Ratings - please read before commenting

[computerrank]

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

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

I will not propose this version at the moment. Reason: As a side effect the new version builds a wide spread ladder for the present Heavyweights for example. This spread effect exceeds the normal rating relation between weight divisions and the normal relation between ratings from different time periods.

[SportsRatings]

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

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

I will not propose this version at the moment. Reason: As a side effect the new version builds a wide spread ladder for the present Heavyweights for example. This spread effect exceeds the normal rating relation between weight divisions and the normal relation between ratings from different time periods.

Yikes! Francis Ngannou caused a ripple in space-time!

[computerrank]

... Yikes! Francis Ngannou caused a ripple in space-time!

[computerrank]

My last approach to handle the Fury - Ngannou ranking problem resulted in building a pronouced ranking ladder system.

Now I found a more sophisticated approach by keeping a standard number of 3 prior bouts for all boxers with at least 6 career bouts. And working with a reduced number o prior bouts only for boxers with less career bouts. 0.5 prior bouts for a boxer with 1 career bout, 1 prior bout for a boxer with 2 career bouts ... 3 prior bouts for boxers with 6 or more career bouts.

This even improves the resulting winner prediction ratio from 85.67 to 85.84 percent!

And it doesn't really touch the ratings of boxers with 6 and more career bouts. Only boxers with less career bouts are more tied now to the rankings of their opponents than to virtual prior opponents. Especially boxers after their first bout are rated closer to their opponent.

Ngannou would be rated at #28 after his SD loss to Fury.

I will propose that solution for the live system now.

[SportsRatings]

My last approach to handle the Fury - Ngannou ranking problem resulted in building a pronouced ranking ladder system.

Now I found a more sophisticated approach by keeping a standard number of 3 prior bouts for all boxers with at least 6 career bouts. And working with a reduced number o prior bouts only for boxers with less career bouts. 0.5 prior bouts for a boxer with 1 career bout, 1 prior bout for a boxer with 2 career bouts ... 3 prior bouts for boxers with 6 or more career bouts.

This even improves the resulting winner prediction ratio from 85.67 to 85.84 percent!

And it doesn't really touch the ratings of boxers with 6 and more career bouts. Only boxers with less career bouts are more tied now to the rankings of their opponents than to virtual prior opponents. Especially boxers after their first bout are rated closer to their opponent.

Ngannou would be rated at #28 after his SD loss to Fury.

I will propose that solution for the live system now.

[computerrank]

My last approach to handle the Fury - Ngannou ranking problem resulted in building a pronouced ranking ladder system.

Now I found a more sophisticated approach by keeping a standard number of 3 prior bouts for all boxers with at least 6 career bouts. And working with a reduced number o prior bouts only for boxers with less career bouts. 0.5 prior bouts for a boxer with 1 career bout, 1 prior bout for a boxer with 2 career bouts ... 3 prior bouts for boxers with 6 or more career bouts.

This even improves the resulting winner prediction ratio from 85.67 to 85.84 percent!

And it doesn't really touch the ratings of boxers with 6 and more career bouts. Only boxers with less career bouts are more tied now to the rankings of their opponents than to virtual prior opponents. Especially boxers after their first bout are rated closer to their opponent.

Ngannou would be rated at #28 after his SD loss to Fury.

I will propose that solution for the live system now.

WHR 1.2.10 is live now

[JCS]

My last approach to handle the Fury - Ngannou ranking problem resulted in building a pronouced ranking ladder system.

Now I found a more sophisticated approach by keeping a standard number of 3 prior bouts for all boxers with at least 6 career bouts. And working with a reduced number o prior bouts only for boxers with less career bouts. 0.5 prior bouts for a boxer with 1 career bout, 1 prior bout for a boxer with 2 career bouts ... 3 prior bouts for boxers with 6 or more career bouts.

This even improves the resulting winner prediction ratio from 85.67 to 85.84 percent!

And it doesn't really touch the ratings of boxers with 6 and more career bouts. Only boxers with less career bouts are more tied now to the rankings of their opponents than to virtual prior opponents. Especially boxers after their first bout are rated closer to their opponent.

Ngannou would be rated at #28 after his SD loss to Fury.

I will propose that solution for the live system now.

WHR 1.2.10 is live now

Assuming Ngannou was the most notable jump. Any others close?

[computerrank]

WHR 1.2.10 is live now

Assuming Ngannou was the most notable jump. Any others close?

We will see tomorrow morning after the first complete daily update.

[pugilisticspecialist]

My last approach to handle the Fury - Ngannou ranking problem resulted in building a pronouced ranking ladder system.

Now I found a more sophisticated approach by keeping a standard number of 3 prior bouts for all boxers with at least 6 career bouts. And working with a reduced number o prior bouts only for boxers with less career bouts. 0.5 prior bouts for a boxer with 1 career bout, 1 prior bout for a boxer with 2 career bouts ... 3 prior bouts for boxers with 6 or more career bouts.

This even improves the resulting winner prediction ratio from 85.67 to 85.84 percent!

And it doesn't really touch the ratings of boxers with 6 and more career bouts. Only boxers with less career bouts are more tied now to the rankings of their opponents than to virtual prior opponents. Especially boxers after their first bout are rated closer to their opponent.

Ngannou would be rated at #28 after his SD loss to Fury.

I will propose that solution for the live system now.

WHR 1.2.10 is live now

What's the prediction rate?

[omalley]

Is the rating of a boxer reduced for prolonged periods of inactivity? For example, Oleksandr Gvozdyk https://boxrec.com/en/box-pro/679709, period of inactivity: almost 40 months.

[computerrank]

Is the rating of a boxer reduced for prolonged periods of inactivity? For example, Oleksandr Gvozdyk https://boxrec.com/en/box-pro/679709, period of inactivity: almost 40 months.

Not necessarilly, it depends on the results after that.

[omalley]

Is the rating of a boxer reduced for prolonged periods of inactivity? For example, Oleksandr Gvozdyk https://boxrec.com/en/box-pro/679709, period of inactivity: almost 40 months.

Not necessarilly, it depends on the results after that.

Would you please provide some details? Enormous inactivity should be "punished", shouldn't it? Another example: Jermall Charlo https://boxrec.com/en/box-pro/471634.
What if Mayweather and Pacquiao become "unretired" and somehow some commission sanctions their bout as official?

[computerrank]

Is the rating of a boxer reduced for prolonged periods of inactivity? For example, Oleksandr Gvozdyk https://boxrec.com/en/box-pro/679709, period of inactivity: almost 40 months.

Not necessarilly, it depends on the results after that.

Would you please provide some details? Enormous inactivity should be "punished", shouldn't it? Another example: Jermall Charlo https://boxrec.com/en/box-pro/471634.
What if Mayweather and Pacquiao become "unretired" and somehow some commission sanctions their bout as official?

The Whole-History Ratings devalue older results after a while. And they adapt quicker to the value of later results. So there is no need of special punishment.

Without his last 2 results Gvozdyk would only be #15.

I don't see, why Charlo should be punished. He had good wins recently too.

Mayweather would be #2 after stopping Pacquiao, and Pacquiao #12. Pacquiao would be #2 after stopping Mayweather and Maywether #13.

[omalley]

Not necessarilly, it depends on the results after that.

Would you please provide some details? Enormous inactivity should be "punished", shouldn't it? Another example: Jermall Charlo https://boxrec.com/en/box-pro/471634.
What if Mayweather and Pacquiao become "unretired" and somehow some commission sanctions their bout as official?

The Whole-History Ratings devalue older results after a while. And they adapt quicker to the value of later results. So there is no need of special punishment.

Without his last 2 results Gvozdyk would only be #15.

I don't see, why Charlo should be punished. He had good wins recently too.

Mayweather would be #2 after stopping Pacquiao, and Pacquiao #12. Pacquiao would be #2 after stopping Mayweather and Maywether #13.

With all due respect i believe that there IS need of special rating reduction for 2-3-4... years of inactivity.
Wins over Ricards Bolotniks (661423) & Isaac Rodrigues (379126) enabled Gvozdyk to jump from #15 to #3? Then why they are so low in ratings themselves now (#63 & #138)? Rather strange, isn't it?

Jermall Charlo inactivity: 29 months. After that only 1 win at SMW, not in MW, against Jose Benavidez Jr, who is not even a MW fighter. Is it good win? I doubt it.

So, if grandfather Mayweather in 2024 will stop Pacquiao in some kind of uh... umm... official bout, he would be #2 acting WW only behind Crawford? Unreal...

[Cap]

Just a suggestion. Why not have a column to the right of country of origin stating the last time individual boxers fought? Or how many fights in the previous calendar year?

[SeanBrennan]

Just a suggestion. Why not have a column to the right of country of origin stating the last time individual boxers fought? Or how many fights in the previous calendar year?

this is a simple but excellent suggestion.

[computerrank]

Just a suggestion. Why not have a column to the right of country of origin stating the last time individual boxers fought? Or how many fights in the previous calendar year?

This ratings thread is dedicated to ratings questions only.

Please post your suggestion here viewtopic.php?t=250810

[omalley]

I'm sorry, what about my questions? Please, please reveal the secret, what positions had Oleksandr Gvozdyk (679709) and Ricards Bolotniks (661423) before and after their bout.

You said Gvozdyk was #15 after bout against Josue Obando (556077), unexpectedly high, it's very hard to justify, taking into account that:
- before the bout Gvozdyk had 40 (!!!) months of inactivity
- Gvozdyk-Obando bout took place 2023-02-11, last Gvozdyk's win before that was 2019-03-30 (!!!)
- Obando obviously had very low rating and Gvozdyk-Obando bout was only 6 rounds.

Gvozdyk jumped very high up after win over Bolotniks. Why did Bolotniks rating сollapse so much, down to #63 (now)?

[omalley]

To clarify, I have nothing personal against Oleksandr Gvozdyk. This is just a prime example.

Fury's rating is defined by his best win against Wilder. But it is 24 months ago now. It once was worth 40.5 points, but now only 16.6 points.

So, the value has decreased by more than two times, 40.5 -> 16.6. What if the best win of a Boxer was 36 months ago? 48 months? 60?

[computerrank]

To clarify, I have nothing personal against Oleksandr Gvozdyk. This is just a prime example.

Fury's rating is defined by his best win against Wilder. But it is 24 months ago now. It once was worth 40.5 points, but now only 16.6 points.

So, the value has decreased by more than two times, 40.5 -> 16.6. What if the best win of a Boxer was 36 months ago? 48 months? 60?

Read the paragraph 'Current Ratings' in https://boxrec.com/wiki/index.php/BoxRe ... escription

[computerrank]

I'm sorry, what about my questions? Please, please reveal the secret, what positions had Oleksandr Gvozdyk (679709) and Ricards Bolotniks (661423) before and after their bout.

You said Gvozdyk was #15 after bout against Josue Obando (556077), unexpectedly high, it's very hard to justify, taking into account that:
- before the bout Gvozdyk had 40 (!!!) months of inactivity
- Gvozdyk-Obando bout took place 2023-02-11, last Gvozdyk's win before that was 2019-03-30 (!!!)
- Obando obviously had very low rating and Gvozdyk-Obando bout was only 6 rounds.

Gvozdyk jumped very high up after win over Bolotniks. Why did Bolotniks rating сollapse so much, down to #63 (now)?

He was at 2.9 points after his loss against Beeterbiev, which would be #15 today.
He was at 5.5 points after his win against Bolotniks, which would be #4 today.
He is at 6.1 points after his win against Rodrigues, which is #3 today.
Bolotniks was at 0.578 after his win against Sep, which would be #58 today.
Bolotniks was at 0.475 after his loss against Gvozdyk, which would be #66 today.
Bolotniks is at .486 after his win against Bezvoda, which is #63 today.

[jujigatame]

What's up with the ranking of Fadhili Majiha? Just happened to notice him since it's rare for a 14-loss fighter to be in the top 5 of a relatively deep division. There don't appear to be any big names on his record, either.

[computerrank]

What's up with the ranking of Fadhili Majiha? Just happened to notice him since it's rare for a 14-loss fighter to be in the top 5 of a relatively deep division. There don't appear to be any big names on his record, either.

No big names, but a solid record the last 4 years.

[jujigatame]

Oh I thought maybe there was some win on his record that was more valuable than it appeared. The guys ranked below him would seem to have bigger wins, like Astrolabio over Rigo, Nishida over Higo, etc.

[JCS]

That does seem awfully peculiar.... expected a bug to be responsible for that one... Tanzania not exactly a hot spot for boxing.