BoxRec

MontyCircus wrote:

computerrank wrote:But currently I am working on new solution for the current p4p ratings, refering to some suggestions from the forum's side:

- The ratings are used without equalisation.
- For p4p purpose a boxer's current rating is transformed to the lowest divsion, he achieved 500 in at least.

-> So boxers like Mayweather, Hopkins, Pacquio, Haye ... get a higher p4p rating. They will no longer lose points for moving up the divisions.
-> The ratings will no longer change daily caused by the division cluster equalisation
-> the p4p rating will no longer change by simply changing the division

Looks like this will bring Pacquiao to #1

Hey computerrank,

Whatever happened to this system? It's necessary in my opinion. And also for the all-time P4P list as well.

It makes sense for points to drop as a fighter moves up, but not on P4P lists.

Cutting those point cuts out of the division jumps would definitely be an improvement I think.

The all-time P4P "division dominance" system (100 points for #1, etc.) is interesting too. Sounds like it would put a much higher reward on long-term success and limit the impact of super-fights and astronomic point totals that result from that (which might be good or bad?). Maybe you could combine the systems somehow, or use them to "check" each other.

Just wondering exactly how things are set up now, or if there's anything else you've been working on.

Kudos for putting points available on each boxer's page. That's cool. And the bars are a good idea, but I don't understand why you decided to display point changes logarithmically. Especially for fighters with a lot of points, those bars just show slivers of red and green, even for pretty dramatic point swings.


Monty

The minimum point amount and long careers probably caused issues with this one.

I apologize if this questions has already been asked and I missed it. I thought I understood the system. However, I always seem to see strange rankings at the lower divisions. Why is 12-0-3 Ryo Miyazaki, who hasn't beaten anyone notable, ranked #2 at light flyweight with 228 points but Ivan Calderon, who has had a borderline HOF career, is a former longtime champion and just lost for the first time, at #10 with 168 points? And why is Calderon behind 6-0 Katzuto Ioka (#9), with 170 points. Is there a different point system based on the location of the fighter?

The Japanese domestic boxers are good at pooling points. They fight only other Japs and they do it often, thus building their point totals higher and higher by only fighting within their group. Sometimes they are that good on the global scale.. sometimes they aren't. Same shit happens in South Africa.

Calderon has built his legacy on title defenses with decisions against soft opponents. Not exactly point-building wins.

MontyCircus wrote:

JCS wrote:The Japanese domestic boxers are good at pooling points. They fight only other Japs and they do it often, thus building their point totals higher and higher by only fighting within their group. Sometimes they are that good on the global scale.. sometimes they aren't. Same shit happens in South Africa.

So if two fighters, say Fighter A and Fighter B take turns knocking each other out...then their point totals escalate over time? That doesn't seem right. Or does it need a larger group to have that effect?

That would take dozens and dozens of fights for two fighters to build up something. But yes, a larger group is required.

MontyCircus wrote:

JCS wrote:

MontyCircus wrote: Hey computerrank,

Whatever happened to this system? It's necessary in my opinion. And also for the all-time P4P list as well.

It makes sense for points to drop as a fighter moves up, but not on P4P lists.

Cutting those point cuts out of the division jumps would definitely be an improvement I think.

The all-time P4P "division dominance" system (100 points for #1, etc.) is interesting too. Sounds like it would put a much higher reward on long-term success and limit the impact of super-fights and astronomic point totals that result from that (which might be good or bad?). Maybe you could combine the systems somehow, or use them to "check" each other.

Just wondering exactly how things are set up now, or if there's anything else you've been working on.

Kudos for putting points available on each boxer's page. That's cool. And the bars are a good idea, but I don't understand why you decided to display point changes logarithmically. Especially for fighters with a lot of points, those bars just show slivers of red and green, even for pretty dramatic point swings.


Monty

The minimum point amount and long careers probably caused issues with this one.

I don't see much problem with either the minimum point or long careers. Isn't the point of an all-time P4P list to figure out who was elite and dominated for the longest time? 500 points is pretty high. Right now only 30 active fighters have 500 points.

And, of course, it doesn't make a lick of sense for a fighter to go DOWN on the current P4P list because he went up a division. That's usually when fans and media start to "ooh" and "ahh" because the fighter is challenging himself.

Anyway, like I said, I'm not entirely sure which systems are in place right now. It could very well be that that's already been implemented (but it doesn't seem like it).

Monty

What I'm saying is... You have a guy who got 501 points at Light Flyweight, 8 years ago who now fights at Featherweight... and a guy who got 499 points at Light Flyweight only two years ago and now fights at Middleweight.

This is hypothetical, but the first guy gets credit while the second doesn't.. even though the second was more recent.

JCS wrote:The Japanese domestic boxers are good at pooling points. They fight only other Japs and they do it often, thus building their point totals higher and higher by only fighting within their group. Sometimes they are that good on the global scale.. sometimes they aren't. Same shit happens in South Africa.

Calderon has built his legacy on title defenses with decisions against soft opponents. Not exactly point-building wins.

I'm not a huge Calderon fan, so this isn't specifically about him. But just in the last three years he's twice beaten Cazares, he beat Mayol, Dieppa and Esquer. All are far better wins than anyone 12-0-3 Ryo Miyazaki or 6-0 Katzuto Ioka have ever fought. I understand pooling, but it seems illogical that victories over those opponents wouldn't, at a minimum, put Calderon above Ioka, who has only had six wins total and none that were not against domestic nobodies.
I guess it's just a quirk in the system because I can't imagine anyone would really argue Ioka's or Miyazaki's accomplishments are greater than Calderon's or that they have beaten better opponents.

giacomino wrote:

JCS wrote:The Japanese domestic boxers are good at pooling points. They fight only other Japs and they do it often, thus building their point totals higher and higher by only fighting within their group. Sometimes they are that good on the global scale.. sometimes they aren't. Same shit happens in South Africa.

Calderon has built his legacy on title defenses with decisions against soft opponents. Not exactly point-building wins.

I'm not a huge Calderon fan, so this isn't specifically about him. But just in the last three years he's twice beaten Cazares, he beat Mayol, Dieppa and Esquer. All are far better wins than anyone 12-0-3 Ryo Miyazaki or 6-0 Katzuto Ioka have ever fought. I understand pooling, but it seems illogical that victories over those opponents wouldn't, at a minimum, put Calderon above Ioka, who has only had six wins total and none that were not against domestic nobodies.
I guess it's just a quirk in the system because I can't imagine anyone would really argue Ioka's or Miyazaki's accomplishments are greater than Calderon's or that they have beaten better opponents.

You also have to consider the types of wins. A KO is worth more than a split decision against the same opponent.

squiggy wrote:I noticed this when looking at Margarito's record and thinking, wow, I really thought he'd been coming into fights with higher ratings than that... Then I see the ratings through Pacquaio's history are a lot lower than I remember them, too. Sorry if I'm missing some place where it was explained, but did the new change mostly affect the ratings of a certain group of fighters? And did it mostly cause ratings to go down? What is it that changed, if it can be put simply?

Recent tests with Boxrec ratings unveilled, that new and low rated boxers were allowed to get too much additional points against opponents with too few recent achievements.

The only point changed is the requirements to get some of the additional points (additional to the standard rating formula) for boxers below 50 points.

This additional point mechanism is necassary to get the ratings running at all - as all boxers start at 0 points by definition.

old release:

- defeating an opponent with at least 5 wins gets up to 50 additional points
- the amount in reduced in relation to less than 5 wins in relation to full 5 wins
- the amount is reduced in relation to less rounds boxed compared with full to 12 rounds
- the amount is reduced in relation to the clearness of a decision on points compared with a clear win
- the amount is recuded in relation the rating already achieved before the bout - down to no additonal points for already 50 points achieved before the bout

new release:

- defeating an opponent with at least launch status 4 gets up to 50 additional points
- the amount in reduced in relation to the opponents launch status in relation to full launch status 4
- the amount is reduced in relation to less rounds boxed compared with full to 12 rounds
- the amount is reduced in relation to the clearness of a decision on points compared with a clear win
- the amount is recuded in relation the rating already achieved before the bout - down to no additonal points for already 50 points achieved before the bout

The launch status system is a ladder system, where a boxer

- steps up by defeating an opponent at higher or same level
- steps down by losing to an opponent at lower or equal level

@MontyCircus

Welcome back here.

Current P4P ratings

As JCS said, the current P4P system was not changed. It simply shows the current all-division-list, the highest rated current boxers.

I was not convinced, the change would be the better option.

All-time ratings

The all-time ratings were changed to a mixed formula:

The All Time Rating for a boxer is the sum of 33% of the sum of annual rank points he gets for his annual ratings - and 33% of the sum of annual rank points of best defeated opponents - and 14% of his career top rating:

1. the annual rating is the rating at the end of every year the boxer was active
2. annual rank points = 200 / annual rank in division cluster
3. the value of 200 annual points for the top boxer is reduced, if the annual rating of #10 in the division cluster is less than 100 for men, - and if the annual rating of #5 in the division cluster is less than 50 for women

Launch mechanism

Actually I am working on improvements for the launch mechanism of the ratings - essential, as all boxers start with 0 points.

This is a tricky issue, as you have

- to provide additional points to the ratings, at least in the very lower regions
- to avoid these points for performances and opponents not worthy enough

P.S.:

Graphics in bout ratings

The scale is logarithmic as the ratings are based on ratios not on differences. The logarithmic scale makes these ratios to differences of same significance.

I assume a ranking update happened this morning.. wondering if there is an issue.

This guy jumped to 22 points upon beating a 0 point fighter

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

This guy jumped 3.

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

I assume its because the first loser was "higher ranked" in the launch zone than the second loser. Is this accurate?

JCS wrote:I assume a ranking update happened this morning.. wondering if there is an issue.

This guy jumped to 22 points upon beating a 0 point fighter

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

This guy jumped 3.

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

I assume its because the first loser was "higher ranked" in the launch zone than the second loser. Is this accurate?

The first case is an issue ... will look.

computerrank wrote:

JCS wrote:I assume a ranking update happened this morning.. wondering if there is an issue.

This guy jumped to 22 points upon beating a 0 point fighter

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

This guy jumped 3.

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

I assume its because the first loser was "higher ranked" in the launch zone than the second loser. Is this accurate?

The first case is an issue ... will look.

The release is corrected and is launched.

Check out Minimumweight, Martin. Surprised to see Kokietgym at #2! I think the new launch mechanism might be too aggressive.

JCS wrote:Check out Minimumweight, Martin. Surprised to see Kokietgym at #2! I think the new launch mechanism might be too aggressive.

Jason,

oh yes, our well-known friend ... it is.

I will work on this.

I had to cut off a lot off unworthy points in the very low regions. But to keep the system alive, as it needs points from nowhere in anyway, I became a bit too aggressive in the higher regions.

Thanks for your hints ... and enjoy your days!

Hi from a first-time poster, some-time reader, curious and impressed with an evidently workable (and broadly sane) objective rating system.

Forgive me that my first post is to suggest that something appears to be going awry.

computerrank wrote: The release is corrected and is launched.

Sir, since your corrected release of the ratings of the 12th, or a few days thereafter, there appears to have been an accumulation error occurring. I see some ratings increasing daily, now having surpassed what they were before your correction of the 12th took away often 5%, occasionally more than 10%, of the previous values.

Also, regarding the old (pre-12th Dec) system:

computerrank wrote: old release:

- defeating an opponent with at least 5 wins gets up to 50 additional points
- the amount in reduced in relation to less than 5 wins in relation to full 5 wins
- the amount is reduced in relation to less rounds boxed compared with full to 12 rounds
- the amount is reduced in relation to the clearness of a decision on points compared with a clear win
- the amount is recuded in relation the rating already achieved before the bout - down to no additonal points for already 50 points achieved before the bout

It was my experience in following actual exmaples that the additional points were still being acquired by boxers with up to 100, not 50, points achieved before the bout. (With less certainty, I'm not sure the number of rounds scheduled for the bout was effecting any difference, either.)

I can't comment on the same regarding the new release as I've been trying to reckon on a cause for the apparent daily accumulation instead of playing with the arithmetic of early boxing careers.

Hope you resolve the issues (or, if none to resolve, can shed light on my misunderstandings) soon.

Regards

Forest Bob
p.s. Your phrase "in relation to", were it means a linear relation, is perhaps best expressed in non-technical English as "in direct proportion to".

forestbob wrote:Hi from a first-time poster, some-time reader, curious and impressed with an evidently workable (and broadly sane) objective rating system.

Forgive me that my first post is to suggest that something appears to be going awry.

computerrank wrote: The release is corrected and is launched.

Sir, since your corrected release of the ratings of the 12th, or a few days thereafter, there appears to have been an accumulation error occurring. I see some ratings increasing daily, now having surpassed what they were before your correction of the 12th took away often 5%, occasionally more than 10%, of the previous values.

Also, regarding the old (pre-12th Dec) system:

computerrank wrote: old release:

- defeating an opponent with at least 5 wins gets up to 50 additional points
- the amount in reduced in relation to less than 5 wins in relation to full 5 wins
- the amount is reduced in relation to less rounds boxed compared with full to 12 rounds
- the amount is reduced in relation to the clearness of a decision on points compared with a clear win
- the amount is recuded in relation the rating already achieved before the bout - down to no additonal points for already 50 points achieved before the bout

It was my experience in following actual exmaples that the additional points were still being acquired by boxers with up to 100, not 50, points achieved before the bout. (With less certainty, I'm not sure the number of rounds scheduled for the bout was effecting any difference, either.)

I can't comment on the same regarding the new release as I've been trying to reckon on a cause for the apparent daily accumulation instead of playing with the arithmetic of early boxing careers.

Hope you resolve the issues (or, if none to resolve, can shed light on my misunderstandings) soon.

Regards

Forest Bob
p.s. Your phrase "in relation to", were it means a linear relation, is perhaps best expressed in non-technical English as "in direct proportion to".

I am a bit short of time these days - but I will come back to your message - I appreciate it, thanks a lot!

Found another issue Martin

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

Boxer comes back from inactivity at 5 points, but had 0 prior to.

JCS wrote:Found another issue Martin

http://boxrec.com/list_bouts.php?human_ ... &cat=boxer

Boxer comes back from inactivity at 5 points, but had 0 prior to.

This is due to the update modus for all bouts older than current_year -2.

These bouts are updated only on days with day's last figure equal to bout year's last figure.

So the next update will be on Jan 2, 2011 as the bout was in 2002 - and the ratings release update was on Dec 23, 2010.

All done to save server ressources during the daily ratings update.

Addition:

No, Jason, there is something off - I will look...

Hmmm, nuts.

Just lost (timed-out, I guess) a big long post saying that whilst, as there have been no further step-ups in ratings, I guess my accumulation theory was wrong, but gave examples of De La Hoya's and Chavez's ratings increases since the initial post-12th Dec drop.

Also, having investigated Boxrec's records for possible erratic increases, found instead (and detailed, whence the time taken) possible cases of ratings points erroneously held-back from them both. Some cases random; quite a few for Chavez (yet none for De La Hoya) where he's only gaining half the points he ought (opponent losing correct amount); and an apparent non-zero-sum type of ratings change not at all referred to in http://boxrec.com/media/index.php/BoxRe ... escription

Q: am I correct to understand that, excluding bonus ratings points for low-rated victorious boxers, the changes in rating for each and every bout are zero-sum? (i.e. points gained by boxer A equals points lost for boxer B)

regards

Forest Bob

forestbob wrote:
Q: am I correct to understand that, excluding bonus ratings points for low-rated victorious boxers, the changes in rating for each and every bout are zero-sum? (i.e. points gained by boxer A equals points lost for boxer B)

No, this is incorrect. There are certain instances of scenario handling that create positive and negative sums. Bottom feeding being an example of the first, and inactivity acceleration being an example of the latter.

JCS wrote:

forestbob wrote:
Q: am I correct to understand that, excluding bonus ratings points for low-rated victorious boxers, the changes in rating for each and every bout are zero-sum? (i.e. points gained by boxer A equals points lost for boxer B)

No, this is incorrect. There are certain instances of scenario handling that create positive and negative sums. Bottom feeding being an example of the first, and inactivity acceleration being an example of the latter.

Eh? Your terms are obscure to me, but having a guess:

1. Isn't what you call "bottom feeding" eaxctly what I'd already allowed for?

2. What is "inactivity acceleration", please? I can't make "inactivity acceleration" or "... deceleration" quite mean anything to me in this context. Optimistically (or desperately?) grasping at a straw to beat you back with: if you mean a drop in rating due to inactivity (or similar for activity having been only against opponents sub 25% of that of the boxer-in-question), in the boxer's records that rating adjustment is calculated before the bout and accounted for in the pre-bout rating, so the following bout itself is still zero-sum (all else being equal).

Like Forestbob, I have noticed that at some point recently, many ratings have increased to slightly more than the pre-Dec 12 values. I am not inclined to try and understand why this happened like some others here, as I simply trust that the rankings are in good hands, but I use Boxrec ratings points as a reference point for compiling my own divisional rankings for my blog based on different criteria, and every time a fighter's historical ratings points change it throws off my consistency and makes me recompile everything, so I'm just wondering if the current criteria can be considered stable or not (ie are more significant updates to the algorithm anticipated in the near future)- and if not- when the final tweaks are anticipated so I can work around that date.

Thank you.

forestbob wrote:

JCS wrote:

forestbob wrote:
Q: am I correct to understand that, excluding bonus ratings points for low-rated victorious boxers, the changes in rating for each and every bout are zero-sum? (i.e. points gained by boxer A equals points lost for boxer B)

No, this is incorrect. There are certain instances of scenario handling that create positive and negative sums. Bottom feeding being an example of the first, and inactivity acceleration being an example of the latter.

Eh? Your terms are obscure to me, but having a guess:

1. Isn't what you call "bottom feeding" eaxctly what I'd already allowed for?

2. What is "inactivity acceleration", please? I can't make "inactivity acceleration" or "... deceleration" quite mean anything to me in this context. Optimistically (or desperately?) grasping at a straw to beat you back with: if you mean a drop in rating due to inactivity (or similar for activity having been only against opponents sub 25% of that of the boxer-in-question), in the boxer's records that rating adjustment is calculated before the bout and accounted for in the pre-bout rating, so the following bout itself is still zero-sum (all else being equal).

The rating adjustment is zero-sum, except for:

- additional points - part of bout evaluation
- drop for inactivity - independent of bout evaluation
- drop for missing opponent quality - independent of bout evaluation

marcianofan wrote:Like Forestbob, I have noticed that at some point recently, many ratings have increased to slightly more than the pre-Dec 12 values. I am not inclined to try and understand why this happened like some others here, as I simply trust that the rankings are in good hands, but I use Boxrec ratings points as a reference point for compiling my own divisional rankings for my blog based on different criteria, and every time a fighter's historical ratings points change it throws off my consistency and makes me recompile everything, so I'm just wondering if the current criteria can be considered stable or not (ie are more significant updates to the algorithm anticipated in the near future)- and if not- when the final tweaks are anticipated so I can work around that date.

Thank you.

The test server is down at the moment.

So some final adjustments for the last rating program release are postponed to January, I fear.

computerrank wrote:

marcianofan wrote:Like Forestbob, I have noticed that at some point recently, many ratings have increased to slightly more than the pre-Dec 12 values. I am not inclined to try and understand why this happened like some others here, as I simply trust that the rankings are in good hands, but I use Boxrec ratings points as a reference point for compiling my own divisional rankings for my blog based on different criteria, and every time a fighter's historical ratings points change it throws off my consistency and makes me recompile everything, so I'm just wondering if the current criteria can be considered stable or not (ie are more significant updates to the algorithm anticipated in the near future)- and if not- when the final tweaks are anticipated so I can work around that date.

Thank you.

The test server is down at the moment.

So some final adjustments for the last rating program release are postponed to January, I fear.

Okay thanks. I'll work around it for the moment and monitor this thread for updates come January.