Thursday, March 27, 2008

NCAA Predictions: The Sweet Sixteen

The predictions below are based on TAPE matchups which compare teams in 20 statistical rate categories to determine how they stack up against each other, and then applies those rates to a possession model to determine how many points each is likely to score against the other.
Thursday's Games
7 West Virginia 69 4 Washington St 73
2 Xavier 68 (65 poss.) 1 N. Carolina 74 (67 poss.)

12 W. Kentucky 63 3 Louisville 73
1 UCLA 74 (66 poss.) 2 Tennessee 72 (72 poss.)
  • The two upsets predicted here are by the slimmest of margins. Louisville has only a 0.39-point edge (73.22-72.83 over 71.95 possessions) on the Volunteers. Tennessee projects to turn the Cardinals over on 24.2% of their possessions, and you might think it'd be hard to win when giving away that many possessions (and it is; teams turning the ball over at least 24% of the time won only 39% of the time this season), Louisville has won all five games in which it's done so this season.
  • West Virginia has a 0.57-point advantage over Xavier (68.68-68.11, 64.85 possessions) in a game that's close across the board. WVU should shoot the ball slightly better and turn the ball over slightly less often, but Xavier makes up for that with a small rebounding edge and getting to the line more often.
  • UCLA is a full standard deviation better than Western Kentucky. The Bruins' expected offensive rebounding percentage is 42.0% in this game (versus just 27.9% for WKU), and they should get to the free throw line more than twice as often as the Hilltoppers (25.5 FTA on average for UCLA against just 12.7 for WKU).
  • Offensive rebounding will be key in the UNC-WSU game. WSU has an expected slight edge in each of the other four factors, but UNC has a huge advantage on the glass, 37.8% to 25.1%.
Friday's Games
10 Davidson 60 3 Stanford 66
3 Wisconsin 68 (64 poss.) 2 Texas 65 (62 poss.)

12 Villanova 64 5 Michigan St. 65
1 Kansas 80 (70 poss.) 1 Memphis 71 (67 poss.)
  • Davidson beat Georgetown by getting ridiculously lucky. I know you're not supposed to say that about a Cinderella team, and the Wildcats are a great story and they're easy to root for, but Georgetown's likelihood of turning the ball over 20 times in 63 possessions against Davidson is about 5.6%. The chances of the Hoyas making 8 or fewer of 17 free throws in a game are 9.6%. Lightning probably won't strike twice for Davidson, especially against the second-best team in the country.
  • Texas-Stanford should be a heck of a game. The Longhorns don't turn the ball over (and the Cardinal doesn't force its opponents to turn the ball over), so big advantage there for Texas; Stanford is a ridiculously good rebounding team (thanks to the Unfrozen Cave Man Basketball Playing Twins) while Texas is merely above average, so big advantage there for Stanford. Everything else is even in this game, so whichever team mitigates the other's advantage (or exploits its own) will win.
  • Kansas doesn't normally take a whole bunch of three point attempts (only 29.4% of their shots this year have come from distance), but they're pretty good at hitting the ones they do take. Villanova's opponents take tons of threes (40.1% of the field goals attempted against the Wildcats have been from outside), and they're not great at defending the three point shot. Look for Kansas to shoot a whole lot more threes than normal, and hit more than 40% of them. Oh yeah, and Kansas is better than Villanova at just about everything else.
  • The exact opposite is true in the MSU-Memphis game. The Spartans don't shoot many threes (or make a large percentage of the ones they take), and Memphis doesn't allow their opponents to take many threes (and they hit even fewer). If Michigan State hits three pointers, they will win the game; everything else is close.

No comments: