MTigerBlue
Heisman
Posts: 5,579
Joined: Apr 2009
Reputation: 421
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 06:54 PM)Brother Bluto Wrote: (01-15-2017 01:41 PM)bubbapt Wrote: (01-15-2017 10:42 AM)Brother Bluto Wrote: (01-15-2017 10:38 AM)bubbapt Wrote: (01-15-2017 10:31 AM)Brother Bluto Wrote: I'm just not sure KenPom is all that to be honest
It's like the new in fad
If you call removing bias to measure efficiency a fad, then it's a fad.
What bias does it remove?
It provides tempo free offensive and devensive stats, which are then weighted against how good your opponents tempo free stats are.
Like I've said before .. Where is the starting point ? What decides who is good when everyone is 0-0? Or does it go by the previous season and returning players?
Previous season counts in Sagarin, not RPI. In the RPI, everyone starts out equal.
|
|
01-15-2017 07:47 PM |
|
bubbapt
Uh, what?
Posts: 12,894
Joined: Dec 2006
Reputation: 480
I Root For: Memphis
Location: St. Louis
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 07:47 PM)MTigerBlue Wrote: (01-15-2017 06:54 PM)Brother Bluto Wrote: (01-15-2017 01:41 PM)bubbapt Wrote: (01-15-2017 10:42 AM)Brother Bluto Wrote: (01-15-2017 10:38 AM)bubbapt Wrote: If you call removing bias to measure efficiency a fad, then it's a fad.
What bias does it remove?
It provides tempo free offensive and devensive stats, which are then weighted against how good your opponents tempo free stats are.
Like I've said before .. Where is the starting point ? What decides who is good when everyone is 0-0? Or does it go by the previous season and returning players?
Previous season counts in Sagarin, not RPI. In the RPI, everyone starts out equal.
Kenpom stats at this point of the season calculate games from the start of the season.
|
|
01-15-2017 07:52 PM |
|
Tiger87
Hall of Famer
Posts: 19,214
Joined: Jan 2012
Reputation: 1251
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
|
|
01-15-2017 09:30 PM |
|
bubbapt
Uh, what?
Posts: 12,894
Joined: Dec 2006
Reputation: 480
I Root For: Memphis
Location: St. Louis
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 09:30 PM)Tiger87 Wrote: My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
Champion Villanova was a 2 seed last year, but had the #1 ranking in Kenpom.
2015 Champion Duke had a #3 ranking in Kenpom, which would be fine for a #1 seed.
2014 Champion UConn was a 7 seed, but ranked #15 in kenpom, which would put them at around a 4 seed.
2013 Champion Louisville had the #1 ranking in Kenpom
2012 champion Kentucky had the #1 ranking
2011 champion UConn had a #10 ranking
2010 champion Duke had the #1 ranking
2009 champion North Carolina had the #1 ranking
2008 champion Kansas had the #1 ranking in kenpom (Memphis was #2)
seven of the last nine champions had #1 rankings in kenpom, with UConn being the spoiler.
Pretty good track record.
(This post was last modified: 01-15-2017 10:15 PM by bubbapt.)
|
|
01-15-2017 09:42 PM |
|
Tiger87
Hall of Famer
Posts: 19,214
Joined: Jan 2012
Reputation: 1251
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 09:42 PM)bubbapt Wrote: (01-15-2017 09:30 PM)Tiger87 Wrote: My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
Champion Villanova was a 2 seed last year, but had the #1 ranking in Kenpom.
2015 Champion Duke had a #3 ranking in Kenpom, which would be fine for a #1 seed.
2014 Champion UConn was a 7 seed, but ranked #15 in kenpom, which would put them at around a 4 seed.
2013 Champion Louisville had the #1 ranking in Kenpom
2012 champion Kentucky had the #1 ranking
2011 champion UConn had a #10 ranking
2010 champion Duke had the #1 ranking
2009 champion North Carolina had the #1 ranking
2008 champion Kansas had the #1 ranking in kenpom (Memphis was #2)
seven of the last nine champions had #1 rankings in kenpom, with UConn being the spoiler.
Pretty good track record.
But doesn't he post his final rankings after the tournament? So you win in the tourney and you're obviously climbing up. And there's the "luck" factor he can bump up, too - after the fact. You let me post my final rankings after the tournament, and I'll be 100%!
(This post was last modified: 01-15-2017 10:26 PM by Tiger87.)
|
|
01-15-2017 10:21 PM |
|
Tiger87
Hall of Famer
Posts: 19,214
Joined: Jan 2012
Reputation: 1251
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
The example of what I'm referring to with KenPom is always found in Virginia and Wisconsin. His algorithm love both those schools. Every year. They're top 10.
No doubt they're good. But they get over-rewarded on KenPom because of their style of play.
|
|
01-15-2017 10:28 PM |
|
Whale of a Guy
Banned
Posts: 83
Joined: Apr 2016
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 10:31 AM)Brother Bluto Wrote: I'm just not sure KenPom is all that to be honest
It's like the new in fad
Honest like telling everyone you "own" season tickets?
|
|
01-15-2017 10:47 PM |
|
bubbapt
Uh, what?
Posts: 12,894
Joined: Dec 2006
Reputation: 480
I Root For: Memphis
Location: St. Louis
|
RE: NCAA to go beyond RPI to assess at large
(01-15-2017 10:21 PM)Tiger87 Wrote: (01-15-2017 09:42 PM)bubbapt Wrote: (01-15-2017 09:30 PM)Tiger87 Wrote: My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
Champion Villanova was a 2 seed last year, but had the #1 ranking in Kenpom.
2015 Champion Duke had a #3 ranking in Kenpom, which would be fine for a #1 seed.
2014 Champion UConn was a 7 seed, but ranked #15 in kenpom, which would put them at around a 4 seed.
2013 Champion Louisville had the #1 ranking in Kenpom
2012 champion Kentucky had the #1 ranking
2011 champion UConn had a #10 ranking
2010 champion Duke had the #1 ranking
2009 champion North Carolina had the #1 ranking
2008 champion Kansas had the #1 ranking in kenpom (Memphis was #2)
seven of the last nine champions had #1 rankings in kenpom, with UConn being the spoiler.
Pretty good track record.
But doesn't he post his final rankings after the tournament? So you win in the tourney and you're obviously climbing up. And there's the "luck" factor he can bump up, too - after the fact. You let me post my final rankings after the tournament, and I'll be 100%!
If he was just posting numbers, then yes, but he is setting up a statistical calculation before the season starts, and showing how it calculated to actual results. He's not looking at what happened and retrofitting his calculation.
|
|
01-16-2017 07:29 AM |
|
Tiger87
Hall of Famer
Posts: 19,214
Joined: Jan 2012
Reputation: 1251
I Root For: Memphis
Location:
|
RE: NCAA to go beyond RPI to assess at large
(01-16-2017 07:29 AM)bubbapt Wrote: (01-15-2017 10:21 PM)Tiger87 Wrote: (01-15-2017 09:42 PM)bubbapt Wrote: (01-15-2017 09:30 PM)Tiger87 Wrote: My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
Champion Villanova was a 2 seed last year, but had the #1 ranking in Kenpom.
2015 Champion Duke had a #3 ranking in Kenpom, which would be fine for a #1 seed.
2014 Champion UConn was a 7 seed, but ranked #15 in kenpom, which would put them at around a 4 seed.
2013 Champion Louisville had the #1 ranking in Kenpom
2012 champion Kentucky had the #1 ranking
2011 champion UConn had a #10 ranking
2010 champion Duke had the #1 ranking
2009 champion North Carolina had the #1 ranking
2008 champion Kansas had the #1 ranking in kenpom (Memphis was #2)
seven of the last nine champions had #1 rankings in kenpom, with UConn being the spoiler.
Pretty good track record.
But doesn't he post his final rankings after the tournament? So you win in the tourney and you're obviously climbing up. And there's the "luck" factor he can bump up, too - after the fact. You let me post my final rankings after the tournament, and I'll be 100%!
If he was just posting numbers, then yes, but he is setting up a statistical calculation before the season starts, and showing how it calculated to actual results. He's not looking at what happened and retrofitting his calculation.
I mean, he is adjusting based on results. So you go on a 6 game win streak against at least 4 other highly rated teams and you're going to climb up. And he has the "luck" variable - which to me just screams "plug number".
|
|
01-16-2017 08:42 PM |
|
bubbapt
Uh, what?
Posts: 12,894
Joined: Dec 2006
Reputation: 480
I Root For: Memphis
Location: St. Louis
|
RE: NCAA to go beyond RPI to assess at large
(01-16-2017 08:42 PM)Tiger87 Wrote: (01-16-2017 07:29 AM)bubbapt Wrote: (01-15-2017 10:21 PM)Tiger87 Wrote: (01-15-2017 09:42 PM)bubbapt Wrote: (01-15-2017 09:30 PM)Tiger87 Wrote: My opinion is that Ken does a good job with the analytics - possessions, tempo, efficiency, etc. But is not a good ranking system. It rewards stats - but undervalues results of the game. You will frequently see top 10 KenPom teams that have a ton of losses - just because they play "efficiently".
Of course, the RPI is worse - it's just a SOS measurement. And Sagarin is comparable to KenPom. Not sure about BPI. Nothing is perfect. Probably best to do a composite of RPI, Sag, KP and BPI.
Champion Villanova was a 2 seed last year, but had the #1 ranking in Kenpom.
2015 Champion Duke had a #3 ranking in Kenpom, which would be fine for a #1 seed.
2014 Champion UConn was a 7 seed, but ranked #15 in kenpom, which would put them at around a 4 seed.
2013 Champion Louisville had the #1 ranking in Kenpom
2012 champion Kentucky had the #1 ranking
2011 champion UConn had a #10 ranking
2010 champion Duke had the #1 ranking
2009 champion North Carolina had the #1 ranking
2008 champion Kansas had the #1 ranking in kenpom (Memphis was #2)
seven of the last nine champions had #1 rankings in kenpom, with UConn being the spoiler.
Pretty good track record.
But doesn't he post his final rankings after the tournament? So you win in the tourney and you're obviously climbing up. And there's the "luck" factor he can bump up, too - after the fact. You let me post my final rankings after the tournament, and I'll be 100%!
If he was just posting numbers, then yes, but he is setting up a statistical calculation before the season starts, and showing how it calculated to actual results. He's not looking at what happened and retrofitting his calculation.
I mean, he is adjusting based on results. So you go on a 6 game win streak against at least 4 other highly rated teams and you're going to climb up. And he has the "luck" variable - which to me just screams "plug number".
Yes, his calculations are entirely based upon results.
|
|
01-19-2017 07:13 AM |
|