Ratings - please read before commenting - Archived

[mattyp151]

Sorry JCS, he was there (or at least top 15). And it was in the last year. There were several versions of the algorithm that put too much weight on retirees coming back.

And that is my point. Martin will publish anything at all, even Bowe in the top 10. There is no plausability check.

What about you? You still haven't said where you would draw the line.


conan

I am not sure what was going on at that time.. I wasn't really active in the ratings discussions/helping out back then. I started contributing about 8 months ago.
That was probably back when the "Traditional ranking" system had major flaws with decay rates.

If Bowe made it into the Top 10 or 15, then I assure you there is a major problem with some programming that affects prediction counts.


BTW, we have a revision in the queue waiting to be posted by the site mod. Not sure when that will make it on live though.

Wrong. There was no programming mistake. There was no "traditional algorithm". There was just the same old algorithm as ever, just with the solution set for multivariable equation back then, it produced the very charming effect of Bowe in the top 10 (or 15). However, as always, because it produced the best "predictability", the farcical ratings were published.

And you still haven't answered the question. Assuming no programming mistakes, would you publish anything at all? Martin would/has.

Besides, what do you mean "major flaws". The last time I was in this debate, Martin was clamiing 80% predictability. Now it's only 75% or so. That would mean this is more flawed than before. And I beg you, what could be more flawed than Chad Dawson #1 and Stipe Drews #2. Maybe 1 in a 1000 boxing fans would choose that. Maybe 800 or so would have one of Adamek Tarver, Tarver Adamek, Johnson Adamek, Adamek Johson, Johnson Tarver, TArver Johson. But none would have Dawson/Drews. That is so flawed.

conan

conan

Can I just say you're complaining an awful lot and offering no help or to help them figure anything out at all?

Seriously, if you have a beef, sit down and discuss better ways instead of spouting off examples that anyone could do.

PS: About Drews, that hometown factor is huge considering he never fights outside and he is in the robbery capital of the boxing world.

And anyone chosing Tarver as the best LHW has no right voting in that poll. I don't care the competition, a guy going 2-2 and one of the wins being against a scared man and losing to a 41 year old jumping up 2 classes has no right being a top fighter anywhere.

[conan_the_cribber]

Can I just say you're complaining an awful lot and offering no help or to help them figure anything out at all?

Seriously, if you have a beef, sit down and discuss better ways instead of spouting off examples that anyone could do.

PS: About Drews, that hometown factor is huge considering he never fights outside and he is in the robbery capital of the boxing world.

And anyone chosing Tarver as the best LHW has no right voting in that poll. I don't care the competition, a guy going 2-2 and one of the wins being against a scared man and losing to a 41 year old jumping up 2 classes has no right being a top fighter anywhere.

So what exactly do you think I'm doing now Matty? All my points are illustrated with examples and theory. If you miss all that because I put a rude word in there or two, then what can I do about your sensibilities?

In the first post I already said to you how you could make it better. By a) using reasonable plausability checks instead of just an autistic statistic and b) by introducing domain relevant aspects into the rankings.

Even the tone of my argument is critical to the debate. The rankings stink so gargantually at the moment, they are an embarrassment to the site and that is something that should not be put in any way more politely than I just did. You can't ignore the elephant in the lounge room and the elephant is over-reliance on the predictability factor, and that is just due to being dogmatic or foolish.

Your point about Drews only illustrates my points a) the ratings stink b) Martin will publish anything, because he is so dogmatic.

And Tarver deserves to be in the pole. He was a deserved linear champ until he got beat by a future Hall of Famer in Hopkins. Hopkins may have equally outclassed all of the candidates in the list, we will never know that. (Personally I would've given Adamek the best chance of a victory).

If you factor in b) Matty, then you will get traditional rankings, that do not suck as now. It's as simple as that. But that will never happen, unless someone tells the emporer that he has no clothes.

Or are you another supporter of Bowe in the top ten? You can complain about me compaining, but apart from misinterpreting my Coke and Pepsi example, I dont see you trying to refute any of my points.

conan

[Blue]

U guys are bucking tradition which has always used ratings to determine objectively who deserves through his accomplishments, a shot at the title.

Exactly what purpose BoxRec’s prediction based system serve other than giving publicist a reference for their bogus propaganda?

Shannon Briggs publicist can now point to Boxrec and claim his fighter is 6th rated even though he hasn’t beat a rated boxer in 9 years.
Except for a very few, most of BoxRec’s members and editors are against it.

U guys need to put on your BoxRec t-shirts and go to shows & see what king of feedback U get. :(
I have stopped wearing mine.

[JCS]

Can I just say you're complaining an awful lot and offering no help or to help them figure anything out at all?

Seriously, if you have a beef, sit down and discuss better ways instead of spouting off examples that anyone could do.

PS: About Drews, that hometown factor is huge considering he never fights outside and he is in the robbery capital of the boxing world.

And anyone chosing Tarver as the best LHW has no right voting in that poll. I don't care the competition, a guy going 2-2 and one of the wins being against a scared man and losing to a 41 year old jumping up 2 classes has no right being a top fighter anywhere.

So what exactly do you think I'm doing now Matty? All my points are illustrated with examples and theory. If you miss all that because I put a rude word in there or two, then what can I do about your sensibilities?

In the first post I already said to you how you could make it better. By a) using reasonable plausability checks instead of just an autistic statistic and b) by introducing domain relevant aspects into the rankings.

Even the tone of my argument is critical to the debate. The rankings stink so gargantually at the moment, they are an embarrassment to the site and that is something that should not be put in any way more politely than I just did. You can't ignore the elephant in the lounge room and the elephant is over-reliance on the predictability factor, and that is just due to being dogmatic or foolish.

Your point about Drews only illustrates my points a) the ratings stink b) Martin will publish anything, because he is so dogmatic.

And Tarver deserves to be in the pole. He was a deserved linear champ until he got beat by a future Hall of Famer in Hopkins. Hopkins may have equally outclassed all of the candidates in the list, we will never know that. (Personally I would've given Adamek the best chance of a victory).

If you factor in b) Matty, then you will get traditional rankings, that do not suck as now. It's as simple as that. But that will never happen, unless someone tells the emporer that he has no clothes.

Or are you another supporter of Bowe in the top ten? You can complain about me compaining, but apart from misinterpreting my Coke and Pepsi example, I dont see you trying to refute any of my points.

conan

Bowe is not in the top 10. You speak of reasonable plausbility... what exactly do you suggest we use? What good is a system that simply mocks others? What other type of statistic would you rather us use, because I doubt any of the powers that be will want to judge their system based on the "quality" of others.

This system is still not a mature one yet and everyone is always wanting to pull the plug simply because it differs from the norm.

BTW, Bowe is not in the top 10 right now.. dunno if you checked. Actually he's inactive, but if he was active, I guarantee you he would not be in the top 10

[conan_the_cribber]

Bowe is not in the top 10. You speak of reasonable plausbility... what exactly do you suggest we use? What good is a system that simply mocks others? What other type of statistic would you rather us use, because I doubt any of the powers that be will want to judge their system based on the "quality" of others.

This system is still not a mature one yet and everyone is always wanting to pull the plug simply because it differs from the norm.

BTW, Bowe is not in the top 10 right now.. dunno if you checked. Actually he's inactive, but if he was active, I guarantee you he would not be in the top 10

Hi JCS,

here is a very polite post regarding each sentence above. I hope it helps.

"Bowe is not in the top 10". No, thank god he is not. I use this as an example for one simple reason. And that reason is to illustrate that Martin (computerrank) is prepared to publish anything. Bowe was in in the top 10, or at least in the top 15 for a while. Whatever it was that caused it, it has been fixed. I hope it was fixed and that the predictability went down, because that would be a precedent that is so desperately needed now.

"You speak of reasonable plausbility... what exactly do you suggest we use?". The plausability I speak of, is what I have mentioned several times, the feedback of your peers. The message is overwhelmingly negative at the moment. The alarm bells in your head should be going off, just as they went off in mine, when I read 98% of people prefer Coke over Pepsi in Wyoming.

"What good is a system that simply mocks others? ". Personally I see a lot of good in producing a set of rankings that a site can be proud of. We have a statement from Editor Blue up above, someone that publicly represents boxrec at boxing shows, that he is ashamed of the rankings. I have been asking the opposing question though. What is the point of producing a set of ratings that being actively laughed at by the boxing community?

"What other type of statistic would you rather us use, because I doubt any of the powers that be will want to judge their system based on the "quality" of others.". Again you are falling into the trap of using statistics to justify statistics. That is exactly my main point, the over-reliance on the predictability statistic is ruining the ratings, because it a) falls the "human" test, and b) it does not respect the principles of the boxing domain space e.g. untested prospects should never be #1. If you absolutely wanted to use a statistic to do a plausability check on your results, then I would suggest correlating your results against the acknowledged best ratings on the net. My personal picks are Scott Schaeffers excellent list on boxingtalk and the editorial teams picks on http://www.boxingpress.de. I consider both lists to be done with the utmost ethical quality. Regarding qualityand powers that be. The first iterations done by John Shep himself were done until in his own words "They looked right". The comparison that he had, were the other hand made lists. I think you have to get off your high horse and think that the other lists have no quality. Scotts list is superb and even complies to the reporting requirements as defined by Muhammed Ali Act and codified by the Association of Boxing Comissions.

"This system is still not a mature one yet and everyone is always wanting to pull the plug simply because it differs from the norm.". Two points. This system has been ranking people for at least two years now. That's a lot of time to get it right. People do not want to pull the plug, people want accurate boxing ratings, based on expectations of the boxing domain space. What they are getting is something so left field, that people are getting upset. People see no value in the predicability percentage, especially me, as I have illustrated that is statistically and mathematically unsound.

"BTW, Bowe is not in the top 10 right now.. dunno if you checked. Actually he's inactive, but if he was active, I guarantee you he would not be in the top 10". As above. I used Bowe in the top 10 as an example of what Martin will publish. I am aware that he is currently not in the top 10, but he was shortly after his comeback. If you garantee that he will nnot be in the top 10, then that only means common sense prevails. All I'm asking from you and Martin is that this common sense is expanded to cover such grave ranking mishaps as Dawson #1 and Drews #2.

I hope that clears up a lot for you.

conan

[conan_the_cribber]

Hi CopWebCat ,

Thanks for your points, I hope I can answer them below.

Conan,
How many ranking systems do you know where even one person is happy with all the placings? Rankings is an imperfect science by its very nature. Whatever "checks" you carry EVERYONE will still be unhappy about some aspect of the system.....that's EVERYONE, not a majority.

That is too general Mr Cat. of course everyone will disagree about the finer points of a rating. If I rank the top 4 at Light Heavy as RJJ, Michaelzwevski, Nunn and Hill, based on an algorithm that over emphasises life time achievements, then of course people will complain. However, if I rank it like Adamak, Erdei, Tarver, Johnson, Dawson, Woods, then people might want to swap a fighter or two, but they will walk away saying that is a defensible list. At the moment, the boxrec ratings are indefensible. Just as my 'bad' algorithm weighted too much for life time achievements, the current one way over emphasises protected fighters.

What about other weight divisions? This system has Mayweather number one at Welterweight, is that right or wrong? In other weight divisions Manny Pacquiao and Calzaghe are first...is that right or wrong.

You can make an argument for Mayweather #1 at welter and a pretty strong argument for both Calzaghe #1 at Super Middlea and Pacquaio at super feather. If you came up with any of these fighters at less than top 3, then that would be a "plausability check" that something is not right. However, it is also a plausability check that Chad Dawson is #1 and light heavy and Stipe friggen Drews is #2. Just because something is about right, doesn't mean that the whole thing is worth keeping.

I think many people would say that that's correct so there's a check for you.

Again you are being too simplistic. I can look at a car and say that the headlights look great, but if the brakes dont work and the engine is leaking oil and the doors dont shut, then it's fair to say that the car sucks. You are using some of the few things that are right, to justify the whole. The whole is seriously wrong.

In the All-time ratings Ali is first wright or wrong?

I haven't looked at the all time rankings and I have spent no time theorizing about it. Ali is top 10 all time, so it's not so bad to have him at #1.

You are picking out one or two examples you are not happy with and ignoring the ones that "look correct"

Rubbish. For the sake of time I picked out the most prominent ones that are bad. I also restricted it to one weight class, so that it's easier to debate. Other people have cited other notable examples including Shannon the Cannon Briggs.

For the record, Briggs, Thompson and Brock are all wrong at heavy.Jean Marc Mormeck at 12 at cruiser is brain dead. Goddamm he was the unified champ one fight ago!!!!! Stipe Drews at 2 is also a crime. Briggs is at 13 and he smashed the shit out of Drews. Adamek beat Briggs and both are lower than Drews??? WBC supermiddle Champ Markus Beyer is 12? The middles are OK. Sechell Powell is 2 at light middle 5 places ahead of DLH who's coming off a great win over Mayorga? Not to mention being ahead of Mosely, Kamarzin and Vargas. No point going on, there are just so many mistakes it aint funny.

If you really are a statitician you must realise there is no equation out there which will produce ratings which even one person is completely happy with so unless you have such a system I think its time for you to stop moaning!

I really am a statistician, which is why I argue with theoretical knowledge as well as my own knowledge as a boxing fan and ex-journalist. I will stop moaning when people like you stop pretending that the current ratings are OK. They are a flat out joke. It is ironic that you acuse me of moaning. I have provided a logical argument for my position. I am the one working and thinking here and it is YOU, with your simple little argument that "nothing is perfect" is the one that is moaning.

My point is, and this is shared by many others, is that the ratings are so imperfect, that it is time to do something about it. My suggestions are to stop relying solely on the predictability statistic as a plausability check and incorporate the boxing domain space in the algoirthm (A beats B A is ahead of B, got to beat people at the top level to be top).

conan

edit: actually I know a whole bunch of complicated ranking systems that everybody is happy with. FIDE Chess rankings. ICC Cricket Team rankings and the ICC Cricket Players rankings. The ATPtour 52 week rankings are good. All of these fine ranking systems are undisputed and guess what, completely automated i.e. based purely on the results. It is possible, boxing is a tick harder, beause of the lack of direct comparisons.

[JCS]

Bowe is not in the top 10. You speak of reasonable plausbility... what exactly do you suggest we use? What good is a system that simply mocks others? What other type of statistic would you rather us use, because I doubt any of the powers that be will want to judge their system based on the "quality" of others.

This system is still not a mature one yet and everyone is always wanting to pull the plug simply because it differs from the norm.

BTW, Bowe is not in the top 10 right now.. dunno if you checked. Actually he's inactive, but if he was active, I guarantee you he would not be in the top 10

Hi JCS,

here is a very polite post regarding each sentence above. I hope it helps.

"Bowe is not in the top 10". No, thank god he is not. I use this as an example for one simple reason. And that reason is to illustrate that Martin (computerrank) is prepared to publish anything. Bowe was in in the top 10, or at least in the top 15 for a while. Whatever it was that caused it, it has been fixed. I hope it was fixed and that the predictability went down, because that would be a precedent that is so desperately needed now.

"You speak of reasonable plausbility... what exactly do you suggest we use?". The plausability I speak of, is what I have mentioned several times, the feedback of your peers. The message is overwhelmingly negative at the moment. The alarm bells in your head should be going off, just as they went off in mine, when I read 98% of people prefer Coke over Pepsi in Wyoming.

"What good is a system that simply mocks others? ". Personally I see a lot of good in producing a set of rankings that a site can be proud of. We have a statement from Editor Blue up above, someone that publicly represents boxrec at boxing shows, that he is ashamed of the rankings. I have been asking the opposing question though. What is the point of producing a set of ratings that being actively laughed at by the boxing community?

"What other type of statistic would you rather us use, because I doubt any of the powers that be will want to judge their system based on the "quality" of others.". Again you are falling into the trap of using statistics to justify statistics. That is exactly my main point, the over-reliance on the predictability statistic is ruining the ratings, because it a) falls the "human" test, and b) it does not respect the principles of the boxing domain space e.g. untested prospects should never be #1. If you absolutely wanted to use a statistic to do a plausability check on your results, then I would suggest correlating your results against the acknowledged best ratings on the net. My personal picks are Scott Schaeffers excellent list on boxingtalk and the editorial teams picks on http://www.boxingpress.de. I consider both lists to be done with the utmost ethical quality. Regarding qualityand powers that be. The first iterations done by John Shep himself were done until in his own words "They looked right". The comparison that he had, were the other hand made lists. I think you have to get off your high horse and think that the other lists have no quality. Scotts list is superb and even complies to the reporting requirements as defined by Muhammed Ali Act and codified by the Association of Boxing Comissions.

"This system is still not a mature one yet and everyone is always wanting to pull the plug simply because it differs from the norm.". Two points. This system has been ranking people for at least two years now. That's a lot of time to get it right. People do not want to pull the plug, people want accurate boxing ratings, based on expectations of the boxing domain space. What they are getting is something so left field, that people are getting upset. People see no value in the predicability percentage, especially me, as I have illustrated that is statistically and mathematically unsound.

"BTW, Bowe is not in the top 10 right now.. dunno if you checked. Actually he's inactive, but if he was active, I guarantee you he would not be in the top 10". As above. I used Bowe in the top 10 as an example of what Martin will publish. I am aware that he is currently not in the top 10, but he was shortly after his comeback. If you garantee that he will nnot be in the top 10, then that only means common sense prevails. All I'm asking from you and Martin is that this common sense is expanded to cover such grave ranking mishaps as Dawson #1 and Drews #2.

I hope that clears up a lot for you.

conan

THIS particular system which focuses entirely on prediction rate is not that old.. Only since early '06 or so.

Sure its a system to be proud of, no other site can boast of a rating system with such a high level of predictability.

[conan_the_cribber]

THIS particular system which focuses entirely on prediction rate is not that old.. Only since early '06 or so.

Sure its a system to be proud of, no other site can boast of a rating system with such a high level of predictability.

Well obviously being polite and going to a lot of trouble to explain things to you didn't help. What a blinkered answer this is. I had exactly the same debate over a year ago with Martin and Monty and a few others. The prediction rate is Martin's baby. It was excuse a year ago and it's excuse now. Except now the results are so appalling that you can't hide behind it any more.

At what the hell is there to be proud of? You've got Mormeck at 12. One loss ago he was the unified freaking champ. He lost to the new champ, an undefeated fighter with an excellent KO record. And Mormeck gets smoked 11 places in the ratings for that. That is garbage.

And your claim that "no other site can boast of a rating system with such a high level of predictability" is a COMPLETELY UNFOUNDED LIE. You have not tested that. You have no proof whatsoever. All you can say is that no other site embarrasses themselves with such a myopical view of the world that produces such crap ratings.

conan

[JCS]

THIS particular system which focuses entirely on prediction rate is not that old.. Only since early '06 or so.

Sure its a system to be proud of, no other site can boast of a rating system with such a high level of predictability.

Well obviously being polite and going to a lot of trouble to explain things to you didn't help. What a blinkered answer this is. I had exactly the same debate over a year ago with Martin and Monty and a few others. The prediction rate is Martin's baby. It was excuse a year ago and it's excuse now. Except now the results are so appalling that you can't hide behind it any more.

At what the hell is there to be proud of? You've got Mormeck at 12. One loss ago he was the unified freaking champ. He lost to the new champ, an undefeated fighter with an excellent KO record. And Mormeck gets smoked 11 places in the ratings for that. That is garbage.

And your claim that "no other site can boast of a rating system with such a high level of predictability" is a COMPLETELY UNFOUNDED LIE. You have not tested that. You have no proof whatsoever. All you can say is that no other site embarrasses themselves with such a myopical view of the world that produces such crap ratings.

conan

Ask Martin about the comparisons to other systems. I believe he has tested it.

Once we get the latest revision online, I will spearhead some comparisons to other rankings though. I would not mind that at all.

[conan_the_cribber]

Well seeming the predictability is sooooooooo good, let's have a little look at what's happening next. Remember 75% of this is likely to happen according to the best rankings on the web.

1) Coming off a career best win against Eric "not in the top 25" Harding, Chad Dawson is going to smoke any of Adamek, Erdei, Johnson, Tarver or Woods.

2) Stipe Drews should enjoy a rematch against Paul Briggs. Despite hardly winning a round in the last fight, the boxrec predictabiltiy rankings imply that Drews #2, should deal with Briggs #13 easily.

3) Zsolt Erdei should be aiming for unification fights. Despite being thought of as by most boxing pundits as the weakest of the four champs, according to the super-duper boxrec predictablitly ratings he should start a slight favourite in each bout as he is ranked higher than all the other champs.

4) Clinton Woods should be looking forward to his rematch with Johnson. Despite going for a draw and a loss previously, boxrec suggests that this time its the right time. I guess Johsons subsequent wins over former p4p champ Roy Jones and unified champ Tarver don't count for shit in this algorithm. Clinton Woods win against the erratic non top 10 fighter Gonzales is really where you soak up the points.

5) Tarver. No comment

6) Adamek. No comment basically. Looks like his wins against Briggs and Ullrich dont match Woods wins over jason de Lisle.

7) Adrian Diaconu. Wow, go Canada. Conquerer of 68th ranked Thysse (who moved up in weight) and already a neck and neck fight with Adamek and a firm favourite over Johnson, Brigggs, Roy Jones etc.

8) Prince Badi Ajamu will be licking his chops against faded Roy Jones Jnr. Despite two losses to two fighters currently ranked lower than 20, it's a case of third time lucky for him as he battles #19 Roy Jones Jnr.

9) Garay. Despite the fact that he IS actually a decent fighter, boxrec shows the real way to get ahead in the rankings. First of all you lose twice to a weakish champ in Zoltan Erdei. Then you fight three stiffs outside of the top 100 and hey presto you're back in the top 10.

10) Glen Johnson. Here boxrec shows you how not to get ahead in the ratings. Despite being 6-1-2 against mostly world class opposition in his last 9 outings, Johnson rots away at 10 in the rankings. Why? Because he chose the hard road and not fight a bunch of stiffs to up his rankings. That has to get punished here. According to boxrec, he should start as an outsider against Drews and Diaconu, who haven't defeated shit yet.

Anybody else get that inkling feeling, that the predictability of the boxrec rankings aren't all that they're hyped up to be? All the predictions here are the predictions given by the current algorithm. And that has an oh so high predictability OR DOES IT ?????

conan

[JCS]

Again Conan.. Wait for matches to actually happen, then analyze the results. Also remember home country advantage of 125 points.

Home country advantage is negated however if the other fighter has fight experience in that country.... I believe it is 2 fights or more then 2 fights.. have to check on that.

Briggs would be favored in Australia vs. Drews at this moment.

Also, who's to say a fighter can't improve from their previous fight while the opponent declines?

BTW - Fights in that range of fighters are 77-78% predict rate. Overall established fighter is about 84%

Once you consider major upsets, screwed decisions, etc. You will realize you can't get much higher than that.

[Lennox]

Conan,

How many ranking systems do you know where even one person is happy with all the placings? Rankings is an imperfect science by its very nature. Whatever "checks" you carry EVERYONE will still be unhappy about some aspect of the system.....that's EVERYONE, not a majority.

What about other weight divisions? This system has Mayweather number one at Welterweight, is that right or wrong? In other weight divisions Manny Pacquiao and Calzaghe are first...is that right or wrong.

I think many people would say that that's correct so there's a check for you. In the All-time ratings Ali is first wright or wrong?

You are picking out one or two examples you are not happy with and ignoring the ones that "look correct"

If you really are a statitician you must realise there is no equation out there which will produce ratings which even one person is completely happy with so unless you have such a system I think its time for you to stop moaning!

I think the IBO is as near as you could get to perfect ratings. I wish Boxrec would use them, I like to use the IBO number and then check the fighter using Boxrec. I have accepted Boxrec's new ratings/rankings'listings are not 'what I want' and I have no interest in knowing that the computer says 77% of the time chad Dawson will defeat Erdei or Adamak, Tarver whoever. I'd say at this stage 77% he would lose. For some reason a minority beleive the best ratings are the one who more accurately pick the winner, Id say 95% of people would say the best ratings are the ones that show the order by merit, achievement, facts, that to most will be IBO ratings. Prospects will look to face better names as their career's progress, in most cases they will beat the fading fighter. I still think JohnShep,CompRank JSC are baking the wrong cake.

[conan_the_cribber]

Hey Webcat.

Thanks for your response. Here are my answers to your points.

Really Conan?
I cant make you out. You have some interesting points and then come out with some awful ones.

Well no-one is perfect. 8)

If I have read you correctly you think if A beats B then B should always be ranked higher than A. Are you serious?

Yes/no/maybe. You have read me correctly, but you only have half the story. The principal points that I want to make are that a) there is an over reliance on the predictability statistic and b) the percularities of the boxing domain must be appreciated. If you persist in ignoring a) and b) then you will get ratings that are astounding bad, as they are now.

As part of main point b), I very briefly described the expectation that if A beats B, then the expectation is that, that until A or B have done something significant in their careers, that A should be ahead of B. There is a lot more theory to this. My opinion is definitely that until A or B fight again, then A must be ahead of B. My opinion is that subsequent fights and time decay mean that at some point in the future that B could be ahead of A. The trick is to define the affect of subsequent fights and to define the decay.

However, and I stress this, the main point is that you lot have to recognise that there's an elephant in the lounge room. You have to stop having a nice cup of tea and and pretend there's nothing unusual going on. The fine print regarding A beats B is not part of my initial agenda.

I run a football rankings site "MANCRANK" are you suggesting if Espanyol beats Barcelona then they are the better team? What utter crap!

Football rankings are football rankings and boxing rankings are boxing rankings. You cannot compare them, nor did I make the statement regarding Barcelona.

Fide uses the ELO system which is very prediction based and does not promote one player over another on the basis of one result!

So? Just like you misintrepretted my use of the Coke vs Pepsi example, here you have misintrepretted my FIDE comments. I brought this in as an example to show that an automated ranking programm can find general (not universal) acceptance amongst a particular community. I did not bring it in to say it should be used for boxing. Nor did I say that it promotes one player over another on the basis of one result.

And to be formal, the ELO is not prediction based. If you have two ELO rankings, then you can theoretically make a prediction about what are the percentage chances of each player e.g. how many games will the weaker player win out of 10 games. That is however, just an interpretation of the ratings.

Incidentally there is a better system that FIDE could use and so I for one am NOT happy with their ratings so when you make sweeping statements like everyone is happy with them or they are generally well regarded, just think that might not be true. Remember? Think...think some more...and again... got it?!

Again I say that the acceptance is general and not universal. I have no idea why you do not like the FIDE rankings. They have been used for 25 years on a world and country stage and produce a set of rankings that are completely acceptatble.

If you want to remand somebody for thinking then I suggest you
a) read my posts carefully, so you don't misquote my examples in a different context and
b) don't post comments or ideas in my name that I didn't say, like you did five times here.

Where did I read that before......?! 8)

I dont know, somebody who thinks a lot probably wrote it. Seriously, at least you're trying to have a debate about this, which I appreciate it. Martin is busy or hiding. JCS hasn't contributed much yet. But you can't ignore the elephant much longer. Have a look at my predictions post. God that sounds rocky to me. I seriously doubt the predicativaty being quoted.

conan

[conan_the_cribber]

Again Conan.. Wait for matches to actually happen, then analyze the results. Also remember home country advantage of 125 points.

Home country advantage is negated however if the other fighter has fight experience in that country.... I believe it is 2 fights or more then 2 fights.. have to check on that.

Briggs would be favored in Australia vs. Drews at this moment.

Also, who's to say a fighter can't improve from their previous fight while the opponent declines?

BTW - Fights in that range of fighters are 77-78% predict rate. Overall established fighter is about 84%

Once you consider major upsets, screwed decisions, etc. You will realize you can't get much higher than that.

Want to put your money where your mouth is?

Established fighter level is 84%. Tell you what, let's keep it at 75/25, that's a 9% advantage to you already. Woods vs Johnson is up next. You give me 75 bucks if Johnson wins and I'll give you 25 if Woods wins. if Briggs fights Drews in the next year, we can do that for 750 to 250. Hell, I'll even take 700 to 300, you get a bonus 5%.

Are you so confident in your predictability that you want to do that?

Are you so proud that you can admit any misgivings that my list indicates? No misgivings whatsoever, everything is rosy? Honestly, none whatsoever? If so, then you should jump all over my Johson/Woods offer, you got nothing to lose.

conan

[JCS]

Again Conan.. Wait for matches to actually happen, then analyze the results. Also remember home country advantage of 125 points.

Home country advantage is negated however if the other fighter has fight experience in that country.... I believe it is 2 fights or more then 2 fights.. have to check on that.

Briggs would be favored in Australia vs. Drews at this moment.

Also, who's to say a fighter can't improve from their previous fight while the opponent declines?

BTW - Fights in that range of fighters are 77-78% predict rate. Overall established fighter is about 84%

Once you consider major upsets, screwed decisions, etc. You will realize you can't get much higher than that.

Want to put your money where your mouth is?

Established fighter level is 84%. Tell you what, let's keep it at 75/25, that's a 9% advantage to you already. Woods vs Johnson is up next. You give me 75 bucks if Johnson wins and I'll give you 25 if Woods wins. if Briggs fights Drews in the next year, we can do that for 750 to 250. Hell, I'll even take 700 to 300, you get a bonus 5%.

Are you so confident in your predictability that you want to do that?

Are you so proud that you can admit any misgivings that my list indicates? No misgivings whatsoever, everything is rosy? Honestly, none whatsoever? If so, then you should jump all over my Johson/Woods offer, you got nothing to lose.

conan

You are once again ignoring the home country factor.

Also, the 77-78% is the average for fights with fighters who are established, have 8 or more wins and a win % of 75% or greater. These fighters are most likely closer in rating than are the average fights between these upper echlon of established fighters. The odds would probably be more like 65/35 at that close of a range.

If Johnson/Woods took place in the states now.. Johnson would get a +125

Johnson = 1493 + 125 = 1618
Woods = 1542 + 0 = 1542


Woods would be ahead if the fight were in the UK


Once the latest revision is posted, I'll start tracking results.

[conan_the_cribber]

Conan: I never mentioned Pepsi! Though, for the record, I prefer it to coke. I was a Pepsi Max junkie, until the caffeine played havoc with my stomach lining. But I digress...

Any fighter is capable of a fluke victory over a top ranked fighter. How bad would the rankings look if someone with a record of 1-15-0 were top because of one punch against a tp fighter having an off night? I know this wasn't your main point but it seriously makes me consider your other points. I think Boxrec used to use that system with some outrageous looking ratings as a consequence.

The WBO rankings are interesting and I'd be interested to see what would happen in a direct comparison Vs Boxrec. I cant see anywhere where they say how they are calculated though and that makes me a little suspicious (probably incorrectly)

Hi Mr Cat.

Indeed it was Matty who missed the coke vs pepsi point. I take that back. My apologies.

The lucky punch stuff is actually pretty irrelevant. Firstly no champ fights a 1-15 fighter. Secondly, if an upset does occur, then it usually an indication that further upsets will occur. Lennox got nailed by McCall and then Rahman. Wlad got beat up by a strong chinned Purrity and then Sanders and Brewster. Zab was an accident waiting to happen. He proved his loss to Spinks was no fluke, because he lost to Baldomir. I can't for the life of me think of a recent top 10 boxer who near the peak of his career lost via lucky punch and then went on to never lose again.

Martin actually knows the answer to this question, he mentioned it in his only response to me. It's all about levels i.e. the perceived level of a fighter. Is the figher an elite p4p, a champion, a serious contendor, a fringe contendor, a journeyman or a tomato can. If an elite fighter loses to 1-15 fighter, then he was never an elite fighter. he should be reduced to a champion level ( a poor one at that) or even a serous contendor. Conversely the journeyman should be elevated to the (poor) champion level or serious contendor level. Amongst this level, then A should be ranked ahead of B, at least until further data is available.

If there are several eilte fighters in a division, then then the former elite fighter should land behind them. By definition, anyone who gets KO'd by a bum is not an elite fighter. I did not say or imply, that the lucky puncher should inherit the previous spot occupied by the elite fighter. If I personally came out and KO'd Manny Pacquaio on my pro debut, then there is no way I should be ranked above Barerra, Raheem, Moralles etc.

The WBo rankings suck. What your mean is the IBO rankings. The IBO rankings rock, when they are up to date that is. Their algorithm is not public property, but they have the concept of 'levels' in it. All you have to do is to look at the points for a fighter and you can guage at what level they are at. You look at boxrec's and there is very little difference between 1 and 20.

You can be rest assured it's an algorithm though.

conan

[JCS]

Yeah I meant IBO.

IBO vs Boxrec anyone?

I'm up for that as soon as JohnShep implements the latest revision which should be soon.. Don't want any changeover midway through and its already in the queue.

[mattyp151]

Conan: I never mentioned Pepsi! Though, for the record, I prefer it to coke. I was a Pepsi Max junkie, until the caffeine played havoc with my stomach lining. But I digress...

Any fighter is capable of a fluke victory over a top ranked fighter. How bad would the rankings look if someone with a record of 1-15-0 were top because of one punch against a tp fighter having an off night? I know this wasn't your main point but it seriously makes me consider your other points. I think Boxrec used to use that system with some outrageous looking ratings as a consequence.

The WBO rankings are interesting and I'd be interested to see what would happen in a direct comparison Vs Boxrec. I cant see anywhere where they say how they are calculated though and that makes me a little suspicious (probably incorrectly)

Hi Mr Cat.

Indeed it was Matty who missed the coke vs pepsi point. I take that back. My apologies.

The lucky punch stuff is actually pretty irrelevant. Firstly no champ fights a 1-15 fighter. Secondly, if an upset does occur, then it usually an indication that further upsets will occur. Lennox got nailed by McCall and then Rahman. Wlad got beat up by a strong chinned Purrity and then Sanders and Brewster. Zab was an accident waiting to happen. He proved his loss to Spinks was no fluke, because he lost to Baldomir. I can't for the life of me think of a recent top 10 boxer who near the peak of his career lost via lucky punch and then went on to never lose again.

Martin actually knows the answer to this question, he mentioned it in his only response to me. It's all about levels i.e. the perceived level of a fighter. Is the figher an elite p4p, a champion, a serious contendor, a fringe contendor, a journeyman or a tomato can. If an elite fighter loses to 1-15 fighter, then he was never an elite fighter. he should be reduced to a champion level ( a poor one at that) or even a serous contendor. Conversely the journeyman should be elevated to the (poor) champion level or serious contendor level. Amongst this level, then A should be ranked ahead of B, at least until further data is available.

If there are several eilte fighters in a division, then then the former elite fighter should land behind them. By definition, anyone who gets KO'd by a bum is not an elite fighter. I did not say or imply, that the lucky puncher should inherit the previous spot occupied by the elite fighter. If I personally came out and KO'd Manny Pacquaio on my pro debut, then there is no way I should be ranked above Barerra, Raheem, Moralles etc.

The WBo rankings suck. What your mean is the IBO rankings. The IBO rankings rock, when they are up to date that is. Their algorithm is not public property, but they have the concept of 'levels' in it. All you have to do is to look at the points for a fighter and you can guage at what level they are at. You look at boxrec's and there is very little difference between 1 and 20.

You can be rest assured it's an algorithm though.

conan

I didn't misunderstand your soda model at all...it's a shit example. You compared a poll of two options to a poll of 15,000.

Put it this way, for your plausibility clause or whatever, a p4p list completely derails any theory of plausibility. Most p4p lists have Mayweather #1 or #2. You ever think he would have a chance to beat Nicolay Valuev? No way. So, why isn't Valuev #1? It's not plausible that Mayweather could beat anyone minimum to heavy, so thus, by your reason, the only #1 fighter could be a LHW, CW or HW pretty much.

[conan_the_cribber]

Yeah I meant IBO.

IBO vs Boxrec anyone?

Boxrec
+ open source
- crap ratings

IBO
- proprietry code
+ great ratings.

predictabiltiy, who gives a toss?

conan

[JCS]

Yeah I meant IBO.

IBO vs Boxrec anyone?

Boxrec
+ open source
- crap ratings

IBO
- proprietry code
+ great ratings.

predictabiltiy, who gives a toss?

conan

Obviously we do... so you're missing the whole basis of the BoxRec system.

[conan_the_cribber]

Conan: I never mentioned Pepsi! Though, for the record, I prefer it to coke. I was a Pepsi Max junkie, until the caffeine played havoc with my stomach lining. But I digress...

Any fighter is capable of a fluke victory over a top ranked fighter. How bad would the rankings look if someone with a record of 1-15-0 were top because of one punch against a tp fighter having an off night? I know this wasn't your main point but it seriously makes me consider your other points. I think Boxrec used to use that system with some outrageous looking ratings as a consequence.

The WBO rankings are interesting and I'd be interested to see what would happen in a direct comparison Vs Boxrec. I cant see anywhere where they say how they are calculated though and that makes me a little suspicious (probably incorrectly)

Hi Mr Cat.

Indeed it was Matty who missed the coke vs pepsi point. I take that back. My apologies.

The lucky punch stuff is actually pretty irrelevant. Firstly no champ fights a 1-15 fighter. Secondly, if an upset does occur, then it usually an indication that further upsets will occur. Lennox got nailed by McCall and then Rahman. Wlad got beat up by a strong chinned Purrity and then Sanders and Brewster. Zab was an accident waiting to happen. He proved his loss to Spinks was no fluke, because he lost to Baldomir. I can't for the life of me think of a recent top 10 boxer who near the peak of his career lost via lucky punch and then went on to never lose again.

Martin actually knows the answer to this question, he mentioned it in his only response to me. It's all about levels i.e. the perceived level of a fighter. Is the figher an elite p4p, a champion, a serious contendor, a fringe contendor, a journeyman or a tomato can. If an elite fighter loses to 1-15 fighter, then he was never an elite fighter. he should be reduced to a champion level ( a poor one at that) or even a serous contendor. Conversely the journeyman should be elevated to the (poor) champion level or serious contendor level. Amongst this level, then A should be ranked ahead of B, at least until further data is available.

If there are several eilte fighters in a division, then then the former elite fighter should land behind them. By definition, anyone who gets KO'd by a bum is not an elite fighter. I did not say or imply, that the lucky puncher should inherit the previous spot occupied by the elite fighter. If I personally came out and KO'd Manny Pacquaio on my pro debut, then there is no way I should be ranked above Barerra, Raheem, Moralles etc.

The WBo rankings suck. What your mean is the IBO rankings. The IBO rankings rock, when they are up to date that is. Their algorithm is not public property, but they have the concept of 'levels' in it. All you have to do is to look at the points for a fighter and you can guage at what level they are at. You look at boxrec's and there is very little difference between 1 and 20.

You can be rest assured it's an algorithm though.

conan

I didn't misunderstand your soda model at all...it's a shit example. You compared a poll of two options to a poll of 15,000.

Put it this way, for your plausibility clause or whatever, a p4p list completely derails any theory of plausibility. Most p4p lists have Mayweather #1 or #2. You ever think he would have a chance to beat Nicolay Valuev? No way. So, why isn't Valuev #1? It's not plausible that Mayweather could beat anyone minimum to heavy, so thus, by your reason, the only #1 fighter could be a LHW, CW or HW pretty much.

You got a real short memory Matty, or you didnt read my response. I used Coke vs Pepsi as an example of how it is important to do a plausability check. I did not compare it as a poll between two options and a poll of 15000. Read the first post again and come back.

And who is talking about the p4p list? What has that got to do with the price of fish in China. Nothing. I am criticizing the regular lists, not the p4p list. I dont even care about the p4p list. What do you mean then by "by your reason". I haven't cited anything about the p4p list. Don't let your anger at me get you all confused and quote things that I never said.

conan

[conan_the_cribber]

Yeah I meant IBO.

IBO vs Boxrec anyone?

Boxrec
+ open source
- crap ratings

IBO
- proprietry code
+ great ratings.

predictabiltiy, who gives a toss?

conan

Obviously we do... so you're missing the whole basis of the BoxRec system.

Predicability is important as one of a few key plausability checks of any algorithm. You have missed the point. You are all worshipping predicatbility as the SOLE verification of a ranking system. And that is dogmatic and wrong. I understand teh boxrec system well, hell I have the old source code lying around at home. Unlike you, I understand the theoretical weaknesses of what you are doing.

You still haven't said, how bad the rankings would have to be before you wouldn't publish them.

The fight johson vs Woods is in the UK and that means your +125 is on Woods side. You give me 65/35 and we got a bet.

You haven't indicated whether you have any reservations about the predictions post I made. None whatsoever. It's a damming list. You say "wait and see". I say that's a cop out. Read it and denounce it, if you can. The current rankings list is so counter-intuuitive, but like a flat earther you are unable to even accept an inkling of evidence to the contrary.

If your just going to say "predictability, predicability, predictability" and not discuss any point that I brought up, then butt out of the debate. I certainly have gone to a lot of trouble to dissect your and other's statements line by line. Let Martin get on centre stage. He's the one that has to defend this mess. At the moment the only one worth talking to is Mr Cat.

conan

[conan_the_cribber]

By the way, who's number one at heavy in the new algoirthm? Calvin Brock or Butterbean or someone sensible?

conan

[JCS]

By the way, who's number one at heavy in the new algoirthm? Calvin Brock or Butterbean or someone sensible?

conan

I thought I wasn't worth talking to? I also thought you were only going to stick around for 1 day per year?

[conan_the_cribber]

So many posts, so much to read. Here's a summary of my position. Refute at will.

1) The current rankings are so bad that it is alarming. Even Editors of this site are embarrassed by them.

2) The reason that they are so bad is because a) the only plausability check being used is the predicatbiltiy stat and b) common ranking factors used in the boxing domain are being wilfully ignored.

3) Improvements will be achieved if you build in common ranking factors in use today.

4) The statements about "great predictability" are misfounded. They have no root in mathematical theory e.g. 10000088 might be the optimum for home town advantage.

5) They are not better than any other sites. There has been no comparison, so how do you know.

6) My post regarding the light heavy match ups indicates how common sense has gone out the window. Fighters who got toasted in their only serious fight ending up being number 2.

7) The problems are present at all weight levels. How did Mormeck end up at #12?

8) I do not care about the p4p rankings.

9) The Coke vs Pepsi example is used solely to illustrate that plausability checks are needed, not to suggest that you can compare a Coke vs Pepsi drink test with a boxrec algorithm.

10) A beats B means for now, A must be ahead of B. It does not mean forever and does not apply to other sports.

11) The best improvement to the rankings can be achieved by introducing levels. Here are the six that I prefer, elite p4p, champion, serious contendor, fringe contendor, journey man, tomato can. A fighter should move up and down amongst the levels depending on who he fights. If Shannon Briggs only fights and beats fringe contendors, then at best he should rise to a serious contendor (but a poor one at that).

12) Martin (computerrank) will publish anythign, so long as his precious predicability stat goes up (see point 2). As proof I offer,a) he has already published rankings where Big Daddy Bowe returned to the top 15 and b) rankings where the 1-2 in the light heavy division is Dawson-Drews.

13) I dont have the time to code up a better algorithm myself. That doesn't mean, the existing algorithm is acceptable.

14) There is an elephant in the lounge room. And that elephant is that the ratings suck. There's no point being nice and chatting about it. The ratings are currently so bad, that an open and frank discussion about it is needed.

That's about it, I think.

conan