Ratings - please read before commenting - Archived

[Daedalus]

I have to say, even I'm a bit confused by what is displayed in the rankings versus fighter pages... the movement, etc. And I have my own rating system for MMA, so that says a lot.

For instance, I saw Casimero drop from #2 to #5 this weekend after a KO win. Was he somehow spotted some points, but then reduced after fighting a much lesser rated opponent? Did he just happen to get another penalty on the same date? I don't know.

Why do almost all boxer's current ratings seem far lower than in previous fights? Has Covid reduced activity to the point where it affects the entire landscape? Is WHR so dependent on "future" results that this is somehow treated as missing data.. and therefore, missing points? Is this a glitch in how past points are displayed?

It must be a major headache to go over the same set of questions week after week... but the WHR-based implementation is enough for folks to get a handle on, let alone all of the ancillary things. To understand WHR from a basic standpoint requires some brain power. To understand the intricacies of it, involves knowledge of some pretty advanced math... and then on top of it, all the BoxRec-specific mods. Yikes.

People would be much better off looking at the ranking pages... rather than to try and build fight-by-fight cases for fighters.

not very obvious, I fear ...
All depends on all ... time and opponents ...

The last Boxrec bout ratings and the Boxrec current ratings are tentatively lower than the the bout ratings before, even for a boxer at his best. This is due to the fact, that the current rating is combined with more uncertainty (due to missing affirmative future results) than former results, Boxrec for this reason does not use the full WHR rating, but it uses the full WHR rating deminished by the estimated mean uncertainty.

I think the 50% point drop for 18 months of inactivity is too harsh and doesn't reflect common sense. Was Ray Leonard half-strength when he fought Howard, and quarter-strength vs Hagler? Absolutely not! A 25% or even 50% SRL has no shot against Hagler; maybe Ray was 70-80%. Was Thurman 50% when he fought Lopez? No, and had Keith's points not been shaved, the fight vs Pacquiao would've reflected an even fight than a gross mismatch. Some may say, these are anecdotes using elites and not applicable to everyone. Well, that's the point: the greats take their craft seriously and should be the example. Look at Julio Chavez Jr. He takes a vacation and loses 7 points, then more time off and -1 point; compare that to the 300+ losses by elites which defies reality.
* Therefore, the 50% penalty should be significantly trimmed to 10-20%, or replaced with a fixed point deduction (100-200) with a cap.

[JCS]

Always thought the 50% inactivity (and previously for lack of quality performance) drop was too harsh. These steep, all or nothing penalties, only seem appropriate in a league where the boxers, promoters, etc, abide by these rules. Barring the adjustment of crazy outliers..

It was I that convinced Martin to reduce this if the boxer had success against some level of opposition within the period... so as to not get the full 50%... but full drop applies with total inactivity.

In my system (MMA), I use a gradual approach.... but when you update the ratings everyday (I do weekly), this causes continuous seemingly unexplainable movement and people question this. At 18 months inactivity, my drop is around 25%... most of which is given back if returning fighter has good comeback performance.

[computerrank]

The 50% ratings drop after 18 months complete inactivity is only for the current ratings. This indicates for the current ratings, that there is a critical uncertainty for the former rating.

The WHR ratings are calculated without that drop after the next bout.

[marcianofan]

Hi computerrank, I normally make a list of fights involving top-50 fighters in certain divisions before they happen and the ranks of the fighters at the time, but I fell asleep this weekend and didn't get around to it in time. If you get a chance, would you mind telling me the rankings of the following fighters as of right before their fights (eg Friday morning) this past weekend?

Balasz Bacskei - result not in - was / is #31
Fedor Chudinov - was #5
Umar Sadiq - was #87
Mark Heffron - was #41
Denzel Bentley - was #98
Anthony Yarde - was #26
Dec Spelman - was #110
Ekow Essuman - was #84
Egidijus Kavaliauskas - was #17
Mikael Zewski - was #35

Much appreciated, if you can manage it.

I added the old ranks to you post.
Method: I just set the results to no contest in my test account and so got the old ranks.

Thanks a ton. Hate to ask again so soon, but any chance you could run the same analysis on Tyrone McKenna vs. Ohara Davies and Serge Michel vs. Liam Conroy from Wednesday?

[computerrank]

@marcianofan, I will send you a PM, when I will be back home next week.

[marcianofan]

@marcianofan, I will send you a PM, when I will be back home next week.

Great thanks- I promise I don't intend to ask regularly.

[felinoboxing]

Greetings..Here https://boxrec.com/media/index.php/BoxR ... al_Ratings.
In short and simple words, can you explain what "The Legends Men" means?
It is calculated in another way.
You could show the calculations that result in Hopkins at "18 Top Rank".

[computerrank]

Greetings..Here https://boxrec.com/media/index.php/BoxR ... al_Ratings.
In short and simple words, can you explain what "The Legends Men" means?
It is calculated in another way.
You could show the calculations that result in Hopkins at "18 Tope Rank".

It is explained on the page, you linked:
You will find the boxers with 3 and more top ranks of any weight division.

[Daedalus]

raw prediction rate now 89.8 percent

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

So the whole history raw figures are now nearly 4% better than the adjusted present Boxrec rankings?! That prediction rate is very high indeed.

So the assumption is that 89.8 can go even higher when they too are adjusted?

What changes are you making to get the raw figures better and better?

If you are making changes when are the raw figures considered the modified figures?

The current figures again look fine to me.

The all time figures still look not as good as the existing all time figures...just look at the all time heavyweights....Povetkin?!

With the prediction rate now so high for the current ratings it must be getting to the stage where you are seriously thinking of using them?

I will not propose the WHR ratings to go live before next year. There is no need for hurry, the evaluation is not finished yet.

In order to evaluate the equivalence of a top career nowadays and at former times regarding the WHR all time ratings, I tested the following scenario:

- For W. Klitschko I determined the division rank of his opponents at bout time.
- I created a dummy boxer W.Klitschko, with the same sequence of bouts, but starting in Joe Louis' debut year. A shift of 62 years, from 1996 to 1934.
- For these dummy bouts I substituted his real opponents by opponents, equivalently ranked at the dummy bout time.
- Then I run WHR ratings.

The result:
- W. Klitschko dummy peak rating and all-time rating is slightly higher than in his real career (about 10 percent)


I did the same with a shift to Sam Langford's debut year. A shift of 94 year, from 1996 to 1902.

The result:
- W. Klitschko dummy peak rating and all-time rating is about 25 percent of the rating in his real career. But he is still #26 all time Heavyweights, Boxrec recorded competition was lower at that times, I guess.
- While Sam Langford is #96 with 1/3 of dummy Klitschko's points.

Over all, the WHR all time rankings are quite founded based on the achievements against equivalent ranked opponents. Boxrec cannot regard styles, skills, social restrictions and other imponderabilities.

What happened to this version of WHR? Was it superseded by the current version because the prediction rate is higher? I understand Langford being ranked at top 100 or even 200s (Gene Tunney is the only pre-WWII career HW in the top 300), but 1300's?

[Daedalus]

I'm still hoping someone will tell me what the old prediction rate was in order to compare to the 83-84% of the new system. I think it's pretty relevant, considering we've completely abandoned the idea of ratings reacting to individual fight results logically in order to achieve whatever the difference is. If it's 15%....okay. But if it's 3 or 4%? I don't see how that could be worth it. And how do draws factor in?

Also...how does the predicting model take into account (without hindsight) an Olympic Gold medalist making his debut against a mediocre opponent with some point value, but not much? Or are those not included in the figures?

For the old Boxrec r_41 release:
- raw winner prediction ratio based on the old rating only = 0.7482
- winner prediction ratio for r_41 with 16 additional parameters = 0.8345

For WHR:
- raw winner prediction ratio based on the WHR rating only = 0.8308
- winner prediction ratio for r_41 with 16 additional parameters = 0.8402

So the raw winner prediction ratio of the old rating was much lower by more than 8 percent points (better the error was 25 percent against 17 percent, so the error ratio of the old rating was 50 percent higher).

The raw prediction power of the old rating was much worse than that of the new WHR.

16 additional parameters could nearly compensate that, but WHR is nearly as good without the 16 additional parameters.

2nd question. I'm wanting to discuss some ideas on pre-fight factors, but so there's no duplication, can you please list the 16 parameters if possible, or link if this has been answered before?

[Offthehook]

Greetings..Here https://boxrec.com/media/index.php/BoxR ... al_Ratings.
In short and simple words, can you explain what "The Legends Men" means?
It is calculated in another way.
You could show the calculations that result in Hopkins at "18 Tope Rank".

It is explained on the page, you linked:
You will find the boxers with 3 and more top ranks of any weight division.

why does jono carroll jump up 33 places when he lost a UD to maxi hughes. Its not like he even won on one card.

The ratings are really poor in my opinion

[computerrank]

Greetings..Here https://boxrec.com/media/index.php/BoxR ... al_Ratings.
In short and simple words, can you explain what "The Legends Men" means?
It is calculated in another way.
You could show the calculations that result in Hopkins at "18 Tope Rank".

It is explained on the page, you linked:
You will find the boxers with 3 and more top ranks of any weight division.

why does jono carroll jump up 33 places when he lost a UD to maxi hughes. Its not like he even won on one card.

The ratings are really poor in my opinion

Jono Caroll was #13 before the bout. He went down to #33. The ranking move shown on the Boxrec page is erroneous.

[computerrank]

So the whole history raw figures are now nearly 4% better than the adjusted present Boxrec rankings?! That prediction rate is very high indeed.

So the assumption is that 89.8 can go even higher when they too are adjusted?

What changes are you making to get the raw figures better and better?

If you are making changes when are the raw figures considered the modified figures?

The current figures again look fine to me.

The all time figures still look not as good as the existing all time figures...just look at the all time heavyweights....Povetkin?!

With the prediction rate now so high for the current ratings it must be getting to the stage where you are seriously thinking of using them?

I will not propose the WHR ratings to go live before next year. There is no need for hurry, the evaluation is not finished yet.

In order to evaluate the equivalence of a top career nowadays and at former times regarding the WHR all time ratings, I tested the following scenario:

- For W. Klitschko I determined the division rank of his opponents at bout time.
- I created a dummy boxer W.Klitschko, with the same sequence of bouts, but starting in Joe Louis' debut year. A shift of 62 years, from 1996 to 1934.
- For these dummy bouts I substituted his real opponents by opponents, equivalently ranked at the dummy bout time.
- Then I run WHR ratings.

The result:
- W. Klitschko dummy peak rating and all-time rating is slightly higher than in his real career (about 10 percent)


I did the same with a shift to Sam Langford's debut year. A shift of 94 year, from 1996 to 1902.

The result:
- W. Klitschko dummy peak rating and all-time rating is about 25 percent of the rating in his real career. But he is still #26 all time Heavyweights, Boxrec recorded competition was lower at that times, I guess.
- While Sam Langford is #96 with 1/3 of dummy Klitschko's points.

Over all, the WHR all time rankings are quite founded based on the achievements against equivalent ranked opponents. Boxrec cannot regard styles, skills, social restrictions and other imponderabilities.

What happened to this version of WHR? Was it superseded by the current version because the prediction rate is higher? I understand Langford being ranked at top 100 or even 200s (Gene Tunney is the only pre-WWII career HW in the top 300), but 1300's?

Sam Langford is #123 at all time Heavyweight.

[computerrank]

...
2nd question. I'm wanting to discuss some ideas on pre-fight factors, but so there's no duplication, can you please list the 16 parameters if possible, or link if this has been answered before?

With current WHR ratings 14 parameters are significant for predicting the winner of a bout ( 2 x 7 parameters)
r = WHR rating of boxer
sd = lowest rating of opponent, who stopped boxer in last 6 bouts
l = lowest rating of opponent, who defeated boxer in last 6 bouts
w = highest rating of opponent, who lost to boxer in last 6 bouts
s = highest rating of opponent, who was stopped by boxer in last 6 bouts
age
years2 = bouts within last 2 years

[Daedalus]

...
2nd question. I'm wanting to discuss some ideas on pre-fight factors, but so there's no duplication, can you please list the 16 parameters if possible, or link if this has been answered before?

With current WHR ratings 14 parameters are significant for predicting the winner of a bout ( 2 x 7 parameters)
r = WHR rating of boxer
sd = lowest rating of opponent, who stopped boxer in last 6 bouts
l = lowest rating of opponent, who defeated boxer in last 6 bouts
w = highest rating of opponent, who lost to boxer in last 6 bouts
s = highest rating of opponent, who was stopped by boxer in last 6 bouts
age
years2 = bouts within last 2 years

Thank you! By the way, I'm a tournament chess player but also a boxing fan. So, on to the ideas, some of which I think are unique and haven't been brought up elsewhere; the actual suggested rating tweaks are marked with (*).
Chess results mostly rely on differences of skill, although given enough time to prepare for the next opponent, I suppose one could research an opening that he thinks the other would be unfamiliar/uncomfortable with, and eke out a trick-victory. But who does that? Boxers! Just kidding.
Boxing success relies on way more things than pure skill and technique. Athleticism/speed, power, stamina/experience to last into later rounds, chin durability, and unfortunately, hometown judging, all come into play.
Unlike team sports, pure defense (let’s say, % opponent landed) is impractical to measure.
* Impose some kind of penalty adjustment based on % knocked down/stopped (the weak chin factor which anyone can exploit). Yes, getting floored as a sign of of poor defense or suspect chin isn’t perfect; some guys are too weight-drained etc.
Most of the other factors can’t be individually quantified, except perhaps power. Power is the one strength that remains the most constant in the boxer’s career, and primarily only affected by going up in weight and facing stronger opponents. I think power would be best expressed not by KO%, but by the strongest opponent a boxer felled in the division; because if you can KO the best, you should be able to take out anyone below. So it would be used not in the rankings, but in scheduled matchups.
* If the opponent is lower-rated than the strongest KO’d, add a bonus (like the 3-5 point home-field advantage in team sports).
Of course, hometown advantage itself has been extensively studied and proven in many sports, including boxing.
* Add a bonus for the local and hope he survives to carry the decision.
Because “drowning in deep waters” is a common strategy by veterans, stamina could be a similar adjustment when facing less-experienced foes.
* Impose a penalty on the fighter not having a minimum number of bouts that have gone for at least 8 rounds.
I read a recent article: https://psycnet.apa.org/record/2018-26204-007. In studying 522 European fights, a mere 3.12 cm (1.22 in) height advantage was found decisive (p<0.05) to winning.
* Add a bonus to the boxer taller by 2".
Hope these help. Since the math is a little beyond me, I'll leave the actual numerical values up to discussion.

[computerrank]

@Daedalus

[Daedalus]

For the old Boxrec r_41 release:
- raw winner prediction ratio based on the old rating only = 0.7482
- winner prediction ratio for r_41 with 16 additional parameters = 0.8345

For WHR:
- raw winner prediction ratio based on the WHR rating only = 0.8308
- winner prediction ratio for r_41 with 16 additional parameters = 0.8402

Your response came out around the same month as the article here: https://www.ringsideboxingnews.com/the- ... in-detail/. So I don't know if the current WHR still reflect the rules published there. Sorry to copy/paste, but that (dubious?) site sometimes goes down.

Ratings Structure
All these ratings evaluate every day all bouts in the database in chronological sequence. A higher rated boxer should be expected to defeat a lower rated boxer with increasing probability by increasing rating difference.

Current Ratings
1. Points are redistributed by every bout. No points come in or get out of the system by this basic rating process
2. But points get out of the system by career end, by point reductions due to inactivity or missing opponent quality
3. So additional points must be fed into the system
4. – 0.01 points, when a boxer wins a bout
5. – 0.1 points, when a boxer defeats an opponent, who already won a bout within 18 months
6. – 1 point, when a boxer defeats an opponent, who already won a bout against a winning opponent within 18 months
7. – when defeating an opponent within the top 15 percent of all active boxers, who already defeated an opponent within the top 15 percent of all active boxers (top 15 percent limit 2019: men = 4.89 points, women = 1.87 points)
8. — 8 points at least
9. — as much points as the defeated opponent had before the bout
10. — 40 points at most
11. — with weight = 1, when the opponent had no loss after his top win, with weight = 1/2 after 1 loss, with weight = 1/4 after 2 losses etc
12. — with weight = 1 within 18 months after the opponents top bout, with weight = 1/2 thereafter and decreasing by a factor of 1/2 per another 18 months
13. – all with weight = cd * v; cd = clear decision factor and v = bout value
14. Every boxer gets a first rating of 0 before his first bout.
15. After every bout, the ratings of the two boxers involved are changed depending on the bout’s official result (KO, TKO, RTD, UD, PTS, NWS, MD, SD, DQ, TD, DRAW).
16. The value of a result varies between v=1 and v=0.
17. The clear decision factor varies between cd=1 and cd=0.
18. The winner cannot lose points for KO, TKO, RTD, DQ, TD and decisions on points with cd=1
19. KO, TKO, RTD are rewarded with full value v=1, cd=1.
20. NWS is rewarded with full value v=1 for 12 rounds boxed and more and a lower value related to the number of rounds boxed. Clear decision factor cd=1.
21. UD, PTS, DQ, TD are rewarded with full value v=1 for 12 rounds boxed and more and a lower value related to the number of rounds boxed, clear decision factor cd=1. If the score cards are available, it may be less.
22. MD, SD are rewarded with full value v=1 for 12 rounds boxed and more and a lower value related to the number of rounds boxed. Clear decision factor is limited to cd=0.5. If the score cards are available cd may be less.
23. DRAW is rewarded with full value v=1 for 12 rounds boxed and more and a lower value related to the number of rounds boxed. Clear decision factor cd=0.
24. If the score cards are available, the value rewarded is in direct proportion to the rounds boxed, with full value v=1 for 12 rounds boxed and more. The clear decision factor is in proportion to rounds boxed and the mean score difference per judge. cd=1 for a mean score difference per judge of 50% of the rounds boxed.
25. All bouts are regarded to have the same weight independent of titles.
26. The winner gets a certain part of the opponent’s points and a certain part of the rating difference to the opponent’s rating.
27. For a DRAW the rating of the higher rated boxer is reduced by some part of the point difference; the rating of the lower rated boxer is enhanced by the same amount of points.
28. The full relative point reward is 33%. It is in direct proportion to the pre-bout rating of the defeated opponent.
29. The rating of a boxer is reduced, if he didn’t box an opponent with a rating of at least 50% of his own rating points within 18 months.
30. The rating of a boxer is reduced by up to 50% in proportion to the difference of 2 times the rating points of his best opponent in this time period minus his own rating
31. The reduction is in proportion to the time the requirement was missed.
32. The rating of a boxer is reduced by 50% for every time period of inactivity of 18 months – and to even less in proportion to a longer time period of inactivity
33. The pre-bout rating of a successfully debuting boxer is set to 25% of his opponents pre-bout rating.
34. The rating points are in relation to a weight division. The rating points are converted with the cube of the upper weight limit ratio of the old and new weight division.
35. The winner is always rated higher than the loser. the minimum margin is: mean pre-bout ratings of both r_m = (r_a + r_b)/2; r_new_winner = r_m + earn_f*v*cd/4; r_new_loser= r_m – earn_f*v*cd/4

Formula
If a boxer with a rating of r_a before the fight defeats a boxer b with a rating of r_b before the fight with result of value v, clear decision factor cd, the new regular ratings r_a_new and r_b_new after a fight are:
• earn = 1/3 * v * (r_b*cd + (r_b-r_a)/(1+2*cd));
• r_a_new = r_a + earn_a
• r_b_new = r_b – earn_b
The winner gets additional points.
Rating reduction caused by missing opponent quality:
career_top_rating = highest career rating adapted to bout division days = days between bout with best opponent before and bout with best opponent after reduction
• r_red-career_top_rating/10 = (r_old-career_top_rating/10) * (1 – 0.5*(1 – 2*best_opp/r_old))**(days/(365.24*1.5))

Examples

1. Boxer a UD 6 boxer b, scores 59:55 58:56 58:56, a has 10 points, b has 5 points, opponent with a prior win, but not within the top 15 percent of all active boxers
• A 6 rounder is rewarded with value 6/12, v=0.5
• UD is rewarded with cd=1 at maximum
• mean score difference per judge is (4+2+2)/3 = 2.667, which is rewarded in direct proportion to half the rounds boxed with cd= 2.667/3 = 0.89 at maximum
• so cd=0.89
• earn= 1/3 * 0.5 * (5*0.89 + (5-10)/(1+2*0.89)) = 0.44
• r_a_new = 10 + 0.44 = 10.44
• r_b_new = 5 – 0.44 = 4.56
• Winner additionally gets 0.01 points + 0.1 points for opponent with win and weight = cd*v = 0.89*0.5= 0.445
• additional points = 0.11*0.445 = 0.05 points
• r_a_new = 10.44 + 0.05 = 10.49

2. Boxer a SD 8 boxer b, scores 77:75 77:75 75:77, a has 100 points, b has 50 points, opponent within the top 15 percent of all active boxers and with an own win against such a top 15 percent boxer 23 months ago and 1 following loss
• A 8 rounder is rewarded with 8/12, v=0.667
• SD is rewarded with cd=0.5 at maximum
• mean score difference per judge is (2+2-2)/3 = 0.667, which is rewarded in direct proportion to half the rounds boxed 0.667/4 at maximum
• so cd=0.167
• earn= 1/3 * 0.667 * (50*0.167 + (50-100)/(1+2*0.167)) = -6.48
• r_a_new = 100 – 6.48 = 93.52
• r_b_new = 50 + 6.48 = 56.48
• Winner additionally gets 0.01 points + 0.1 points for opponent with win + 1 point for opponent with a win against a winner
• + 40 points for opponent with top 15 percent win with weight=1/2 for 1 following loss and weight=0.41 for top win before 23 months
• and overall weight = cd*v = 0.5*0.67= 0.333
• additional points = (1.11 + 40*0.5*0.41)*0.333 = 3.10 points
• r_a_new = 93.52 + 3.10 = 96.62

All Time Ratings
The All Time Rating for a boxer is the sum of annual rank points of best defeated opponent at time of bout and the square root of his career top rating:
1. best opponent rank points = 200/ (best opponent rank – 1) for annual rating (if rank is 1 use rank 1)
2. best opponent p4p rank points = 200 / (best opponent p4p rank – 1) for annual rating (if rank is 1 use rank 1)
3. the ranks are interpolated in relation to the ratings of boxer ranked before him and the boxer behind him
4. the best opponent rank points are limited to the lowest rank points of the opponent, he lost to in this year – pair by pair
5. the value of 200 annual points is reduced, if the annual rating of #30 in the division is less than 76.3 for men, or less than 3.86 for women
6. the value of 200 annual p4p points is reduced, if the annual rating of #50 p4p is less than 279 for men, or less than 12.2 for women

I know I asked about part of it in a previous post, but just to clarify, which, if any, of the Ratings Structure rules are NOT in effect (for example: CR 4-10; Ex 2; ATR 5-6)?

[computerrank]

I know I asked about part of it in a previous post, but just to clarify, which, if any, of the Ratings Structure rules are NOT in effect (for example: CR 4-10; Ex 2; ATR 5-6)?

Only the basic rules for the bout value are still in place - rules 15 to 25. All time rating is the career top rating of the boxer.
WHR ratings for each bout regard all bout results of all boxers in history before and after that bout. So the WHR ratings represent the maximum likelihood for all results in history, measured by the minimum sum of deviations of the real results from the predicted results.

[Offthehook]

It is explained on the page, you linked:
You will find the boxers with 3 and more top ranks of any weight division.

why does jono carroll jump up 33 places when he lost a UD to maxi hughes. Its not like he even won on one card.

The ratings are really poor in my opinion

Jono Caroll was #13 before the bout. He went down to #33. The ranking move shown on the Boxrec page is erroneous.

nope he went down to number 46 straight after the bout, and maxi hughes was ranked 217 before their bout, so carroll should of been a lot lower than that based on who he lost to at the time. As i say theres a lot of people ranked incorrectly, with some fighters given a huge amount of credit for some results, and others not so. Perhaps just rank the top 100 in each weight class properly and by inspection, rather than use a formula, that seems to be inaccurate, to rank everyone.

[computerrank]

why does jono carroll jump up 33 places when he lost a UD to maxi hughes. Its not like he even won on one card.

The ratings are really poor in my opinion

Jono Caroll was #13 before the bout. He went down to #33. The ranking move shown on the Boxrec page is erroneous.

nope he went down to number 46 straight after the bout, and maxi hughes was ranked 217 before their bout, so carroll should of been a lot lower than that based on who he lost to at the time. As i say theres a lot of people ranked incorrectly, with some fighters given a huge amount of credit for some results, and others not so. Perhaps just rank the top 100 in each weight class properly and by inspection, rather than use a formula, that seems to be inaccurate, to rank everyone.

I noticed Maxi Hughes had a bout after his bout against Caroll.
So without these 2 bouts: Jono Caroll would be #13 and Maxi Hughes would be #142. With the Caroll vs Hughes bout Caroll would be #62 and Hughes would be #60. And with Hughes' last win against Viktor Kotochigov Hughes is #26 and Caroll is back to #33.

[Daedalus]

I know I asked about part of it in a previous post, but just to clarify, which, if any, of the Ratings Structure rules are NOT in effect (for example: CR 4-10; Ex 2; ATR 5-6)?

Only the basic rules for the bout value are still in place - rules 15 to 25. All time rating is the career top rating of the boxer.
WHR ratings for each bout regard all bout results of all boxers in history before and after that bout. So the WHR ratings represent the maximum likelihood for all results in history, measured by the minimum sum of deviations of the real results from the predicted results.

Thank you for explaining! Not sure I got the last paragraph.

[Daedalus]

I understand WHR is more stable with old-timers so I looked at the #1 all time Ezzard Charles whose 1940-59 career I thought wouldn't show many wild swings. I was wrong.

Here's the dates/opponent (their post-fight rating)/change in Charles' rating (comments):

9/23/46 W O. Smith (157.6) 1394 pts post-fight

11/12/46 W Bivins (151.2) lost 822 pts (Why?)

2/17/47 KO O. Smith (145.9) gained 1024 pts (Why? Smith went 3-0-1 in the interim including a draw with Archie Moore but had 57.61 the previous fight, losing points despite a stellar performance)

3/10/47 KO Bivins (364.2) gained 23 pts (Bivins went 4-0 in the interim but had 143.8 pts the previous fight. Why did Bivins gain 220 points in an expected loss?)

4/14/47 W Sarlin (12.7) lost 987 pts (Why? Was fighting a Sarlin much worse than a 100+ pre-fight Bivins?)

5/5/47 W Moore (472.3) gained 1002 pts (Why? Charles last beat Archie [403.3] on 5/20/46. Archie went 4-0-2 in the interim. He drew successively with Smith and Chase [40.91 on 11/6/46])

7/14/47 KO Fitzpatrick (14.83) lost 996 pts (Why?)

7/25/47 L Ray (117.3) gained 1 pt for losing to a nobody (Why? And why didn’t Ray gain any points [119.1 pts on 7/14, 11 days earlier]?)

9/16/47 W Matisi (8.844) gained 19.7 pts

9/29/47 KO Marshall (82.01) gained 1055 pts (Why? Charles last beat Lloyd [108.6] on 7/29/46. Marshall went 2-0 in the interim but had 37.91 pts after the previous fight, so why did Marshall double his points by losing?)

10/16/47 KO Al Smith (0.799) lost 1044 pts (Why? Bivins went 4-2. Moore went 4-0 and beat Bivins but lost 302 pts!)

The result after Matisi is the only thing I get (gain minimal points by beating a nobody). Everything else is incomprehensible. I can't see any pattern between Charles' post-fight rating and winning, losing, opponent's rating, opponent's record before/after their meeting or the timing of the fluctuation. Can someone explain?

[computerrank]

I understand WHR is more stable with old-timers so I looked at the #1 all time Ezzard Charles whose 1940-59 career I thought wouldn't show many wild swings. I was wrong.
...
The result after Matisi is the only thing I get (gain minimal points by beating a nobody). Everything else is incomprehensible. I can't see any pattern between Charles' post-fight rating and winning, losing, opponent's rating, opponent's record before/after their meeting or the timing of the fluctuation. Can someone explain?

I guess you didn't follow the posts in this thread since autumn 2019, when I started to shift to Whole-Historing Rating. And I guess you didn't read the wiki entry for BoxRec ratings, with a link to the Whole-History Rating paper of Remi Coulom:
https://boxrec.com/media/index.php/BoxR ... escription

Main point is - rating points depend on the weight division of the bout.

[Daedalus]

I understand WHR is more stable with old-timers so I looked at the #1 all time Ezzard Charles whose 1940-59 career I thought wouldn't show many wild swings. I was wrong.
...
The result after Matisi is the only thing I get (gain minimal points by beating a nobody). Everything else is incomprehensible. I can't see any pattern between Charles' post-fight rating and winning, losing, opponent's rating, opponent's record before/after their meeting or the timing of the fluctuation. Can someone explain?

I guess you didn't follow the posts in this thread since autumn 2019, when I started to shift to Whole-Historing Rating. And I guess you didn't read the wiki entry for BoxRec ratings, with a link to the Whole-History Rating paper of Remi Coulom:
https://boxrec.com/media/index.php/BoxR ... escription

Main point is - rating points depend on the weight division of the bout.

What do you mean? All those fights were contested at one weight division, light heavyweight.

[Daedalus]

I understand WHR is more stable with old-timers so I looked at the #1 all time Ezzard Charles whose 1940-59 career I thought wouldn't show many wild swings. I was wrong.
...
The result after Matisi is the only thing I get (gain minimal points by beating a nobody). Everything else is incomprehensible. I can't see any pattern between Charles' post-fight rating and winning, losing, opponent's rating, opponent's record before/after their meeting or the timing of the fluctuation. Can someone explain?

...
Main point is - rating points depend on the weight division of the bout.

What do you mean? All those fights were contested at one weight division, light heavyweight. Charles and his opponents were not cruiserweights back then, even going by BoxRec annual ratings.