Saturday, December 8, 2007

'Bama highlights, and Hoya slow starts

I finally got around to putting together some highlights from the Alabama game (link). Enjoy.


I also noticed a couple of threads on HoyaTalk discussing KenPom's schedule prediction page and game plan page for Georgetown, and so I'll take this opportunity to make a few of points:
  • Sample size! The season is only 6 games in, so worrying that Pomeroy projects an 11-7 record in the Big East is a bit premature. His predictions are updated almost every night, based upon games played that day, and expect a lot of movement through January before the predictions stabilize.
  • Schedule strength! Frankly, outside of the game at Memphis, this year's OOC for the Hoyas is going to be much weaker than last year. In fact, Memphis may be the only NCAA tournament team Georgetown will play before the BE schedule kicks in; last season, the Hoyas played 5 NCAA-bound teams by December 9th. Of course, Georgetown was 6-3 at that point last year, while they sit 6-0 so far this season.
  • As of this writing, Georgetown is ranked 80th in RPI and 41st by Pomeroy's ratings, which is causing a bit of consternation to some on Hoyatalk. While certainly no guarantee of future success, last year's slow start offers some solace that the Hoyas will improve upon those ranks as the season progresses. But, can we compare the two starts quantitatively? Sure, so long as you recognize a couple of caveats:
    • Sample size! I know, I already said that, but remember that analysis for this season is a bit premature.
    • Context! I've got numbers for last season, but they are weighted by how the teams finished the season. Using the full season data for just early season games may be a bit misleading, especially if a team lost or got back an important player after playing Georgetown.
So how do you compare the two starts? As I discussed several times last year, you can simply compare the predicted score to the actual score for each game to determine performance. One way to do this is to use the Vegas point spread for each game, versus the final score. I'd think this would work very well, but I don't actually keep track of the Vegas line, so I'll use KenPom's data instead. Here, you use each team's offensive and defensive efficiencies, adjusted for competition, to determine expected point totals for each team.

Here are the expected point spreads and actual outcomes for last season and this season, through Dec 8th:


2006-7 Season


Final Rank
Site Predict.
Margin


Result
GU Opp
Diff. from
predict.
11-Nov-06 Hartford 253
H 34
W
69 59
-24
15-Nov-06 Vanderbilt 35
A 5
W
86 70
11
19-Nov-06 Old Dominion 73
H 16
L
62 75
-29
22-Nov-06 Fairfield 198
A 21
W
73 60
-8
27-Nov-06 Ball St. 195
H 27
W
69 54
-12
29-Nov-06 Oregon 18
H 10
L
50 57
-17
2-Dec-06 Duke 11
A 0
L
52 61
-9
5-Dec-06 James Madison 263
H 34
W
89 53
2


2007-8 Season


Current Rank
Site Predict.
Margin

Result GU Opp
Diff. from
predict.
10-Nov-07 William & Mary 248
H 26
W 68 53
-11
15-Nov-07 Michigan 129
H 15
W 74 52
7
21-Nov-07 Ball St. 191
A 12
W 57 48
-3
28-Nov-07 Old Dominion 100
A 5
W 66 48
13
1-Dec-07 Fairfield 219
H 19
W 61 49
-7
5-Dec-07 Alabama 138
A 9
W 70 60
1


In Georgetown's first 8 games last season, the team played up to or above expectations only twice (and the James Madison game was essentially a wash). The team had two horrible games: ODU, as everyone knows, but also the season opener against Hartford.

Now keeping in mind that this season's number will change as we learn more about all the teams, Georgetown's worst game is the season opener against William & Mary. The ODU and Michigan games both ended up well above the spread.

In essence, while the season is young and there's a lot of basketball yet to be played, there's no reason to think that the Hoyas' slow start is any different than last year.

No comments:

Post a Comment