Ratings - please read before commenting - Archived

Lennox · Post by **Lennox** » 25 Apr 2006, 11:04

You clearly know nothing about Boxing. boxrec ratings are the let down of the site. Ray Mercer number 314 and Mike McDuffie number 259. Give me a break McDuffie has only had two fights and he lost them both. The formula stinks.

JCS · Post by **JCS** » 25 Apr 2006, 11:08

Lennox wrote:You clearly know nothing about Boxing. boxrec ratings are the let down of the site. Ray Mercer number 314 and Mike McDuffie number 259. Give me a break McDuffie has only had two fights and he lost them both. The formula stinks.

McDuffie's problem is due to someone entering the judge's scores in reverse in his 2nd fight thus causing a mathematical error when considering the decision's closeness.

Errors like these do not help the cause much.. however, Martin, when it is a UD, you could probably just do a mean difference between the cards and assume its for the guy who got the W.

Post by **John** » 25 Apr 2006, 11:11

I agree with German B, what have the IBO got to hide ? for all we know their ratings could be produced by a room full of monkeys sitting at typewriters

They have no claim to objectivity when their ratings cannot be independently checked and certainly should not be taken seriously by anybody in the boxing industry.

Lennox · Post by **Lennox** » 25 Apr 2006, 11:18

Check out other 2-0 fighters ranked higher than Ray Mercer. One is still doing 2 minute round fights. The rankings are littered with tripe stats. Shannon Briggs not top 100 and a lot of his opponents are really ranked shit. Abarham Okine is ranked 300+.
I dont believe there is anyone who understands boxing that would think Boxrecs rankings are better than IBO. I think at the top20 there is not so much of a problem, just an occasional stinker, but the lower 51-100 really suck.

Lennox · Post by **Lennox** » 25 Apr 2006, 11:22

JohnShep wrote:I agree with German B, what have the IBO got to hide ? for all we know their ratings could be produced by a room full of monkeys sitting at typewriters
They have no claim to objectivity when their ratings cannot be independently checked and certainly should not be taken seriously by anybody in the boxing industry.

They are the ones the boxing industry take note about. Why cant there ratings be independently checked?

JCS · Post by **JCS** » 25 Apr 2006, 11:29

Lennox wrote:
JohnShep wrote:I agree with German B, what have the IBO got to hide ? for all we know their ratings could be produced by a room full of monkeys sitting at typewriters
They have no claim to objectivity when their ratings cannot be independently checked and certainly should not be taken seriously by anybody in the boxing industry.
They are the ones the boxing industry take note about. Why cant there ratings be independently checked?

In my opinion, an established fighter like Mercer, should not end up with a rating under 1,000 because he has not lost his way to that point. Same goes for Briggs.

The base rating for a fighter is 1,000... I believe Martin assesses some kind of early career penalty in his system.. but beyond that, I don't understand why any fighter who's winning, even vs. journeymen.. should end up with a rating below 1,000. Perfect example is Mercer. Regardless of his past 4 years... unless he's only winning vs. guys like Danny Wofford, he should definitely be in the Top 200 at the very least.

When Martin gets some time hopefully he can look at giving fighters SOME credit for their rating previous to 4 years ago. Also improvements can be made to draw results... and in a few other areas that I've been exploring.

Lennox · Post by **Lennox** » 25 Apr 2006, 11:35

That seems sensible JCs. Would explain why Shannon Briggs never got much credit/points for beating Mercer.

Latuch! · Post by **Latuch!** » 26 Apr 2006, 06:29

Lennox wrote:Boxrec Ratings are a disgrace.

You are wrong. Boxrec ratings are very respected.

In my country all newspapers, boxing webages and boxing live commentators prefer to give boxrec ratings to any other alphabet ratings.

Lennox · Post by **Lennox** » 26 Apr 2006, 08:55

Latuch.. but surely not better than IBO ratings. If you say yes check. Boxrec ratings are a disgrace. Look at Cruiserweight.
Johnny Nelson number 16
Virgil Hill number 31
If you still say yes you may need some help!!!!!
If you need further evidence look at some of the boxers that at rated after 2 fights...they are ahead oof some world champions!!!!
Boxrec ratings are okay only for the first 25 or so, after that they do not have the boxers in a good order.
The good people at boxrec should hold their heads in shame.

Post by **Marina** » 26 Apr 2006, 09:19

The IBO place the emphasis on making their ratings seem objective and accurate and boxrec place the emphasis on making thier ratings objective and accurate.
Boxrec publishes their rating formula, what are the IBO hiding and why ?
If the IBO are so good why don't they release the formula and then computerank and jcs83md can verify the IBO ratings as being fair and accurate. Maybe you could even collaborate and produce a common rating formula you could both use.
But as things are boxrec now has credibility and the IBO doesnt.

JCS · Post by **JCS** » 26 Apr 2006, 09:29

Lennox wrote:Latuch.. but surely not better than IBO ratings. If you say yes check. Boxrec ratings are a disgrace. Look at Cruiserweight.
Johnny Nelson number 16
Virgil Hill number 31
If you still say yes you may need some help!!!!!
If you need further evidence look at some of the boxers that at rated after 2 fights...they are ahead oof some world champions!!!!
Boxrec ratings are okay only for the first 25 or so, after that they do not have the boxers in a good order.
The good people at boxrec should hold their heads in shame.

Both those fighters should be ranked higher. I definitely agree with you. You must remember though, any changes made in the rating system will affect every fighter's rating, past and present, so these changes must be made with care.

As Martin gets some time, I'd love to collaborate with him further and help improve the ratings. His current system is too cryptic for me to tackle alone, so I can hopefully play the role of an educated idea man.

Rather than saying the ratings suck, why not look at the encyclopedia where the official formula/procedures are broken apart and explained in decent detail. Use this as a forum to voice your specific opinions.

I would like to see the following tested.

1. Giving fighters partial credit for their activity previous to 4 years from present date.

2. Increase impact of draws. Create alternate routine for technical draws (with much smaller impact, close to the current draw's impact).

3. Vary the weight of a TKO, or KO depending on round it occurred in. Example: Make a TKO2 mean more than a TKO12.

4. Increase the weight of a blow-out decision. A decision with a mean difference of more than 8 points per card (Considering 10-point must) should increase impact. Only use this for current scoring system as older/alternate scoring systems would be negatively affected.. unless a percentage based system is used.

5. Lessen the hit of the 1-year inactivity period as it appears to be too much.

Lennox · Post by **Lennox** » 26 Apr 2006, 11:41

germanbee wrote:The IBO place the emphasis on making their ratings seem objective and accurate and boxrec place the emphasis on making thier ratings objective and accurate.
Boxrec publishes their rating formula, what are the IBO hiding and why ?
If the IBO are so good why don't they release the formula and then computerank and jcs83md can verify the IBO ratings as being fair and accurate. Maybe you could even collaborate and produce a common rating formula you could both use.
But as things are boxrec now has credibility and the IBO doesnt.

I do not see the logic here. i am looking at boxrec ratings which are very poor and i am looking at IBO ratiings which are very good. Let us look at cruiserweight. Is IBO better than Boxrec as a fair set of boxing rankings? a straight yes or no? who cares if you publish your formula and IBO does not, surely you look at the ratings and say accurate or inaccurate. http://www.iboboxing.com
IBO ratings are used by all the boxing promoters and TV companies to make the matches, all the boxing promoters have their ratings, why are they not credible and boxec are?

Post by **John** » 26 Apr 2006, 12:00

Just like all the other minor titles like the IBU, the WBU, the WBE etc. etc. the IBO claims their ratings are fair and unbiased. Until they publish the formula their ratings have as much credibility as the others.

Lennox, I realise that in your opinion the IBO ratings are better but the facts show that BoxRec's are more accurate. You are putting forward an opinion where we are putting forward evidence. As we say in Yorkshire 'opinions are like arse holes, everybody has got one'

Lennox wrote: IBO ratings are used by all the boxing promoters and TV companies to make the matches, all the boxing promoters have their ratings, why are they not credible and boxec are?

This just isn't true, I get emails from TV companies and managers every day asking about the boxrec ratings and who they have to beat to gain a certain ranking. I know of three sanctioning bodies that use the ratings as a starting point for theirs. Whilst the IBO ratings were very popular 5 years ago (I used them myself) they are used less and less now.

Lennox · Post by **Lennox** » 26 Apr 2006, 12:05

JohnShep wrote:Just like all the other minor titles like the IBU, the WBU, the WBE etc. etc. the IBO claims their ratings are fair and unbiased. Until they publish the formula their ratings have as much credibility as the others.

Lennox, I realise that in your opinion the IBO ratings are better but the facts show that BoxRec's are more accurate. You are putting forward an opinion where we are putting forward evidence. As we say in Yorkshire 'opinions are like arse holes, everybody has got one'

What facts?

Post by **John** » 26 Apr 2006, 12:11

Martin did an evaluation or the ratings :-

BoxRec 305-27 8.9% upsets
IBO 251-37 14.7% upsets

Lennox · Post by **Lennox** » 26 Apr 2006, 12:21

JohnShep. Thats something Computer rank has decided is the way to construct a set of rankings.
Forget your theories and answer the question like a gentleman. Do you think Boxrec cruiserweight rankings are better than IBO?

JCS · Post by **JCS** » 26 Apr 2006, 12:22

JohnShep wrote:Martin did an evaluation or the ratings :-

BoxRec 305-27 8.9% upsets
IBO 251-37 14.7% upsets

Trust me, a great deal of those are like #15 vs #500.

In the bouts that comprise of the fighters in the Top 25% by rating points , we are at 33.6% upsets? In my best system I am at 33.1% upsets. Definitely harder to call where it counts, thats for sure.

Maybe we simply can't get much better than that.. Lord knows I have tried.

Post by **John** » 26 Apr 2006, 12:29

Lennox wrote:JohnShep. Thats something Computer rank has decided is the way to construct a set of rankings.
Forget your theories and answer the question like a gentleman. Do you think Boxrec cruiserweight rankings are better than IBO?

Yes absolutely, because they are as accurate and objective as possible and those are my only criteria. If you want my opinion then I would have to say it depends on what day it is and which way the wind is blowing.

jujigatame · Post by **jujigatame** » 26 Apr 2006, 13:26

The BoxRec rankings are pretty good in my book, but I'm not convinced that predictive rate should be the sole accepted criteria for evaluating a computerized ranking system. A system with a 100% predictive rate would have had Wlad Klitchko above Chris Byrd before their rematch. Sometimes the higher ranked man is the underdog, and I don't think a rankings system should aspire to eliminate these occurences.

JCS · Post by **JCS** » 26 Apr 2006, 13:30

jujigatame wrote:The BoxRec rankings are pretty good in my book, but I'm not convinced that predictive rate should be the sole accepted criteria for evaluating a computerized ranking system. A system with a 100% predictive rate would have had Wlad Klitchko above Chris Byrd before their rematch. Sometimes the higher ranked man is the underdog, and I don't think a rankings system should aspire to eliminate these occurences.

Technicall you're right.. but I can blame that one on the hometown factor. Usually, Klitschko fights in the USA - NOT on his home turf.. while Byrd has always fought in his home country. So, Wlad's rating has been formed away from his home (far away) and Byrd's was formed in his homeland.

So, in the Byrd vs. Klitschko fight.. obviously Byrd fighting away is going to hurt his rating.. and Wlad back in the uber-bias Germany will escalate his rating drastically. Now, you and I both know that Wlad could've beat Chris Byrd in his own home with Duane Ford and Glen Hamada running the show.. but if you want to get technical.. such a small difference in rating before the bout suggests its nearly even :).

But.. anyway.. What other statistically proven criteria is there? There has to be some way to test new versions of the system. Without that.. it is simply in the eye of the beholder.

jujigatame · Post by **jujigatame** » 26 Apr 2006, 13:51

But.. anyway.. What other statistically proven criteria is there? There has to be some way to test new versions of the system. Without that.. it is simply in the eye of the beholder.

That's my point. I don't think there is one.

Lennox · Post by **Lennox** » 26 Apr 2006, 14:08

In my opinion a rating should be based on what has happened not what is or might/ going too. So wins get points basically. A win over a good opponent gets x points and if its not such a good opponent lesser, perhaps even no points if the gulf is say 100 differences in the ratings. If a lower ranked fighter beats a higher one so be it. If the number 40 beats the number 20, then I suppose the formula should produce a result that would see the winner at maybe 25 and the loser at 26. i've been studying boxrec & ibo, in boxrec many fighters are in the wrong division to ibo or vice versa, I have been studying Cruiserweight. Rudy Markussen is a boxrec cruiser but IBO 168, the problem is he fought his last fight at 178. I can see how boxrec have him cruiser, but he should probably be at 175. This is the same with Jose Herrera, who beat Jorge Castro. I dont think either were cruiserweights as such.
I can only say what I see and despite the answers from the site owners or rating developers that the boxrec ratings are better because they predict fights to have less upsets, they still dont 'look right'. I dont really know how Computerrank or johnShep or GermanBee can look at Virgil Hill at 31 or Johnny Nelson at 16 and think they have got it right. Virgil Hill only needs to figt someone in the top 30 and win and it will be an upset, which is what they are trying to avoid.
I am going to look at the fights over the next month or so and look at the accuracy of fights and see if Boxrec or IBO produces better prediction. Glancing back over the last few fights IBO had Guinn to beat Audley Harrison Boxrec said Audley would win.

Lennox · Post by **Lennox** » 26 Apr 2006, 14:13

jujigatame wrote:
But.. anyway.. What other statistically proven criteria is there? There has to be some way to test new versions of the system. Without that.. it is simply in the eye of the beholder.
That's my point. I don't think there is one.

Yes I agree.
A good rating or bad rating is an opinion (subjective).
Thats not to say you cant have an objective way of making that rating and clearly Boxrec & IBO have diff ways.
You cant measure a rating and call it objective by the very method of how you developed yours and call the a fact.

Lennox · Post by **Lennox** » 26 Apr 2006, 14:51

Supermiddle 168.

Mger Mkrtchian number 4 and kessler number 7. how can this be?

Kessler has fought approx 6 good foes
Mkrtchian has fought 2 good foes (lost 1)

I am astonished that people have not noticed how bad these ratings are.

JCS · Post by **JCS** » 26 Apr 2006, 14:57

Lennox wrote:Supermiddle 168.

Mger Mkrtchian number 4 and kessler number 7. how can this be?

Kessler has fought approx 6 good foes
Mkrtchian has fought 2 good foes (lost 1)

I am astonished that people have not noticed how bad these ratings are.

We have noticed the problems, we just don't complain as publicly and as often. I have taken the matter up privately w/ Martin, although he claims a lack of time has brought updates to a standstill. Perhaps someone else should assist with the live system?

The problem with Mkrtchian is the fact that Chanet was too high when he was beaten.