I'm not a subscriber to him, I don't love him, I don't hate him, but I do like looking at his rankings and hearing his take on things. But I have a question:
I know his model is based on several different factors, but what role do actual wins play in the model?
I see MU with wins over #5, #7 (twice), #10, and #15 and I'm just curious to know how much weight those wins hold in his rankings.
Not to be pedantic, the Pomeroy Rankings are all based on a simple concept and formula (Points per Possession)
The ranking is Offensive Points per Possession (OPPP) less Defensive Points Per Possession (DPPP). Every time a team wins, OPPP is higher than DPPP, so that goes into the ranking.
His model is adjusted further for several levels, including the quality of the opponents in aggregate. For example, Marquette's Offensive Efficiency is given a boost because it's been done against the #1 overall Defensive SOS. However, MU's Defensive Efficiency is bumped down because it's been against the #66 overall Offensive SOS.
Everything is based on offensive and defensive efficiency (how many points you are expected to score/allow in a given possession). So there is no explicit inclusion of winning or losing anywhere in his method. In consequence, a one point road loss to a good team can increase your ranking, while a home victory over a bad team may cause a decrease if you do not blow them out. And a one point loss vs. a one point victory will have very little differential effect. In contrast, the RPI only counts victories, and ignores efficiency/scoring margin, though it does account for home and away wins differently.
Quote from: THRILLHO on February 26, 2013, 09:28:58 AM
Everything is based on offensive and defensive efficiency (how many points you are expected to score/allow in a given possession). So there is no explicit inclusion of winning or losing anywhere in his method. In consequence, a one point road loss to a good team can increase your ranking, while a home victory over a bad team may cause a decrease if you do not blow them out. And a one point loss vs. a one point victory will have very little differential effect. In contrast, the RPI only counts victories, and ignores efficiency/scoring margin, though it does account for home and away wins differently.
This is why I like Pomeroy's site for all the in-depth stats behind the pay wall, but I like Sagarin's rankings a bit better. It basically merges Pomeroy's overall efficiency rankings with a W/L component that factors in resume with likelihood of future victory.
Quote from: Warrior's Path on February 26, 2013, 09:28:09 AM
The ranking is Offensive Points per Possession (OPPP) less Defensive Points Per Possession (DPPP). Every time a team wins, OPPP is higher than DPPP, so that goes into the ranking.
Is that necessarily true? For instance, what if a team gets 60 possessions in a game and scores 1.2 points per possession thereby giving them 72 whereas the losing team gets 50 possession with 1.4 points per possession giving them 70.
I may very well be misunderstanding the concepts.
Quote from: windyplayer on February 26, 2013, 02:56:40 PM
Is that necessarily true? For instance, what if a team gets 60 possessions in a game and scores 1.2 points per possession thereby giving them 72 whereas the losing team gets 50 possession with 1.4 points per possession giving them 70.
I may very well be misunderstanding the concepts.
The way Pomeroy does it, possessions are given for the entire game. Last night's game was played at a tempo of 61 possessions for both teams.
Even if you calculate possessions on a per-team basis, it rarely differs by more than one or two per team.
Quote from: Warrior's Path on February 26, 2013, 03:13:44 PM
The way Pomeroy does it, possessions are given for the entire game. Last night's game was played at a tempo of 61 possessions for both teams.
Even if you calculate possessions on a per-team basis, it rarely differs by more than one or two per team.
The math on possessions is field goals attempted + Turnovers - offensive rebounds + (.475 x free throws attempted).
Then you average the possessions for both teams and round to get the possessions for the game.
Quote from: Brewtown Andy on February 26, 2013, 03:48:16 PM
The math on possessions is field goals attempted + Turnovers - offensive rebounds + (.475 x free throws attempted).
Then you average the possessions for both teams and round to get the possessions for the game.
Yes, I know how to calculate possessions and determine the game tempo.
Quote from: windyplayer on February 26, 2013, 02:56:40 PM
Is that necessarily true? For instance, what if a team gets 60 possessions in a game and scores 1.2 points per possession thereby giving them 72 whereas the losing team gets 50 possession with 1.4 points per possession giving them 70.
I may very well be misunderstanding the concepts.
It wouldn't be possible for one team to get 60 possessions and the other to get 50. The greatest disparity in possessions in a regulation game is +2, if a team won the tip, had the last possession of the first half, had the possession arrow in their favor to start the second half, and had the final possession of the game they would be +2.
Even if you turn the ball over on an inbound, it's considered a possession and turnover.
Quote from: Warrior's Path on February 26, 2013, 04:17:21 PM
Yes, I know how to calculate possessions and determine the game tempo.
Wasn't necessarily for you. :D
Kenpom knows what he's doing. If you compare the Vegas line to his predictions they are within 1-3 points 95% of the time. The over/under is just as close. My guess is Vegas has been doing this for years and websites have brought attention to it.
surprised Wisconsin did not move up 4 spots in Pomeroy with their crushing of Nebraska....
Quote from: madtownwarrior on February 27, 2013, 11:29:52 AM
surprised Wisconsin did not move up 4 spots in Pomeroy with their crushing of Nebraska....
We get it. You have a beef that Madison and other low possession, high efficency teams are 'overvalued' by Pomroy. Thing is, when you make your methodology transparent then people reading your rankings can adjust as they see fit.
You're not adding anything to the conversation when you keep harping on a point that everyone is well aware of.