body shot KO and gets a big magical boost over Hasegawa just because Golden Boy was the promoter? Ridiculous.

As for the HW situation, who has Thompson beaten to be ranked so highly? His biggest win is Timor Ibragimov who has never been within sniffing distance of the top 10. Povetkin beat Chambers who was borderline top 10, and Sultan beat Briggs who was top 10 at the time, and made it to the final bell against Klitchko with can't be said for Thompson. It looks to me like Thompson is just getting bonus points for completely unremarkable performances on other promoters' cards.

These changes have completely impugned the credibility of the system.

Posted: **06 Nov 2008, 14:39**

I'd wager to say Thompson's biggest win was over Krasniqi, who was still a legit fringe heavyweight contender at the time of the bout.

I guess what I'm saying is I understand why the fighters are ranked where they are considering the changes that have been made. Hasegawa has been a homebody his entire career.

Posted: **06 Nov 2008, 15:26**

I think it's painfully obvious that this new system is horribly unfair to fighters that fight at home but still win convincingly against quality opposition. These rankings look much worse than they did a month ago and if we did a poll of this site's users I doubt more than 5-10% would disagree with me.

Posted: **26 Nov 2008, 07:30**

A repeat of the post in the Miranda/Banks thread.

JCS wrote:The home promoter advantage factor is based on research done by a sports science journal with regards to home country advantage.

They broke down the 1900s (I believe) into three parts. The first two of the three had no significant differences between home and away fighters of comparable records. They blamed this on shoddy data.

However, the third showed a significant advantage for the home fighter. This is how the promoter home advantage originated. Computerrank is taking those factors found in the study, splitting them in half, and applying them to the pre-bout ratings. As you can see, its application is a bit different.

32% DEC
23% STOPPAGE

Well thanks for the explanation. I will attempt not to shoot the messenger as I dont have all the facts.

The following comments are without exact knowledge of the study. It may or may not have been done well. Firstly I would dare say however, that the main reason the first two thirds of the 1900s didn’t show any significant effect was because the majority of fights were hometown fights for both fighters. People just didn’t travel those days. The number of international fights was a mere fraction of today’s. Secondly, I wonder where they got their data from? Thirdly, there is the difficulty in comparing similar fighters. Is JC Chapman a similar fighter to David Haye? The only real assessment of equality would be their boxrec or IBO rankings at the time of the fight. I doubt these researchers had that available.

Regardless of how the results were obtained, you have to think about what you are doing with the results. By dividing them by two and reducing the value of the result, you are effectively saying that the result is dependent on the venue/promoter. However, you don’t have a control group for this statement. If the rematch between the fighters would take place in the other fighters home city, then you might have a comparison. If a third fight would take place on a neutral venue/card, then you would have some real data to compare. It’s like saying smoking causes cancer, without measuring the cancer rates in non-smoker’s. You just don’t know without the comparative group.

Look, I’m not against experimenting to improve the ratings. However, you have to keep the results plausible. I bet well over 90%, maybe even 100% of boxrec posters would answer the following question with ‘yes’.

Should fighter A still be ranked ahead of fighter B, if fighter A has fought and decisively beaten fighter B in the last twelve months and has only one subsequent loss to a p4p top 30 fighter, whilst fighter B hasn’t fought anyone in the top 150 of his division?

Of course the only answer to this question is ‘yes’. It’s obvious and you guys know it, and don’t try and wriggle out of it. In this case, it’s the story of Miranda and Banks, but it could be another example.

If you remember correctly, there was a time where you had the predictive ratings, which were roundly dismissed on this forum as worthless. Then the algorithm was changed back to a “proven in the ring” algorithm, and practically all the serious complaints stopped. Now like jugi says, weird ratings are appearing and the threads are starting again. Take that as a feedback signal to the change, even if you want to deny the answer to the question above.

Cheers

conan

Posted: **26 Nov 2008, 07:47**

Without wanting to get distracted about jugi's very legitimate complaints, does the round of the KO influence the ratings? I mean most people would rate a KO1 higher than a KO12. Especially if it was against the same opponent.

KO12 means that the opponent was at least good enough to hang in there for 12 rounds. A KO in the first four rounds usually indicates that the match was one sided, or at least, the fighter who was knocked out does not have the chin for a high ranking OR that the fighter with the KO punch has the power to be a force in that division.

cheers

conan

Posted: **26 Nov 2008, 10:29**

conan_the_cribber wrote:Without wanting to get distracted about jugi's very legitimate complaints, does the round of the KO influence the ratings? I mean most people would rate a KO1 higher than a KO12. Especially if it was against the same opponent.

KO12 means that the opponent was at least good enough to hang in there for 12 rounds. A KO in the first four rounds usually indicates that the match was one sided, or at least, the fighter who was knocked out does not have the chin for a high ranking OR that the fighter with the KO punch has the power to be a force in that division.

cheers

conan

interesting

Posted: **26 Nov 2008, 10:30**

KO round consideration was factored into the old prediction system.

BoxRec

Ratings - please read before commenting - Archived

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings

Re: Boxrec Ratings - Read first before commenting on the ratings