Following up on the last post, the final score in the NCAA championship game was of course, UNC 89 Michigan State 72 in a game that was not even as close as the score indicated.
Obviously a sample size of one is meaningless, although the Pomeroy ratings did not distinguish themselves with the 14-game sample discussed here either.
Showing posts with label RPI. Show all posts
Showing posts with label RPI. Show all posts
Wednesday, April 8, 2009
Monday, April 6, 2009
GIGO?
Ken Pomeroy has done a terrific job documenting and analyzing data relating to college basketball, much like Bill James for baseball a generation ago. His blog (and writing at Basketball Prospectus) has introduced new stats such as "Effective Height" and "Defensive Fingerprint." Stats such as these have given observers a set of tools to use to evaluate College Basketball teams.
That being said, the core stat -- the Pomeroy Rating -- has been less than stellar during this NCAA tournament. Although there were few major upsets in the first two rounds of the tournament (meaning that the "power conference" teams mainly survived to the Sweet Sixteen Round), Pomeroy's ratings have had middling results.
With that in mind, let's take a tour back through each of the four regions, comparing Pomeroy's ratings against other ranking systems: the NCAA Men's Committee itself (NCAA), the final AP poll (AP), and RPI (Ratings Percentage Index, as calculated here. (RPI))
Only Sweet Sixteen and later games are used, to try and emphasize the differences between the systems. (After all, every system will get the 1-16 and 2-15 games correctly; by the time you reach the round of 16, the choices are hard, as anyone who's ever done a bracket will tell you.) Note that while KenPom systems remains 'dynamic', adding information from each successive game and updating results, the NCAA, AP, and RPI ratings used here were 'fixed' at the end of the regular season.
EAST
Sweet Sixteen actual teams: Pitt vs. Xavier; Villanova vs. Duke
SOUTH
Sweet Sixteen actual teams: UNC vs. Gonzaga; Syracuse vs. Oklahoma
MIDWEST
Sweet Sixteen actual teams: Louisville vs. Arizona; Kansas vs, Michigan State.
WEST
Sweet Sixteen Actual teams: UConn vs. Purdue; Missouri vs. Memphis
FINAL FOUR:
UNC vs. Villanova; MSU vs. UConn.
CHAMPIONSHIP GAME:
UNC vs. MSU
As all four systems have the same prediction for tonight's game, there can be no change in the final standings, which leaves the much maligned RPI as the 'most accurate', at least for this year.
That being said, the core stat -- the Pomeroy Rating -- has been less than stellar during this NCAA tournament. Although there were few major upsets in the first two rounds of the tournament (meaning that the "power conference" teams mainly survived to the Sweet Sixteen Round), Pomeroy's ratings have had middling results.
With that in mind, let's take a tour back through each of the four regions, comparing Pomeroy's ratings against other ranking systems: the NCAA Men's Committee itself (NCAA), the final AP poll (AP), and RPI (Ratings Percentage Index, as calculated here. (RPI))
Only Sweet Sixteen and later games are used, to try and emphasize the differences between the systems. (After all, every system will get the 1-16 and 2-15 games correctly; by the time you reach the round of 16, the choices are hard, as anyone who's ever done a bracket will tell you.) Note that while KenPom systems remains 'dynamic', adding information from each successive game and updating results, the NCAA, AP, and RPI ratings used here were 'fixed' at the end of the regular season.
EAST
Sweet Sixteen actual teams: Pitt vs. Xavier; Villanova vs. Duke
Predicted results:
NCAA: (1) Pitt over (4) Xav (correct); (2) Duke over (3) Nova (incorrect). (1-1)
AP: (4) Pitt over (20) Xav (correct); (6) Duke over (14) Nova (incorrect) (1-1)
RPI: (2) Pitt over (17) Xav (correct); (1) Duke over (13) Nova (incorrect) (1-1)
KenPom: (5) Pitt over (20) Xav (correct); (11) Duke over (14) Nova (incorrect) (1-1)
Finals: Pitt vs. Villanova
NCAA: (1) Pitt over (3) Nova (incorrect). (1-2, cumulative)
AP: (4) Pitt over (14) Nova (incorrect) (1-2)
RPI: (2) Pitt over (13) Nova (incorrect) (1-2)
KenPom: (5) Pitt over (14) Nova (incorrect) (1-2)
SOUTH
Sweet Sixteen actual teams: UNC vs. Gonzaga; Syracuse vs. Oklahoma
NCAA: (1) UNC over (4) Gonzaga (correct); (2) OU over (3) Cuse (correct) (3-2, cumulative)
AP: (2) UNC over (10) Gonzaga (corret); (7) OU over (13) Cuse (correct) (3-2)
RPI: (3) UNC over (26) Zaga (correct); (5) OU over (12) Cuse (correct); (3-2)
KenPom: (3) UNC over (8) Zaga (correct); (13) OU over (15) Cuse (correct); (3-2)
Finals: UNC vs. OU
NCAA: (1) UNC over (2) OU over (4-2, cumulative)
AP: (2) UNC over (7) OU (correct) (4-2)
RPI: (3) UNC over (5) OU over (correct); (4-2)
KenPom: (3) UNC over (13) (correct); (4-2)
MIDWEST
Sweet Sixteen actual teams: Louisville vs. Arizona; Kansas vs, Michigan State.
NCAA: (1) Lou over (12) Ariz (correct); (2) MSU over (3) KU (correct) (6-2, cumulative)
AP: (1) Lou over (NR) Ariz (correct); (8) MSU over (14) KU (correct) (6-2)
RPI: (4) Lou over (62) Ariz (correct); (6) MSU over (11) KU (correct) (6-2)
KenPom: (4) Lou over (40) Ariz (correct); (7) MSU over (10) KU (correct) (6-2)
Finals: Lou vs. MSU
NCAA: (1) Lou over (2) MSU (incorrect) (6-3, cumulative)
AP: (1) Lou over (8) MSU (incorrect) (6-3)
RPI: (4) Lou over (6) MSU (incorrect)(6-3)
KenPom: (4) Lou over (7) MSU (incorrect) (6-3)
WEST
Sweet Sixteen Actual teams: UConn vs. Purdue; Missouri vs. Memphis
NCAA: (1) UConn over (5) Purdue (correct); (2) Mem over (3) Mizzou (incorrect) (7-4)
AP: (5) UConn over (17) Purdue (correct); (3) Mem over (9) Mizzou (incorrect) (7-4)
RPI: (8) UConn over (20) Purdue (correct); (7) Mem over (10) Mizzou (incorrect) (7-4)
KenPom: (2) Uconn over (17) Purdue (correct); (1) Mem over (6) Mizzou (incorrect) (7-4)
Finals: UConn vs. Mizzou
NCAA: (1) UConn over (3) Mizzou (correct) (8-4, cumulative)
AP: (5) UConn over (9) Mizzou (correct) (8-4)
RPI: (8) UConn over (10) Mizzou (correct) (8-4)
KenPom: (2) Uconn over (6) Mizzou (correct) (8-4)
FINAL FOUR:
UNC vs. Villanova; MSU vs. UConn.
NCAA: (1) UConn over (2) MSU (incorrect); (1) UNC over (3) Nova (correct) (9-5, cumulative)
AP: (5) UConn vs. (8) MSU (incorrect); (2) UNC over (11) Nova (correct); (9-5)
RPI: (6) MSU over (8) UConn (correct); (3) UNC over (13) Nova (correct) (10-4)
KenPom: (2) UConn over (7) MSU (incorrect); (3) UNC over (14) Nova (correct)(9-5)
CHAMPIONSHIP GAME:
UNC vs. MSU
NCAA: (1) UNC over (2) MSU
AP: (2) UNC over (8) MSU
RPI: (3) UNC over (6) MSU
KenPom: (3) UNC over (7) MSU
As all four systems have the same prediction for tonight's game, there can be no change in the final standings, which leaves the much maligned RPI as the 'most accurate', at least for this year.
RPI: 11-4 or 10-5
NCAA: 10-5 or 9-6
AP: 10-5 or 9-6
KenPom: 10-5 or 9-6
Sunday, February 24, 2008
Walkin' in Memphis
Last night's (#1) Memphis vs. (#2) Tennessee game (won by the Vols, 66-62) had plenty of interest for the Ivy League basketball fan.
Memphis, as has been well-documented, has been running an extension of the "Princeton"-offense, called "Dribble-Drive Motion" by Coach John Calipari. DDM moves the center to the weak side, rather than to the high post, but the net effect is similar: opening up the lane for cuts (Princeton) or dribble penetration (DDM).
(Calipari earlier in the year was calling DDM "Princeton-on-steroids" although now he seems to be exclusively using the "DDM" terminology; no word on whether Calipari has attempted to trademark "DDM", as he did with "Refuse to Lose" at UMass.)
Tennessee, meanwhile, appears to incorporate some elements of the Princeton-sets into its quick strike offense (at least on those possessions where the Vols aren't fast-breaking), using guard dribble-exchanges and having the center backscreen to create opportunities for cutters. On longer possessions, Tenn sent their big men down to the low post, and resulting series of hook shots were reminiscent of Georgetown's Roy Hibbert.
But as important, last night's game featured two teams that could easily face the prospective Ivy champ, Cornell (with all due respect to Barack Obama's brother-in-law, whose Brown Bruins are 3 games behind). Sports Illustrated projects Cornell as a #14 seed, and ESPN has them as a #13; the RPI puts Cornell at #72 and KenPom rates them as #127.
But with the trip to Penn and Princeton upcoming in two weekends, Cornell has not clinched an undefeated Ivy season yet. The Penn/Princeton weekend has been traditionally difficult, and the pressure of closing out a 14-0 season will likely make both games hotly contested.
The Tournament Committee (and the RPI/KenPom computers) will punish Cornell for a late-season loss. If so, don't be surprised to see the Big Red at #15 (or worse), and playing (perhaps) Memphis or Tennessee.
Memphis, as has been well-documented, has been running an extension of the "Princeton"-offense, called "Dribble-Drive Motion" by Coach John Calipari. DDM moves the center to the weak side, rather than to the high post, but the net effect is similar: opening up the lane for cuts (Princeton) or dribble penetration (DDM).
(Calipari earlier in the year was calling DDM "Princeton-on-steroids" although now he seems to be exclusively using the "DDM" terminology; no word on whether Calipari has attempted to trademark "DDM", as he did with "Refuse to Lose" at UMass.)
Tennessee, meanwhile, appears to incorporate some elements of the Princeton-sets into its quick strike offense (at least on those possessions where the Vols aren't fast-breaking), using guard dribble-exchanges and having the center backscreen to create opportunities for cutters. On longer possessions, Tenn sent their big men down to the low post, and resulting series of hook shots were reminiscent of Georgetown's Roy Hibbert.
But as important, last night's game featured two teams that could easily face the prospective Ivy champ, Cornell (with all due respect to Barack Obama's brother-in-law, whose Brown Bruins are 3 games behind). Sports Illustrated projects Cornell as a #14 seed, and ESPN has them as a #13; the RPI puts Cornell at #72 and KenPom rates them as #127.
But with the trip to Penn and Princeton upcoming in two weekends, Cornell has not clinched an undefeated Ivy season yet. The Penn/Princeton weekend has been traditionally difficult, and the pressure of closing out a 14-0 season will likely make both games hotly contested.
The Tournament Committee (and the RPI/KenPom computers) will punish Cornell for a late-season loss. If so, don't be surprised to see the Big Red at #15 (or worse), and playing (perhaps) Memphis or Tennessee.
Subscribe to:
Posts (Atom)