Ratings - please read before commenting - Archived

[computerrank]

Can you explain why Dzemal Bosnjak gained so many points for defeating Osei

http://boxrec.com/en/boxer/834109

The requirement for additional points may have to be restricted even more. So that the winner has to be in state 1 too at least (just as currently the loser has to). This would avoid such a points win for Bosnjak.

[computerrank]

@Teflondon19

Hi computerank.... i was just wondering if you could explain how its feasible that someone with 16.92 points defeats someone with 6.43 points by ko and recieves just 1.08 points for it....But a fighter who has 0.045 points beats a fighter with 0.053 points.....and recieves 24.12 points for their win?...thats over a 58000% increase....theres quite a few of these popping up.

http://boxrec.com/en/event/786217/2342146

as you can see...no-one is connected.

RESPONSE

I will answer in parts for better communication.

In the first bout no additional points were in account. The loser 6 points had a prebout rating of only 35 percent of the winner's 17 points. So the winner got only about 1 point for it. Just as the formula says.

In the second bout the loser WAS connected and in state 1. So the winner could get additional points - up to about 25 - as the description and formula says - even, when the losers rating points were much lower.

The ratings need such extra point injection in order to start from scratch with the first recorded bout with all boxers at 0. And in order to compensate for point losses in the sytem due to inactivity, missing opponent quality and career end.

The other point is, that I will restrict the additional point wins to winners in at least state 1 too. This will prevent most of the winning pointless boxers from getting additional points, even against connected boxers.

There is a network of connected boxers, there is a state system for the boxers with start at 0 and there is a rating system for the boxers with start at 0.

I invite you, to go into the description again.

[computerrank]

@Teflondon19

ALSO...…..ANDMOST IMPORTANTLY …..did you mention that a fighter has an increase of ten times their amount, from the previous update to this one?...so basically you've multiplied the fighters previous score at their last update by 10?...so someone sitting on around 3 points should have 30+? - is this what your saying?...if it is, then why was alex dilmaghani on 4 then 3 points on the last update.....and now sits on 17?... I mean why is he the only fighter to have gone down 60 places and the rest didn't really fluctuate too much....I don't want to insinuate bias, I really don't, but come on theres no way to defend what ive just shown you, even though I feel you will, with some extrangeous 'magical' factor.......its a shame really, and don't get me wrong its good of you and john and claude and everyone else who contributes to this website, because its hard work, but when things like this occur it misleads 'casual fans' and diminishes some fighters achievements and inflates others, when the system shouldn't. That's my whole point.

RESPONSE

You embezzle, that the new release already restricted the requirements to be connected. You should read the messages more thoroughly, I guess.

So the points level was up by factor 10. But some boxers were no longer connected with the new release and so produced less additional points - giving some boxers less than factor 10.

[computerrank]

@Teflondon19

Please stop flooding the thread.

I would appreciate, if you would thoroughly study the messages. I said to answer in parts, step by step.

I need my time.

Some patience and good manners appreciated.

[oogiebe]

AKA - you cant answer/argue the points,

Please don't insult me with the 'reading more' jibe....ive dissected your own formula....showed you with factual mathematical evidence that there are mistakes with it, and you have continued to ignore.

As i said before, please appreciate that you are talking to someone who has a better understanding of mathematics and statistics than yourself. I also believe i know about boxing, but thats subjective. I feel you are someone that is unwillingly to budge on glarying mistakes, so i will not take the rankings seriously anymore, something i feel the majority of knowledgeable users are starting to do.

now watch alex dilmaghani go down 100 places on the next 'update' because i voiced a factual critiscm....WATCH

Tef - Please stop. I don't want to unsubscribe from this thread, but you are making me consider it.

[computerrank]

AKA - you cant answer/argue the points,

Please don't insult me with the 'reading more' jibe....ive dissected your own formula....showed you with factual mathematical evidence that there are mistakes with it, and you have continued to ignore.

As i said before, please appreciate that you are talking to someone who has a better understanding of mathematics and statistics than yourself. I also believe i know about boxing, but thats subjective. I feel you are someone that is unwillingly to budge on glarying mistakes, so i will not take the rankings seriously anymore, something i feel the majority of knowledgeable users are starting to do.

now watch alex dilmaghani go down 100 places on the next 'update' because i voiced a factual critiscm....WATCH

Ok, this is my last response to you. Fare well.

[JCS]

@oogie...sorry but that system is so wrong, even in the face of irrefutable evidence.

@computerank - have the decency to admit when your wrong .....youre bias, and manual changes alienate anyone who questions your opinion with factual evidence.....then again your german., you have a history of that kind of thing, like i said il await your next 'updates' where u will show your bias towards certain fighters again, whilst ignoring the facts infront of u and babble incorrect statistics and 'this fighter is connected today but wasn't last week' stuff to defend what you are doing.


You are an enthusiast without the necessary level of statistical qualifications to warrant your position, unfortunately for you, as shown, your system needs serious overhauling.

You show a unique blend of jerk and jackass.. keep it up.

How do "statistical qualifications" factor into a ranking system for a sport that does not officially consider said system in its matchmaking?

[oogiebe]

OMG! Please Mods...get rid of this guy.

[jamamb]

hes been sending me pms and searching my post history

[oogiebe]

hes been sending me pms and searching my post history

Something really wrong with that bloke!

[JCS]

ahh if it isn't the mma man who single handley invented the worlds greatest software....whilst simultaneously destroying his liver with roids and drink......u show a blend of stupid and American......actually I think that's one thing :)

Well when you operate a system based on statistics...incorrect ones as I have shown multiple times throughout this thread, you have to have some understanding of it.

this system is run on statistics isn't it.....that's why it matters to know what ur doing......I hope I spelt it out for u

come back a 10000 times I wont bother with u, ur not worth the effort

Good.

Hey if you get a good on deal on roids, let me know.. never tried em. Not saying I wouldn't though.

[Manrae]

Until next week.

Things got a bit heated in here

Any updates for the annual rankings?

[computerrank]

Things got a bit heated in here

Any updates for the annual rankings?

Still some tweaks in work. Let's wait for it.

[VegetaPhD]

A question on the points received on fight below which is rather confusing, to me anyway.

http://boxrec.com/en/event/777313/2290896

Ricardo Espinoza Franco defeated Yeison Vargas by second round KO.
Espinoza Franco came into the fight with 72.46 points.
Vargas came into the fight with 2.307 points
A whopping 70.153 point difference. If they were to fight now with those same starting points, Espinoza Franco would be rated 35th and Vargas 244th (ironically, where he is now).
So why is it Espinoza Franco received 19.62 points after the fight?

I have seen many fighters with much less points than Espinoza Franco win by KO against opponents with more points than Vargas, but hardly receive anything (1 or 2 points). How could that be possible?

I don't understand this whole 'connected' thing. Vargas' best opponent was 12-1-1, nearly 3 years earlier. However, he only had 0.286 points to begin with. Although, he did beat someone who had 4.522 points before their fight, but his record was a much less desirable 4-6-1. Strange, is it not?

Many thanks,
Vegeta

[computerrank]

A question on the points received on fight below which is rather confusing, to me anyway.

http://boxrec.com/en/event/777313/2290896

Ricardo Espinoza Franco defeated Yeison Vargas by second round KO.
Espinoza Franco came into the fight with 72.46 points.
Vargas came into the fight with 2.307 points
A whopping 70.153 point difference. If they were to fight now with those same starting points, Espinoza Franco would be rated 35th and Vargas 244th (ironically, where he is now).
So why is it Espinoza Franco received 19.62 points after the fight?

I have seen many fighters with much less points than Espinoza Franco win by KO against opponents with more points than Vargas, but hardly receive anything (1 or 2 points). How could that be possible?

I don't understand this whole 'connected' thing. Vargas' best opponent was 12-1-1, nearly 3 years earlier. However, he only had 0.286 points to begin with. Although, he did beat someone who had 4.522 points before their fight, but his record was a much less desirable 4-6-1. Strange, is it not?

Many thanks,
Vegeta

It is a problem, to inject enough additional points into the system - enough to keep the over-all level of ratings, although boxers drop out of the ratings, taking their points out of the system, and boxers lose points due to inactivity and missing opponent quality penalties.

These additional points are not related to the prebout points of the opponents. And they are restricted to boxers, who are connected to the seeded boxers by a chain of good results. This prevents the system from giving points to boxers not connected to the kernel of proved boxers.

I will try a solution to limit the amount of additional points for wins against connected boxers to:
- 1 point at least
- twice the opponent's rating, if higher
- 25 points at most

This will result in a lower rating points level.

[Manrae]

Still some tweaks in work. Let's wait for it.

Cool

[computerrank]

You can test release 41 (proposed) - comments appreciated:

- large additional point wins against opponents with much lower ratings were in question
- this release limits the additional point wins to the level of the defeated opponent

http://151.boxrec.com/~martin/ratings.php

changes;

- additional points now in relation to defeated opponent's pre-bout rating (before: at most 25 points in relation to decision and against a connected opponent in state 1 at least)
-- at least 1 point fix or
-- at least opponent's pre-bout_rating
-- at most 25 points
-- all in relation to decision and against a connected opponent in state 1 at least

- state 1 can now only be reached by a win against a connected opponent (before: by a win against a winner defeating a winner)

[Manrae]

You can test release 41 (proposed) - comments appreciated:

- large additional point wins against opponents with much lower ratings were in question
- this release limits the additional point wins to the level of the defeated opponent

http://151.boxrec.com/~martin/ratings.php

changes;

- additional points now in relation to defeated opponent's pre-bout rating (before: at most 25 points in relation to decision and against a connected opponent in state 1 at least)
-- at least 1 point fix or
-- at least opponent's pre-bout_rating
-- at most 25 points
-- all in relation to decision and against a connected opponent in state 1 at least

- state 1 can now only be reached by a win against a connected opponent (before: by a win against a winner defeating a winner)

I'll have to wait until release to see what historical changes have taken place, but for now...

-Fury's oppoent, Tom Schwartz..... what a drop
-Miguel Berchelt's inflated score takes a hit
-Poor David Benevidez and Moruti Mthalane... some really good fighters taking hits, but it is what it is
-I'm assuming these scores don't reflect this past weekend's action since Lucas Browne is still a bit and David Allen...

Overall I'm liking what I'm seeing. Again, I'm really more interested in how this changes the past.

You know how you factor in inactivity and weight class changes into scoring? Why not age? I understand your position on objective ratings but, aren't inactivity and weight class changing also subjective to how much effect it has on a fighter? Just a thought, not a request

[computerrank]

I'll have to wait until release to see what historical changes have taken place, but for now...

-Fury's oppoent, Tom Schwartz..... what a drop
-Miguel Berchelt's inflated score takes a hit
-Poor David Benevidez and Moruti Mthalane... some really good fighters taking hits, but it is what it is
-I'm assuming these scores don't reflect this past weekend's action since Lucas Browne is still a bit and David Allen...

Overall I'm liking what I'm seeing. Again, I'm really more interested in how this changes the past.

You know how you factor in inactivity and weight class changes into scoring? Why not age? I understand your position on objective ratings but, aren't inactivity and weight class changing also subjective to how much effect it has on a fighter? Just a thought, not a request

Allen vs Browne was not in at that time.
Age in years and aging of an individual boxer is very difficult to catch in numbers.

[Manrae]

Allen vs Browne was not in at that time.
Age in years and aging of an individual boxer is very difficult to catch in numbers.

1) Ah ok
2) Yes, indeed it is. We see some fighters burn out in their 20's while others not until their late 30's, early 40's

One thing I noticed is that different sports have different "prime" ages. Of course there are always outliers but once the new system is in play, I'm gonna make a request to try and see if there's a prime age for boxers.

Based on the ratings now, it seems to be 30 years old, but I'm not a statistician so my word can't be gold on the matter

[Jens S]

Hi there

I have been trying to get the grip of the new system.

Firstly I am very glad, that we now can see decimals on the fighters. It gives a way better understanding of the fighters. Though I would wish, that it was possible to switch, like it is with kilos, lbs og st. Could it be possible to have a choice, where you can pick one decimal after "."? Like 1430.3, 357.3, 25.0, 7.9 and 0.1. It would make it more manageable. Easier to get the picture of a career fast when looking on a record.

About the connection thing. I understand it, as you are trying to avoid the "Ali Raymi"s of boxing. (http://boxrec.com/en/boxer/561186) That is boxers who fight nobodies, who only fought nobodies in some sort of closed circuit. They shouldn't be allowed to rise high in the rankings. And I agree with that position. These fights are often bogus, and former Soviet-state Georgia is an example of that. But I think you may create a bigger problem, with the way the rankings are now.

Here are some examples:
Britton Norwood is a very ordinary fighter. Nothing special, but there nothing wrong with being that.
http://boxrec.com/en/boxer/782991

But he is affecting a lot of fighters ranking.
James Ballard is a fighter, who came in at 9-0 fighting Norwood at light heavyweight. He gets knocked out, but wins the rematch. Norwood came in at 0.050 and goes out at 27.36, Ballard entered at 1.501 and went out at 0.715. Comparing with the rankings Norwood must have risen from number 6-700 to 87 by beating a guy who was around 268. In the rematch Ballard win a clear decision. He now moves up to 7.008, despite Norwood had lost most of his points and end entered at 12.57. This put Ballard around no. 185 in the world. So splitting fight with a very ordinary fighter moved him more than 80 spots up.

Norwood gained a lot on the fights as well. But we’ll come back to that.

After his initial win, Norwood - now a top100 fighter - fought in cruiserweight against David Gilberto Mendoza (2-7) with 0.046 points, which means, Mendoza came in just outside number 1000 of less than 1200 cruisers. Norwood had 18.35 at cruiser. Mendoza wins RTD2 and exits the fight with 34.85 points and a spot around no. 65 in the world. Jumping almost a thousand spots. In his next fight Mendoza fights Norwood, who has lost half his points since their first meeting. Norwood wins TKO4 and moves to 45.89 points. After splitting wins with a fighter outside top1000, he is now close to top50 at ca. no. 53. Norwood now moves down to light heavyweight and is at 68.48 points in his next fight. That is no. 39 in the world.

Norwood now fights Gervonte Davis (not Gervonta) who is 4-7-1. Davis has lost his last three and his last win is 10 years old. He enters at 0.377 points and around no. 350. He beats Davis and goes to 34.91 points and is no. 74 at 175 lbs. Norwood goes to 33.95 points and is no. 75. What happens, if Norwood wins a rematch KO1?

So here are a lot of points in the hands of some very ordinary fighters. And they are going to spread in the rankings of boxrec. Managers/promoters will want to put their fighters in against such highly rated fighters. They are a shortcut to recognition. But they are clearly overrated and easy prey.

If it is not them, then it will be the likes of:
Adewale Masebinu (2-2) #87
http://boxrec.com/en/boxer/726999

Sylvain Laval (3-5-2) #105
http://boxrec.com/en/boxer/795270

Tommy Spahija (5-13-2) #132
http://boxrec.com/en/boxer/555612

That was an example of a group. Here is what I would call a chain example:

Rob Nichols, an ok fighter, starts a chain reaction.
http://boxrec.com/en/boxer/477120

Nichols (5-1 at the time) enter the heavyweight fight against Jared Kilkenny (4-0-2) with 1.357 points. Nichols was ranked around #293. Kilkenny had 0.418 points #350. Nichols wins KO2 and goes to 28.04 points and #102.

Nichols then loses at cruiser to Jordan Smith (2-0-1) who goes from 0.050 to 15.85 points or from #6-700 to #109.

Jordan Smith then loses to Taylor Bull (2-4). Taylor Bull goes from 0.049 points to 33.74 or from ca. #800 to #68 in the world.

Taylor Bull loses to Gary Kopas (6-11) who goes from 0.844 to 42.40 points and from #253 to #55 in the world.

Examples like these are quite frequent, and the points from journeymen like these will spread in the system.
Viacheslav Yarkin is an example: http://boxrec.com/en/boxer/742372 vs Mykyta Trubchanin and Viktor Kornienko
Adam Fitzsimmons: http://boxrec.com/en/boxer/679196 vs Joshua Jenkins
Dave Nash: http://boxrec.com/en/boxer/54718 vs Mark Flynn

These fighters get up to around top100 with a win against a very mediocre fighter, who in reality is nowhere near top100, 200 or 300.

I think, that the idea of stopping stacking up points by using the system is a good idea. But I think, that it has created more problems, than it has solved. I think, that these bouts between fighters, where the points are “exploding” is at best inexplicable. Yes, the system calculates that way. But as a measure stick for the level of the fighters, it is so bad for the credibility of the rankings, because the points is spread on.

I think the intention was great, but there might be a better way to eliminate the Ali Raymis high rankings?

Best regards Jens

[computerrank]

Jens,

Thanks a lot for looking into the ratings and your comments.

In former releases at lot of boxers got not achieved points. Now some boxers with some achievements get too many points in bulks. So I already looked for a solution and set the additional point wins in relation to the pre-bout ratings of the defeated opponent. This is the proposed release 41.

Some post above I already invited to inspect this proposed release. From my perspective a lot of the problems you found with current release 40 is solved there.

Would you please look into it - and give me your comments?

http://151.boxrec.com/~martin/ratings.php

Just select division, sex and nationality or all. You can compare release 40 and proposed release 41. And you can click to the boxer ID and see their records and point developments for both releases.

[computerrank]

You can test release 41 alternative option (proposed) - comments appreciated:

- alternative option for additional points


http://151.boxrec.com/~martin/ratings_test2.php

changes;

- additional points now for defeating an opponent in the top 10 percent ranks of division (not for being connected to other connected boxers by good results)
- this requirement is more transparent for the user and he can directly check it

Just select division, sex and nationality or all. You can compare release 40 and proposed release 41 alternative. And you can click to the boxer ID and see their records and point developments for both releases.

[Manrae]

You can test release 41 alternative option (proposed) - comments appreciated:

- alternative option for additional points


http://151.boxrec.com/~martin/ratings_test2.php

changes;

- additional points now for defeating an opponent in the top 10 percent ranks of division (not for being connected to other connected boxers by good results)
- this requirement is more transparent for the user and he can directly check it

Just select division, sex and nationality or all. You can compare release 40 and proposed release 41 alternative. And you can click to the boxer ID and see their records and point developments for both releases.

I really like this one A LOT more, you're doing an awesome job
This additional points change has me eagerly awaiting a release, I can't wait!

[Jens S]

Well, it is definitely better, though I think, that it still has some weird ratings. Do fighters get the same bonus fighting a top10% fighter? Or does it vary on the opponents ranking?
I think, that the ratings as they used to be were better, but I don't have the time to go into details now.

I did look into the Ali Raymi story. It is extremely suspect. One thing is the results here at Boxrec, but if you combine it with some facts about the situation in Yemen, it looks clear to me, what really happened. No doubt that these fights are fake.

If we start with boxrec:
Before 2011 there was only held four fightcards in Yemen. All with only one fight and with same fighter Awad Tamim of Tanzania in 2004-2005.

Then from 2011 to 2015 80 fightcards are held. More than 100 new Yemini fighters and around 30 somali fighters suddenly appear. Ali Raymi debuts at the first of these 80 fightcards. He fights 25 times winning all on KO. All against Yemen, Somali (and a single Djibuti) fighters. And these fighters only fight each other. The fightcards ends March 28th, 2015 - less than two months before Raymi gets killed in the Yemeni civil war on May 23rd, 2015. After his dead the fightcards disappeared.

Now how come so many new fighters appeared? Ali Raymi was a colonel in the Yemeni army. The army was not only Yemeni soldiers at the time, but also mercenaries from Somalia. These mercenaries were desperate somali refugees, that became soldiers to survive. Starve or kill.

Just to understand it - no Somali boxers ever fought as a pro in Somalia. There has been some who were migrants to South Africa, where there is a boxing culture. There is no such culture in Somalia, and out of 43 fighters 28 were in Yemen. The rest lived in South Africa. There has only been held four fightcards in Somalia. They were all in 1936-1937 when Italy had occupied the country. It was italian fighters fighting each other.

So it looks obvious, that Ali Raymi let his soldiers fight each other and him self (as the highest ranking officer). There are a few of Ali Raymis fights on YouTube. It is very clear, that the opponents didn't try to land a real punch. They didn't want to knockout the commanding officer. Ali Raymi clearly didn't know how to box, but had the stripes on his shoulders, and is seen arriving at a fight surrounded by men with automatic weapons.

All this makes it so more understandable, why "the connection" thing started in the Boxrec ratings.

It is so obviously fraud, and Boxrec should just ask the (so called?) Yemen Boxing Commission proof of these fights. And then delete those fights, that couldn't be proved. That would be the best solution.
I honestly doubt, whether the Yemen Boxing Commission still exists. I bet it was made by... well, on the order of Ali Raymi.

Some of the sources:

About Ali Raymi:
https://en.wikipedia.org/wiki/Ali_Raymi

Somali mercenaries:
http://www.thenewhumanitarian.org/repor ... ercenaries
https://www.pri.org/stories/2011-09-06/ ... ries-yemen

How to beat up a somalis refugee:


Best regards Jens