Ratings - please read before commenting - Archived

Lennox · Post by **Lennox** » 02 Apr 2007, 11:35

jujigatame wrote:Well, let's face it, Hill just lost to a guy who'd been retired for over 10 years.

I never thought anyone would agree with computerranks' Maske/Hills rating. Hill should be ranked above WBA champ Brudov surely, and Maske above Hill. 10 year lay off is irrelvant. These ratings are really poor. Virgil Hill was a top 10 fighter at Cruiserweight pre this bout, he was the WBA champ for gods sake!!!!

JCS · Post by **JCS** » 02 Apr 2007, 11:38

Even though he's WBA Champ.. right now Hill seems like a #15-18 guy right now to me.. Before the bout, I thought he was around 10.

I think we have to use this most recent result to better place Maske's re-insertion point (even though he re-retired) to more accurately place Virgil Hill. This may seem technically incorrect, but I think situations like this are worthy of such a process.

jujigatame · Post by **jujigatame** » 02 Apr 2007, 12:47

I don't have a problem with Hill below Brudov. That fight was about 15 months ago, and since then Brudov is 3-0 with a win over a solid contender (Pineda) while Hill was inactive for the entire time and then lost to a 43 year old retiree.

I agree that 70-something may be a little low for Hill but I don't think he's top 20 like most of you are implying.

computerrank · Post by **computerrank** » 02 Apr 2007, 13:26

... I will run a test with:

- current inactivity drop of 50% per 18 months reduced to 25% per 18 month - as for missing quality opponents

I will publish the results here - may take me 2 weeks, as I will be on the road

JCS · Post by **JCS** » 02 Apr 2007, 13:30

jujigatame wrote:I don't have a problem with Hill below Brudov. That fight was about 15 months ago, and since then Brudov is 3-0 with a win over a solid contender (Pineda) while Hill was inactive for the entire time and then lost to a 43 year old retiree.

I agree that 70-something may be a little low for Hill but I don't think he's top 20 like most of you are implying.

I could see as low as #30 for Hill, but no lower.

JCS · Post by **JCS** » 02 Apr 2007, 13:40

computerrank wrote:... I will run a test with:

- current inactivity drop of 50% per 18 months reduced to 25% per 18 month - as for missing quality opponents

I will publish the results here - may take me 2 weeks, as I will be on the road

I'll give it a shot, but I don't have access to your new database to change the active statuses of Maske/Hill... and we don't have their fight in the bouts file anyway. Have to use guys like Spadafora, Holyfield, Gainer, etc., to gauge it. I think Gainer at #108 is a bit too low. He should DEFINITELY be Top 50.

Changing the 0.5 in the fade calculations to 0.75 should do the trick I believe..

Any thoughts on Gainer guys? Maybe its just me, but #108 seems low. IBO has him at #8, which I think is too high. FightNews has him at #12. I was thinking around #25.

JCS · Post by **JCS** » 02 Apr 2007, 14:16

Reducing the penalties from 50% to 25% takes..

Gainer from #108 to #56
Holyfield from #58 to #52
Spadafora from #75 to #34
De La Hoya from #2 to #1
Juan Carlos Gomez from #23 to #17

Judging by these results, I'm guessing Hill, with his bouts of inactivity... and Maske, who had a ton of inactivity, should bump Virgil up a good 20-30 spots.

I like the looks of 25% vs 50%.

jujigatame · Post by **jujigatame** » 02 Apr 2007, 14:31

I don't think it's really a big improvement. DLH at #1 isn't really desirable considering his big stretches of inactivity. Where was Hill at before the loss to Maske?

JCS · Post by **JCS** » 02 Apr 2007, 14:32

jujigatame wrote:I don't think it's really a big improvement. DLH at #1 isn't really desirable considering his big stretches of inactivity. Where was Hill at before the loss to Maske?

Hill was probably too low to begin with, unfortunately I do not know because he was classified as inactive.

Can you think of any other prominent boxers who has recent bouts of inactivity... only to come back and be placed REALLY low by the ratings?

I checked on Herbie Hide, but he had a fight between my data point and the live ratings.

BTW, every little bit helps. I may try to set it to 33%.. 25% might be a tad too much... then again, in some cases I would say its great.

JCS · Post by **JCS** » 02 Apr 2007, 15:01

Code: Select all

50% = Live | 33% & 25% Data from 3/22 | IBO from 3/1? |

PENALTY              50%    33%    25%  |  IBO
Derrick Gainer      #106    #48    #56  |   #8
Evander Holyfield    #58    #55    #52  |  #31
Oscar De La Hoya      #2     #2     #1  |   #2
Juan Carlos Gomez    #23    #17    #17  |  #25
Paul Spadafora       #75    #51    #34  |  #48
Harry Simon          #85    #38    #22  |   NA (Came back in March)
Raul Marquez         #43    #39    #39  |  #42
Joe Mesi             #68    #59    #38  |  #48
Tommy Morrison      #435   #353   #302  |  #91
Michael Moorer       #57    #52    #41  |  #36

Interesting how Gainer was higher at 33%...

I like 33% the best.. Martin, what do you think?

JCS · Post by **JCS** » 03 Apr 2007, 10:15

Just for curiosity sake.. I'm gonna run this with the 3/22 data.

- 33% penalties (inactivity + opposition quality) vs. 50% for live

- 10% home advantage for those w/ more than 5 bouts in country (vs 100% and 2 bouts in old system / this is turned off on live)

- TD and DQ factor raised from 0.5 to 0.75

-----

This didn't change things up much... I think now, the ratings are fairly solid... but we need to focus on isolated scenarios and work on improving them. I think thats the only way they get better.

The scenario at hand for me anyway.. is reduction of the penalties. 33% is a marked improvement from 50% IMO.

computerrank · Post by **computerrank** » 03 Apr 2007, 12:47

... I think, a reduction of the inactivity drop from 50% to 33% is discussable ... why not, if it really looks better ...

JCS · Post by **JCS** » 03 Apr 2007, 12:50

computerrank wrote:... I think, a reduction of the inactivity drop from 50% to 33% is discussable ... why not, if it really looks better ...

I'm all for it. It improves on what I think is a problem... even if its only a minor problem, its still a problem nonetheless.

Will you be doing more testing with it?

JCS · Post by **JCS** » 04 Apr 2007, 16:16

Well.. after we get the inactivity thing straightened out, I think we need to finally look at some kind of protection. Case in point, Lyakhovich, who has dropped completely off the Top 10 (and Top 25 for that matter) for losing to Briggs... far below Brewster (who has since done nothing), who would be in the Top 10 if he was still active status.

This would be complicated, but I think its worth trying...

I'll use the scenario of Lyakhovich-Briggs-Brewster

First off..I think we use the last 6 fights in this whole protection scheme.. at least as a start to things.

In a case like Lyakhovich who just lost to Briggs, we record the IDs of the fighters he has beaten in his last 6 bouts (Walton-Guerrero-Maxime-Guinn-Brewster). We refer to their current ratings after Briggs-Lyakhovich, or their ratings at the time Lyakhovich fought them, use whichever rating is greater, and say that he cannot drop below any of the ratings in the set unless its by means of inactivity, or he loses to a fighter whose rating was lower than those he has beaten in his last 6 bouts. In this case, Lyakhovich would not have fallen past Brewster.

If one of those boxers who Sergei has beaten in the last 6 have since risen their rating above Lyakhovich's, then so be it, as they earned it by defeating quality opponents. He will then be prevented from falling below the next best opponent he has beaten in the last 6. Then again, this is also the flipside of the coin we can look at later.

In addition to the above, there probably needs to be a protection to not be able to advance a boxer past someone who has recently beat them. All variables would be similar to the previous protection.
----

With the standard equations, we do a great job of assuming... but with protection schemes, I think we can move into a different level of quality.

jujigatame · Post by **jujigatame** » 04 Apr 2007, 17:39

How did that Brewster/Lyakovich/Briggs situation develop, anyway? I thought maybe Lyakhovich dropped down so low because Briggs was rated poorly before they fought, but by the same token, wouldn't Brewster have dropped down very low since Lyakhovich was rated poorly before he and Brewster fought?

JCS · Post by **JCS** » 04 Apr 2007, 20:48

jujigatame wrote:How did that Brewster/Lyakovich/Briggs situation develop, anyway? I thought maybe Lyakhovich dropped down so low because Briggs was rated poorly before they fought, but by the same token, wouldn't Brewster have dropped down very low since Lyakhovich was rated poorly before he and Brewster fought?

At the beginning.. I believe Lyakhovich was probably around #20, Brewster at about #4, Briggs at about #30.

Lyakhovich beat Brewster by a fairly wide decision... Since it wasnt a blowout edcision, Brewster only dropped to about #8, and Lyakhovich was thrusted above Brewster. That is the one protection we have now... the winner MUST be placed above the loser by at least 1 pt. So now you have Lyakhovich #7, Brewster #8, Briggs #30.

Well Briggs, KO'd Lyakhovich, which caused him to lose a lot of points. In turn, it caused Briggs to gain a lot of points. There is no protection to keep Lyakhovich ahead of Brewster like The Ring and FightNews have it, so Lyakhovich did a freefall down the ratings and Briggs went up to #9 or 10.. wherever he is now.

In reality you'd likely have Briggs, before Lyakhovich, before Brewster, all in a tight group.

jujigatame · Post by **jujigatame** » 04 Apr 2007, 23:28

While it's not exactly fitting to the "traditional" rankings style we're going for, it's interesting to note that these rankings are essentially calling Lyakhovich a one-hit wonder, which is a sentiment a lot of people are starting to share.

TerribleTim2 · Post by **TerribleTim2** » 08 Apr 2007, 23:56

Why did Travis Walkers position not change at all after beating another undefeated heavy weight? I surprise that he didn't improve by at least 1 or 2 spots

JCS · Post by **JCS** » 09 Apr 2007, 00:02

TerribleTim2 wrote:Why did Travis Walkers position not change at all after beating another undefeated heavy weight? I surprise that he didn't improve by at least 1 or 2 spots

It was a close split decision. Since Walker was already 10 or so spots above Garcia.. he performed to the level which he was expected to, not above or below, hence his rating's stability.

computerrank · Post by **computerrank** » 09 Apr 2007, 11:31

... I just tested the inactivity drop reduced from 50% to 33% ...

Now I find:

- Tua from #45 to #26
- Gomez from #23 to #16

I will keep the 50% - as elsewise there are other odd impacts ...

JCS · Post by **JCS** » 09 Apr 2007, 11:34

computerrank wrote:... I just tested the inactivity drop reduced from 50% to 33% ...

Now I find:

- Tua from #45 to #26
- Gomez from #23 to #16

I will keep the 50% - as elsewise there are other odd impacts ...

I don't think that is too crazy.. How about 40%? I still think 50 is a bit too much.

At 40%, we have Tua at #31, and Gomez at #21

computerrank · Post by **computerrank** » 13 Apr 2007, 03:44

Cobwebcat wrote:How do the editors decide which weight category to place a fighter in the all-time list?

I would have thought whilst he was still fighting he would be in the category he last fought but at the end of his career it would be in the category he fought most of his fights in.

Not a criticism, just a question as I can't see a pattern so just assume it's random?

- while active, they are assigned to the division, they last fought

- after retired, the process is not really defined - it's up to the editors, where to assign them by opinion
- the accepted guideline seems to be, to assign them to the division with most achievements - whatever this may be
- but often, the boxers are left in the division, they were last assigned to

- idea:
- use the daily rating program to assign retired (inactive) boxers to the division, where they achieved their best division specific all time rating (average of divison specific best performances)

- but for this you have to track all divisions for every boxer for all of his bouts

JCS · Post by **JCS** » 18 Apr 2007, 12:17

Cobwebcat wrote:Computerrank:

I noticed that Hearns is now appearing as a Welterweight in the all-time stats.

What procedure changed?

I asked the editors to change him.

dynamicop · Post by **dynamicop** » 26 Apr 2007, 14:00

YOur site states that boxers all start off with a rating score of 0. It seems that given your updating system, all boxers would retain a rating score of 0 regardless of their results. So what am I missing that they actually have positive scores?

Also, who is best to contact about a large scale project I want to do with boxrec, that will be very exciting to this site.

Thanks,

dyanmicop

JCS · Post by **JCS** » 26 Apr 2007, 14:08

dynamicop wrote:YOur site states that boxers all start off with a rating score of 0. It seems that given your updating system, all boxers would retain a rating score of 0 regardless of their results. So what am I missing that they actually have positive scores?

Also, who is best to contact about a large scale project I want to do with boxrec, that will be very exciting to this site.

Thanks,

dyanmicop

I believe boxers enter an initial launch process that differs from the typical rating structure.