boxrec top 50 ... my new project

[mh2365]

I started off just wanting to see what the top 50 in boxrec's world rankings were ranked in the game and decided to mess with all the top 50 lists and find pics of fighters not yet rated ... I will post pics I can find in chris's thread (as you noticed in the last two days).

I will also post my list of the world top 50 and what they are rated in my database (I say my db because I may have changed some of the ratings). When you look at the list the # in () to the far right is their rating. if they are rated in another weight class only that is noted. Also I put a little footnote in there on wether or not I think the rating is right.

Now just so everyone knows I don't rate fighters and have the utmost respect for those who do, so it's just my opinion and no reflection on the rating abilities who rated the fighters (except maybe for the English Homer who rated Skelton a 10 ) so please do not take offense.

Also I will list in this thread fighters I couldn't find pics for but merit a look as far as being rated.

[mh2365]

25 new photos without files found
7 missing photos found
0 fighters without photos who merit rating

[mh2365]

30 new photos without files
4 missing photos
7 fighters w/o photos that merit rating

ola afolabi 91647
nicolay popov 92872
frederic serrat 15492
charles davis 28953
dan harvison 197434
daniel bispo 72844
claudio rasco 245659

All of the guys listed above are world top 50 - no pics found


more weights to follow in the next few days

[mh2365]

30 new photos without files
3 missing photos
6 fighters w/o photos that merit rating

denis lebedev 34020
berry butler 6993
antonio brancalion 9053
istvan varga 252155
andrew lowe 44027
fred moore 19627 - not world top 50

all the guys above except where noted are world top 50 no pics found

[CONN CHRIS]

I wonder how advisable it is for us to rate many of these folks Mike, it sounds like Mr. Trunzo is going to rate them. He certainly would do a better job.

I also wonder how Boxrec rates these fighters, number 26 in the world for this guy seems a bit inflated to me.

OLA AFOLABI CW (2002 - ) Overall Rating = 4

Hometown - West Hills, California, USA
Born - London, England

9(3)-1-3

Afolabi has beaten Willie Chapman and James Walton and drawn with Jonathan Young. He has not been stopped to-date in 13 pro bouts.

[mh2365]

I thought the same thing .. so what I will do now is still do the list but I'll note on pics when they are a world top 50 .... I'm going through almost all of the top 50 lists though, It'll probably knock out my list of fighters who should be rated without pics, because I really was only doing world top 50 guys on that.

Okay edited and noted above

[Dave1877]

I can't believe they have Skelton so high, he should be no higher than 25-30 in my opinion, if that.

Also why do you say Vitali is overrated?

[bear]

If you think those ratings are funny I suggest taking a look at their all-time heavyweights. Joe Louis is 15th and Ingo Johanson 7th. Yowzah.

Bear

Originally Posted by Dave1877

I can't believe they have Skelton so high, he should be no higher than 25-30 in my opinion, if that.

Also why do you say Vitali is overrated?

[CONN CHRIS]

Originally Posted by bear

If you think those ratings are funny I suggest taking a look at their all-time heavyweights. Joe Louis is 15th and Ingo Johanson 7th. Yowzah.

Bear

I really do wonder what the criteria is. Now I like Connecticut's own Lee Q Murry, but number 29 all-time heavyweight? As long as we are on the subject of Tyson, boxrec may need to work on their formulas confuciusly.

I think it must favor bussiness in some way, look where they put Jimmy Bivins (who fought LHs mainly) and Gunboat Smith on the all-time HW list - double yowzah!

Christopher

[bear]

I never realized that Jack Dempsey is the 72nd best heavy of all time either.

Bear

Originally Posted by CONN CHRIS

I really do wonder what the criteria is. Now I like Connecticut's own Lee Q Murry, but number 29 all-time heavyweight? As long as we are on the subject of Tyson, boxrec may need to work on their formulas confuciusly.

I think it must favor bussiness in some way, look where they put Jimmy Bivins (who fought LHs mainly) and Gunboat Smith on the all-time HW list - double yowzah!

Christopher

[mh2365]

Originally Posted by Dave1877

I can't believe they have Skelton so high, he should be no higher than 25-30 in my opinion, if that.

Also why do you say Vitali is overrated?

I looked at my list I don't have any notations about Vitali, you might of read it wrong. I thought Valuev is way underrated and whoeever rated Skelton originally at 10 was drinking Matt-aide at the time. (Sure Mr. Jones Kool-aid is great).

[mh2365]

General

* The ratings program uses the records of the boxers to calculate their rating. There are no other factors involved.
* Ratings are not influenced by opinion in any way.
* There are inaccuracies (especially lower down the ratings) because of incomplete records. However the ratings are continually improving as new bouts are entered into the database.
* The ratings are generated at approximately 10:35 GMT every day.

[edit]
Ratings structure

* The effective ratings consist of
o a basic rating
o an overlay cut mechanism

which every day evaluate all bouts in the database in chronological sequence.

All parameter values are optimized to get best prediction quote of bout results in BoxRec database. So these ratings are optimized performance ratings. A higher rated boxer should be expected to defeat a lower rated boxer with increasing probability by increasing rating difference.
[edit]
Effective rating

The effective rating is the minimum of the basic rating and the cut plus 1/1000 of the basic rating.

r_eff = min(r_bas,cut) + r_bas/1000
[edit]
Basic rating

1. Every boxer gets a first rating of 1.000 before his first bout.
2. After every bout the rating of the two boxers involved is changed depending on the bout's official result (KO, TKO, TD, UD, PTS, MD, SD, DQ, D).
3. All bouts are regarded to have the same weight independent of titles.
4. The winner gets a basic gain / the loser a basic loss of points depending on the type of result.
5. This gain / loss is corrected by some amount depending on the difference between the ratings of the opponents before the bout. If the winner was rated higher than his opponent before the bout, his gain will be decreased. If his rating was lower, his gain will be increased. The loser's loss is handled in the same way.
6. For Unanimous Decisions, Points Win, Majority Decisions, and Split Decisions, the gain/loss will be calculated directly from the judges' scores, if available.
7. A winner's rating will not decrease, if the result was KO, TKO, TD, or DQ. The loser's will not increase in this case.
8. If a winner's rating would be lower than the loser's, the winner will be set to the mean value of their ratings before the bout. The loser will be set to 1 point lower.
9. The ratings are decreased for moving up to higher weight divisions by some factor - and they are increased by some smaller factor for moving down the divisions. This factor depends on the the weight limits of the divisions.
10. All ratings are decreased by a small factor on a daily basis.
11. For every 12 months of inactivity the rating is decreased by some value.
12. Parameter values

* TKO = 90
* TD = 90
* KO = 80
* UD = 80
* PTS = 60
* MD = 50
* SD = 50
* DQ = 50
* D = 0
* SCORES for UD, PTS, MD, SD: 20 * mean score difference per judge; maximum 90
* R_DIFF = correction for rating difference of opponents: 13.8 percent
* DECAY = continuous rating decay of 1.67 percent/year on a daily basis
* INACT = after every 12 month of inactivity a boxer loses 90 points
* WEIGHT_UP: r_up = r*(weight_div/weight_div_up)**0.3
* WEIGHT_DOWN: r_down = r*(weight_div/weight_div_down)**0.15

Formula

If a boxer with rating of r_a before the fight defeats a boxer b with a rating of r_b before the fight with result RES the new ratings r_a_new and r_b_new after a fight are:

r_a_new = r_a + RES + (r_b - r_a)*R_DIFF

r_b_new = r_b - RES + (r_a - r_b)*R_DIFF
[edit]
Cut mechanism

1. Every boxer gets a first cut of 90 before his first bout.
2. The cut is set to the effective rating of the best defeated opponent + 90 points
3. If one or both boxers have less than 8 bouts the cut may be reduced by some factor.
4. If the effective rating of the winner would be lower than the loser's, the cut of the winner will be set to the mean value of the their calculated cuts. The loser's value will be set 1 point lower.
5. The cut is decreased by a small factor on a daily basis.
6. For every 12 months of inactivity the cut is decreased by some value.
7. Parameter values

* DECAY = continuous cut decay of 1.67 percent/year on a daily basis
* INACT = after every 12 month of inactivity a boxer's cut loses 90 points

Formula for effective cut after reduction for less than 8 bouts:

cut_eff = min(cut,r_eff_opp+90-675*(1-min(bouts,8)*min(bouts_opp,8)/64)


How boxrec rates their fighters. It's done by computer formula which any college football fan knows is the best . With that said it'd be nice to see that type of formula incorporated into TBCB, since obviously we can't have a subjective (ie human) rating.

I esp like this part -
"4. The winner gets a basic gain / the loser a basic loss of points depending on the type of result."

That's one thing I've been harping on since day one ... a KO should get more points than a MD, but as far as I can tell in TBCB it doesn't.

[Dave1877]

Originally Posted by mh2365

I looked at my list I don't have any notations about Vitali, you might of read it wrong. I thought Valuev is way underrated and whoeever rated Skelton originally at 10 was drinking Matt-aide at the time. (Sure Mr. Jones Kool-aid is great).

I get it now, thanks for clearing it up

[mh2365]

This division took me almost 2 days to go through .... Jim has his work cut out for him as 19 of the World Top 50 have not been rated by anyone and who knows how many of the 31 left were rated by Jim

45 new photos without files (14 in WT50)
4 missing photos
8 fighters w/o photos that merit rating (5 in WT50)

alberto colajanni 14468
christian cruz 10522
joe spina 46847
mohammed said 50575
francis cheka 185091

victor oganov 16978
ryan kerr 65640
dimitri sartison 213976


WT50 in bold

[mh2365]

Here's something I didn't know ... as I am going through the middleweights I noticed that boxrec counted the Contender fights toward the fighters records.

[CONN CHRIS]

Originally Posted by mh2365

That's one thing I've been harping on since day one ... a KO should get more points than a MD, but as far as I can tell in TBCB it doesn't.

I still think that the simplest way to fix the ranking in TBCB is to allow the user to move fighters up and down the ranking scale generated by the program. Then include a checkbox to lock the rankings. This way you could set them where you want and have them stay. Later, if you want a new ranking, uncheck the lock checkbox and allow the program to re-rank, shuffle as you wish, then lock again.

Christopher

[Tosti]

[QUOTE=CONN CHRIS]I still think that the simplest way to fix the ranking in TBCB is to allow the user to move fighters up and down the ranking scale generated by the program. Then include a checkbox to lock the rankings. This way you could set them where you want and have them stay. Later, if you want a new ranking, uncheck the lock checkbox and allow the program to re-rank, shuffle as you wish, then lock again.

Christopher[/QUOTE

Good Idea

Personal input would be V useful

[Brooks TVB]

Originally Posted by mh2365

Here's something I didn't know ... as I am going through the middleweights I noticed that boxrec counted the Contender fights toward the fighters records.

Those fights were sanctioned fights and do count on their record. I think they should have been longer fights, say 8 rounds but thats what they were. You really don't see much of a fighter in five rounds.

[mh2365]

Lots of middleweights also

62 new photos without files (13 in WT50)
2 missing photos
7 fighters w/o photos that merit rating (5 in WT50)

khoren gevor 18807
david alfonso lopez 42862
paulo alejandro sanchez 14665
lorenzo di giacomo 16996
toshihiko Itagaki 54764


james obede toney 65473
antonio arras 33901


WT50 in bold

[mh2365]

here are the jr. middle weights

59 new photos without files (11 in WT50)
4 missing photos
8 fighters w/o photos that merit rating (5 in WT50) (list below)


nobuhiro ishida 24450
attila kovacs 137199
hassan al 7703
hamlet petrosian 75787
albert airapeitian 14515

mike miranda 76345
alexey chirkov 41986
cornelius bundrage 31800

WT50 in bold