Page 42 of 180
Posted: 23 Oct 2007, 13:42
by JCS
computerrank wrote:JCS wrote:computerrank wrote:... would make some 40 points difference for Povetkin ... perhaps
Larry Donald's rating also dropped before Povetkin's win against him.
40 points can mean a dozen ranking spots or more.
Why not implement gradual penalties? They are undoubtedly more accurate and less anomalous in appearance.
... 40 points would make 1 or 2 ranks for Povetkin, if he should win.
The basic reason, why the penalty comes after 18 months completed is - a boxer's rating should be tangled only, as far as he has missed valid opposition within this period - to the last day.
Otherwise the implicit requirement would be to box this level of opposition in a shorter time period - and this is not intended.
And in addition - we would have daily sliding ratings - which are not accepted too, as we now.
Set-backs after periods for missing opposition quality or inactivity are quite usual in common ABC rankings too.
Perhaps this inactive boxer would be penalized by the ABCs after long activity, but if they were to prove themselves after the inactivity.. the penalty would be totally erased. I realize we do some corrections upon returning from inactivity.. but I'm not sure they are enough.
Furthermore, organizations don't do daily shifts for penalties because it is not feasible as a human would have to perform these tasks.
My main problem is the cliff-like drop that I explained. At the very least, I believe there should be a gradual slope from 18-24 months to soften those just around the cut-off.
Povetkin didn't get the credit he deserved for beating Donald, as Donald's rating was pummeled by inactivity.
Posted: 24 Oct 2007, 15:15
by JCS
Hill #123 vs Arslan #10 on BoxRec ratings..
but IBO has Hill #12 vs Arslan #13.. Which is more realistic.
I mean.. inactivity or not, title or not.. whatever. Hill HAS to be in the Top 100.. At least!!
Posted: 24 Oct 2007, 15:19
by m1kee50
JCS wrote:Hill #123 vs Arslan #10 on BoxRec ratings..
but IBO has Hill #12 vs Arslan #13.. Which is more realistic.
I mean.. inactivity or not, title or not.. whatever. Hill HAS to be in the Top 100.. At least!!
I guess you are only going to get penalised for losing to a man who hasnt fought in ten years.... I agree its a bit harsh, but its the vagaries of a computerised impartial system, at least everyone suffers the same penalities here, I dont know how the IBO organise themselves
Posted: 24 Oct 2007, 15:25
by JCS
MIKEE wrote:JCS wrote:Hill #123 vs Arslan #10 on BoxRec ratings..
but IBO has Hill #12 vs Arslan #13.. Which is more realistic.
I mean.. inactivity or not, title or not.. whatever. Hill HAS to be in the Top 100.. At least!!
I guess you are only going to get penalised for losing to a man who hasnt fought in ten years.... I agree its a bit harsh, but its the vagaries of a computerised impartial system, at least everyone suffers the same penalities here, I dont know how the IBO organise themselves
IBO almost certainly has a human affecting the results. Their "blocking" looks too good for a computerized system. (Example: The Briggs/Lyakhovich situation)
This gets back to what I was saying though... in that a fighter should be given a great deal of the penalty back when they come off of inactivity and prove themselves. Martin implemented my ideas concerning adjustments in a situation like Maske/Hill.. but I think Hill got tagged by some other penalty recently. He was around #60.
Posted: 26 Oct 2007, 11:42
by Lennox
IBO ratings are all done on a formula JCS. Human input is for division changes, some injury losses have apparently been over-ridden.
Posted: 26 Oct 2007, 11:51
by JCS
Lennox wrote:IBO ratings are all done on a formula JCS. Human input is for division changes, some injury losses have apparently been over-ridden.
That counts as human input.
Posted: 26 Oct 2007, 11:56
by Lennox
JCS wrote:Lennox wrote:IBO ratings are all done on a formula JCS. Human input is for division changes, some injury losses have apparently been over-ridden.
That counts as human input.
Yes, I agree
Posted: 31 Oct 2007, 14:47
by Lennox
Chris Byrd at 56 and behind Peter Okello. Boxrec needs a new computer.
Posted: 31 Oct 2007, 14:49
by JCS
Lennox wrote:Chris Byrd at 56 and behind Peter Okello. Boxrec needs a new computer.
That is quite harsh.
I'm working on a modified system in my spare time. Hope to have results in a few days, maybe I'll change Martin's mind on these tremendous penalties.
Posted: 02 Nov 2007, 19:34
by JCS
I think I need a new computer too. Here's my first stab at a computer ranking..
HW
1 Wladimir Klitschko
2 Ruslan Chagaev
3 Nikolay Valuev
4 Samuel Peter
5 Sultan Ibragimov
6 Alexander Povetkin
7 Calvin Brock
8 Oleg Maskaev
9 Lamon Brewster
10 Matt Skelton
Byrd at #20... Although I gotta figure out why Golota is at #13 :)
Duddy is at #38.. so I know something is working right.
Posted: 02 Nov 2007, 22:11
by sockdolager
JCS wrote:I think I need a new computer too. Here's my first stab at a computer ranking..
HW
1 Wladimir Klitschko
2 Ruslan Chagaev
3 Nikolay Valuev
4 Samuel Peter
5 Sultan Ibragimov
6 Alexander Povetkin
7 Calvin Brock
8 Oleg Maskaev
9 Lamon Brewster
10 Matt Skelton
Byrd at #20... Although I gotta figure out why Golota is at #13 :)
Duddy is at #38.. so I know something is working right.
At HW?
Posted: 02 Nov 2007, 22:13
by The Irish Assassin
sockdolager wrote:JCS wrote:I think I need a new computer too. Here's my first stab at a computer ranking..
HW
1 Wladimir Klitschko
2 Ruslan Chagaev
3 Nikolay Valuev
4 Samuel Peter
5 Sultan Ibragimov
6 Alexander Povetkin
7 Calvin Brock
8 Oleg Maskaev
9 Lamon Brewster
10 Matt Skelton
Byrd at #20... Although I gotta figure out why Golota is at #13 :)
Duddy is at #38.. so I know something is working right.
At HW?
He's that good, he is ranked in a division he doesn't even fight in!! I TOLD YOU HE DOESN'T SUCK
He should fight Mora.
Posted: 02 Nov 2007, 22:24
by JCS
sockdolager wrote:JCS wrote:I think I need a new computer too. Here's my first stab at a computer ranking..
HW
1 Wladimir Klitschko
2 Ruslan Chagaev
3 Nikolay Valuev
4 Samuel Peter
5 Sultan Ibragimov
6 Alexander Povetkin
7 Calvin Brock
8 Oleg Maskaev
9 Lamon Brewster
10 Matt Skelton
Byrd at #20... Although I gotta figure out why Golota is at #13 :)
Duddy is at #38.. so I know something is working right.
At HW?
Sorry. MW :)
Posted: 03 Nov 2007, 03:03
by m1kee50
JCS wrote:Lennox wrote:Chris Byrd at 56 and behind Peter Okello. Boxrec needs a new computer.
That is quite harsh.
I'm working on a modified system in my spare time. Hope to have results in a few days, maybe I'll change Martin's mind on these tremendous penalties.
Is that before or after he got sparked by a novice?
Posted: 05 Nov 2007, 14:51
by JCS
Martin,
http://www.boxrec.com/list_bouts.php?hu ... &cat=boxer
How can this guy be ranked above 100+ others? Indicative of an underlying problem? I think so.
Posted: 05 Nov 2007, 14:56
by JCS
Cobwebcat wrote:Question for JCS.
In the all time rankings how many fighters have a perfect record (all wins, no draws) plus a rating above 450.
I'm guessing only:
Marciano
Mayweather
Calzaghe
Hatton
Ottke
Cotto
and do you have their current all time rating?
Ta
EDIT: I'll check later.
Posted: 05 Nov 2007, 14:56
by computerrank
Looks strange ... where did he get his 3 points from?
Posted: 05 Nov 2007, 15:00
by JCS
computerrank wrote:
Looks strange ... where did he get his 3 points from?
You tell me. :) I don't see how he got points. Even if he got any, why would he retain them?
I'll see if I can determine this shortly.
Posted: 05 Nov 2007, 15:40
by JCS
The 3 points are not earned through the bouts. It must be something you're doing on the output.
I wonder if it is something you do during weight equalization?
Posted: 05 Nov 2007, 16:52
by JCS
Cobwebcat wrote:JCS wrote:Cobwebcat wrote:Question for JCS.
In the all time rankings how many fighters have a perfect record (all wins, no draws) plus a rating above 450.
I'm guessing only:
Marciano
Mayweather
Calzaghe
Hatton
Ottke
Cotto
and do you have their current all time rating?
Ta
450 is way, way too low of a number to use.
However, ironically enough, 1450 produces the exact list you gave me.
FMJ - 1653
Calzaghe - 1537
Ottke - 1481
Marciano - 1520
Hatton - 1558
Cotto - 1473
Dropping to 1400 adds 3 more
Juan Diaz - 1405
Samson Dutch Boy Gym - 1416
Ivan Calderon 1403
Thanks for that JCS.
Something strange though...before the all time values vanished they said:
Marciano 1425
Mayweather 1116
Calzaghe 753
Hatton 724
Ottke 516
Cotto 498
Hence why I used 450...what's changed?
Oops Sorry. I grabbed those values from an old database.
Posted: 05 Nov 2007, 17:21
by JCS
Martin,
I believe this boxer is getting 3 points in the equal_rating_for_division function just before the database update.
Posted: 05 Nov 2007, 20:11
by JCS
JCS wrote:computerrank wrote:
Looks strange ... where did he get his 3 points from?
You tell me. :) I don't see how he got points. Even if he got any, why would he retain them?
I'll see if I can determine this shortly.
Martin,
On Version M. This boxer goes in as 0.38 points.
Posted: 05 Nov 2007, 20:37
by JCS
Cob,
Give me some time. The test server is really slow.
Posted: 05 Nov 2007, 20:45
by JCS
Cobwebcat wrote:I wonder how the computer using Boxrec ratings before every fight would be fairing in your prediction competition?
Pretty good, but not excellent. PTBF requires a bit more foresight than the current ratings are configured to consider or reflect.
Posted: 05 Nov 2007, 21:01
by JCS
Cob, here's your list.
Mayweather - 1117
Calzaghe - 754
Ottke - 517
Marciano - 1426
Hatton - 723
Cotto - 498
I've been tinkering with a rating system of my own.. here's my Top 20 Heavyweights
Code: Select all
Wladimir Klitschko 380
Ruslan Chagaev 277
Nikolay Valuev 242
Samuel Peter 185
Oleg Maskaev 176
Tony Thompson 156
Alexander Povetkin 154
Sultan Ibragimov 149
Matt Skelton 142
Eddie Chambers 133
Vladimir Virchis 122
Hasim Rahman 116
Lamon Brewster 111
Calvin Brock 111
Chris Byrd 107
Juan Carlos Gomez 100
John Ruiz 99
Shannon Briggs 89
Luan Krasniqi 87
Michael Sprott 87