Ratings - please read before commenting

[JCS]

In all seriousness, I expected Ngannou to at least slip into the Top 100....

My guess is that he is limited by the fact that he has no wins, but I do think this is an extenuating circumstance.

He has no win and he is tied by 3 prior dummy bouts to a low start initialization.

I am confused by this result...

A boxer has an incredible performance in his first bout.. should that not be given a massive amount of credit, similar to a Glicko type rating system -- i.e. the boxer's rating is more in question than someone who has 30 fights, 2 or 3 in the past year?

Is Ngannou's rating more suppressed by the fact that the 3 prior dummy bouts exist? Or that this was not a win.. and there is a penalty as a result that he has no wins?

Logically, I can understand Fury not moving... Though I think he should've been #1 before this bout... I'd be OK with him at #2 or #3 after this result. So #3 seems accurate.

If this was a draw, what would the result have been? What if Ngannou won the split decision with the same cards?

[The Asleep Lamps]

That is a good question. The fight was about a draw so I'd be curious to know where he would have ranked with a draw as well. I'm not sure if we will get an answer from the com-pu-ter though.

[SportsRatings]

In all seriousness, I expected Ngannou to at least slip into the Top 100....

My guess is that he is limited by the fact that he has no wins, but I do think this is an extenuating circumstance.

He has no win and he is tied by 3 prior dummy bouts to a low start initialization.

I am confused by this result...

A boxer has an incredible performance in his first bout.. should that not be given a massive amount of credit, similar to a Glicko type rating system -- i.e. the boxer's rating is more in question than someone who has 30 fights, 2 or 3 in the past year?

Is Ngannou's rating more suppressed by the fact that the 3 prior dummy bouts exist? Or that this was not a win.. and there is a penalty as a result that he has no wins?

Logically, I can understand Fury not moving... Though I think he should've been #1 before this bout... I'd be OK with him at #2 or #3 after this result. So #3 seems accurate.

If this was a draw, what would the result have been? What if Ngannou won the split decision with the same cards?

If he had won, the "winner above loser" rule would put him at least at #3, and interestingly the three "dummys" he fought would also end up in the top 10.

[SportsRatings]

I do have to say, I do my own boxing ratings (HW only) and my system is not prepared to rank someone like Ngannou. He won't even be ranked in my system, which for simplicity only logs about 10 bouts per month, and a fighter doesn't get put in the system until he gets a win or a draw with another fighter in the system.

Someone who comes close to getting a draw or win in his first fight, with the champ, is just out of luck. The assumption is that anyone fighting the champ has had some prior success and would already be rated, but in today's world you have the McGregors and Ngannous who jump the line from MMA. It actually would work fine if the newcomer wins; at least that's a tangible result, but a very close loss, which implies some level of skill that should be recognized, is just hard to deal with in a simple system.

Boxrec's system is more involved that mine, uses every known fight, and recognizes close finishes, so I would think it should be able to capture Ngannou's level better, but #587 is essentially where Ngannou is in my ratings, too (I track only a few hundred 'active' fighters right now)

So while I understand why this is an edge case that hasn't really come up before, maybe it needs a tweak that makes sense because this sort of thing—"novice" fighter vs. a current champ—might become more common

[JCS]

I do have to say, I do my own boxing ratings (HW only) and my system is not prepared to rank someone like Ngannou. He won't even be ranked in my system, which for simplicity only logs about 10 bouts per month, and a fighter doesn't get put in the system until he gets a win or a draw with another fighter in the system.

Someone who comes close to getting a draw or win in his first fight, with the champ, is just out of luck. The assumption is that anyone fighting the champ has had some prior success and would already be rated, but in today's world you have the McGregors and Ngannous who jump the line from MMA. It actually would work fine if the newcomer wins; at least that's a tangible result, but a very close loss, which implies some level of skill that should be recognized, is just hard to deal with in a simple system.

Boxrec's system is more involved that mine, uses every known fight, and recognizes close finishes, so I would think it should be able to capture Ngannou's level better, but #587 is essentially where Ngannou is in my ratings, too (I track only a few hundred 'active' fighters right now)

So while I understand why this is an edge case that hasn't really come up before, maybe it needs a tweak that makes sense because this sort of thing—"novice" fighter vs. a current champ—might become more common

Once upon a time, I believe BoxRec handled this situation more appropriately.

Years ago, I called out Lomachenko as a great example... He debuted and won against a 25-2 fighter. Given his pedigree and obvious skill, it was unfair to the opponent to essentially treat this as a loss to a standard debuting fighter. Thus, you should treat very early career fighters similar to how Glicko handles inexperience... you move them more than you move the more experienced opponent, because of greater uncertainty.

Even though this was not a win by Ngannou, I think coming this close to a win should absolutely mean something. Perhaps unexpected successes in the very early career should -boost- the quality of the dummies involved in those dummy bouts and/or if it is some kind of minimum win penalty, I think a close loss should earn some kind of partial credit.

[pugilisticspecialist]

he is tied by 3 prior dummy bouts to a low start initialization.

What does this mean?

[computerrank]

he is tied by 3 prior dummy bouts to a low start initialization.

What does this mean?

see https://boxrec.com/wiki/index.php/BoxRe ... escription:

Seeding the boxers at start of their career is a critical task

The standard WHR rating seeds a boxer by 1 prior win and 1 prior loss against an opponent with rating 1.

BoxRec seeds the boxers depending on the depth of their win hierarchy. A boxer without win is set to level 0. A win against a level 1 opponent sets him to level 2 etc ...

A boxer with at least level 10 is regarded to be connected to relevant competition.
BoxRec seeds a boxer by 3 prior bouts with with result 0.05 against
an opponent with rating of 1 for connected boxers.
an opponent with rating of 1/3 for a boxer with a close result > 0.13 against a connected opponent.
an opponent with rating of 1/6 for boxer with at least 1 win.
an opponent with rating of 1/20 else.

[computerrank]

I am confused by this result...

A boxer has an incredible performance in his first bout.. should that not be given a massive amount of credit, similar to a Glicko type rating system -- i.e. the boxer's rating is more in question than someone who has 30 fights, 2 or 3 in the past year?

Is Ngannou's rating more suppressed by the fact that the 3 prior dummy bouts exist? Or that this was not a win.. and there is a penalty as a result that he has no wins?

Logically, I can understand Fury not moving... Though I think he should've been #1 before this bout... I'd be OK with him at #2 or #3 after this result. So #3 seems accurate.

If this was a draw, what would the result have been? What if Ngannou won the split decision with the same cards?

If Nganou would have had the SD with same cards the winner_above_loser rule would have brought him to #30. A draw would have brought him to #550.

[JCS]

I am confused by this result...

A boxer has an incredible performance in his first bout.. should that not be given a massive amount of credit, similar to a Glicko type rating system -- i.e. the boxer's rating is more in question than someone who has 30 fights, 2 or 3 in the past year?

Is Ngannou's rating more suppressed by the fact that the 3 prior dummy bouts exist? Or that this was not a win.. and there is a penalty as a result that he has no wins?

Logically, I can understand Fury not moving... Though I think he should've been #1 before this bout... I'd be OK with him at #2 or #3 after this result. So #3 seems accurate.

If this was a draw, what would the result have been? What if Ngannou won the split decision with the same cards?

If Nganou would have had the SD with same cards the winner_above_loser rule would have brought him to #30. A draw would have brought him to #550.

Wow, so the difference between the SD win and draw is massive! I would not expect that at all...

If #30 is an SD win, I think I'd expect a draw to see him get in the Top 100....

[Cap]

Just my humble opinion but while it makes sense to rank active fighters with complete records, trying to rank All-Time boxers is a mug's game. No way should a guy with a 9-7-0 record (Gordon Coghill) be ranked over a guy with a 31-12-4 record (Aussie champion Ern Waddy). Just an example.

[computerrank]

Just my humble opinion but while it makes sense to rank active fighters with complete records, trying to rank All-Time boxers is a mug's game. No way should a guy with a 9-7-0 record (Gordon Coghill) be ranked over a guy with a 31-12-4 record (Aussie champion Ern Waddy). Just an example.

Yes it is a mug's game
But BoxRec doesn't work with the pure record. At a level far below world tops it is all about result against opponents and the career top BoxRec rating.
Here Coghill had better 5 top performances and Waddy had the higher career top rating. Very close ...

[Cap]

I'm puzzled. What were Coghill's top five performances? He beat Charlie Horn (?) a second rater, and got a TKO over Pat Doran who dislocated his shoulder and could not continue. Doran had stopped Coghill in an earlier fight. Coghill KO'd Kelly Mansfield and a very young Alec Pooley. Aside from that he got KO'd by Colin Bell, Fred Fulton, Les O'Donnell, Frank Goddard and Tom Cowler. Are these losses included in his good performances?

[computerrank]

I'm puzzled. What were Coghill's top five performances? He beat Charlie Horn (?) a second rater, and got a TKO over Pat Doran who dislocated his shoulder and could not continue. Doran had stopped Coghill in an earlier fight. Coghill KO'd Kelly Mansfield and a very young Alec Pooley. Aside from that he got KO'd by Colin Bell, Fred Fulton, Les O'Donnell, Frank Goddard and Tom Cowler. Are these losses included in his good performances?

Only the best 5 performances included - losses not considered.

[SportsRatings]

I am confused by this result...

A boxer has an incredible performance in his first bout.. should that not be given a massive amount of credit, similar to a Glicko type rating system -- i.e. the boxer's rating is more in question than someone who has 30 fights, 2 or 3 in the past year?

Is Ngannou's rating more suppressed by the fact that the 3 prior dummy bouts exist? Or that this was not a win.. and there is a penalty as a result that he has no wins?

Logically, I can understand Fury not moving... Though I think he should've been #1 before this bout... I'd be OK with him at #2 or #3 after this result. So #3 seems accurate.

If this was a draw, what would the result have been? What if Ngannou won the split decision with the same cards?

If Nganou would have had the SD with same cards the winner_above_loser rule would have brought him to #30. A draw would have brought him to #550.

Wow, so the difference between the SD win and draw is massive! I would not expect that at all...

If #30 is an SD win, I think I'd expect a draw to see him get in the Top 100....

In my ratings (where Fury is #1, Usyk #2), a win would have put Ngannou #2, drop Fury to #3, & Usyk would end up #1. A draw would have put Ngannou at #3, Fury stays at #1 but loses points.

[JCS]

If Nganou would have had the SD with same cards the winner_above_loser rule would have brought him to #30. A draw would have brought him to #550.

Wow, so the difference between the SD win and draw is massive! I would not expect that at all...

If #30 is an SD win, I think I'd expect a draw to see him get in the Top 100....

In my ratings (where Fury is #1, Usyk #2), a win would have put Ngannou #2, drop Fury to #3, & Usyk would end up #1. A draw would have put Ngannou at #3, Fury stays at #1 but loses points.

That feels like an extreme "ladder" style of ranking.

Though IBO seems to have taken a similar approach... with Ngannou at #5 behind the two champs Fury/Usyk, Joshua and Zhang.
https://www.iboboxing.com/ibotop100male

I actually forgot they existed until just now..

[SportsRatings]

Wow, so the difference between the SD win and draw is massive! I would not expect that at all...

If #30 is an SD win, I think I'd expect a draw to see him get in the Top 100....

In my ratings (where Fury is #1, Usyk #2), a win would have put Ngannou #2, drop Fury to #3, & Usyk would end up #1. A draw would have put Ngannou at #3, Fury stays at #1 but loses points.

That feels like an extreme "ladder" style of ranking.

Though IBO seems to have taken a similar approach... with Ngannou at #5 behind the two champs Fury/Usyk, Joshua and Zhang.
https://www.iboboxing.com/ibotop100male

I actually forgot they existed until just now..

I think that's defensible, too. A fighter who is, by all evidence, a hair's breadth from being as good as the #1 guy, can be #5.

Or having him anywhere in the top 100 seems ok, too.

And really, a draw could fairly put him exactly equal to Fury (the most extreme ladder!). Or anywhere behind him, depending on how much you discount it being just one result to go on.

It's an outlier case, so most systems aren't built to handle it, and it's hard to intuitively know what feels right in order to make sure the system works.

The closest thing I've seen is in 1922 when Harry Wills (#1) fought Bill Tate (#132) and lost due to an alleged foul. They fought against 4 days later to a dirty Draw. Then later in the month Wills beat Sam Langford so by my next (monthly) rankings Wills was still at #1, Bill Tate at #2 by a narrow margin.

[computerrank]

Considering the Fury vs Ngannou case and WHR ratings I found that all depends on the number of prior bouts, which bind the ratings to an initial level.

The original WHR ratings use 2 prior bouts, a win and a loss against a virtual opponent of rating 1.

I once found a better winner prediction ratio using 3 prior bouts.

But I never tested a smaller number of prior bouts. And now voila I found an increased winner prediction ratio using a much smaller value of only 0.5 prior bouts. It indeed improves the winner prediction ratio from 85.67 to 85.79 percent.

And it clearly shows a looser binding of the ratings to an initial value.

It lets Ngannou left at #16 after his SD loss against Fury ...

[JCS]

#16 seems like a decent area for him to land... far better than #587...

Does Ngannou still land at #3 with a draw?

[computerrank]

#16 seems like a decent area for him to land... far better than #587...

Does Ngannou still land at #3 with a draw?

that would bring him to #5 and let Fury at #3 ...

and an equivalent SD in favour of Ngannou would bring him to #3 and let Fury at #4 ...

[SportsRatings]

Considering the Fury vs Ngannou case and WHR ratings I found that all depends on the number of prior bouts, which bind the ratings to an initial level.

The original WHR ratings use 2 prior bouts, a win and a loss against a virtual opponent of rating 1.

I once found a better winner prediction ratio using 3 prior bouts.

But I never tested a smaller number of prior bouts. And now voila I found an increased winner prediction ratio using a much smaller value of only 0.5 prior bouts. It indeed improves the winner prediction ratio from 85.67 to 85.79 percent.

And it clearly shows a looser binding of the ratings to an initial value.

It lets Ngannou left at #16 after his SD loss against Fury ...

Whoa, so both 1 more, or 1.5 fewer "binding" bouts improves prediction? Interesting.

I say run with it, #16 sounds just right for Ngannou. Keeps an 0-1 guy out of the top 15 but rewards him for being this close to beating Fury.

[SportsRatings]

#16 seems like a decent area for him to land... far better than #587...

Does Ngannou still land at #3 with a draw?

With 3 prior bouts the draw put him at only #550. #5 is a big improvement there, too.

[JCS]

#16 seems like a decent area for him to land... far better than #587...

Does Ngannou still land at #3 with a draw?

that would bring him to #5 and let Fury at #3 ...

and an equivalent SD in favour of Ngannou would bring him to #3 and let Fury at #4 ...

[Manrae]

Jermell Charlo ranked 3rd at 154?
Gilberto Ramirez #1 at cruiserweight?

At what point does this nonsense end?
If the rankings do not reflect common sense, what good are they?

[computerrank]

Jermell Charlo ranked 3rd at 154?
Gilberto Ramirez #1 at cruiserweight?

At what point does this nonsense end?
If the rankings do not reflect common sense, what good are they?

I guess, it is rather difficult to comply with something like common sense. Even more, what would common sense mean regarding a ratings algorithm? BoxRec relies on choosing the always best proven algorithm available.

[JCS]

When are you launching the new version w/ the lower dummy fights?

Is there other notable large movement by fighters as a result?