Big East Hoops » Blog Archive » Conference Perspective (the ACC sucks!)

Log5 analysis of future contests is nothing new, but I haven’t seen it applied to the standard debate about conference performance. I applied the technique to the tournament and summed expected wins for each conference on 3/13 — in other words, before the first tip — and today through two rounds. Here are the results:

Conf  #  W-L  Chalk  3/13  3/20  Delta  Champ
P10   6  7-3    9     5.9   8.9   3.0     3%
SEC   5  7-2    7     7.3   9.5   2.2    14%
MWC   2  2-1    2      .8   2.4   1.7     0%
Horz  2  2-1    1     1.1   2.3   1.1     0%
CUSA  1  2-0    3     2.0   2.6    .6     2%
B12   4  5-2    8     8.0   8.5    .5    29%
WAC   2  1-2    1      .7   1.0    .3     0%
BSth  1  1-1    0      .7   1.0    .3     0%
SWAC  1  0-1    0      .0    .0    .0     0%
OVC   1  0-1    0      .0    .0    .0     0%
MAAC  1  0-1    0      .0    .0    .0     0%
NEC   1  0-1    0      .0    .0    .0     0%
SB    1  0-1    0      .0    .0    .0     0%
BSky  1  0-1    0      .1    .0   -.1     0%
MVC   2  2-1    2     2.3   2.2   -.1     0%
Ivy   1  0-1    0      .2    .0   -.2     0%
CAA   2  1-2    0     1.2   1.0   -.2     0%
ASun  1  0-1    0      .2    .0   -.2     0%
Slnd  1  0-1    0      .3    .0   -.3     0%
B10   6  6-5    8     7.5   7.2   -.3     4%
MAC   1  0-1    0      .3    .0   -.3     0%
BE    6  5-4    8     6.9   6.5   -.4     3%
AE    1  0-1    0      .4    .0   -.4     0%
BW    1  0-1    0      .4    .0   -.4     0%
Pat   1  0-1    0      .4    .0   -.4     0%
MCon  1  0-1    0      .6    .0   -.6     0%
SC    1  0-1    0      .7    .0   -.7     0%
A10   2  1-2    0     1.9   1.0   -.9     0%
WCC   1  0-1    0      .9    .0   -.9     0%
ACC   7  6-6   11    12.1   8.8  -3.2    44%

Terms:

Conf, #, and W/L — conference, teams, and tournament record.
Chalk — expected wins based on seeding. Final Four games are ignored (since they’d be between #1 seeds), so there’s only 60 chalk wins available.
3/13 — expected wins based on log5 projections before the tournament started.
3/20 — expected wins based on log5 projections after two rounds.
Delta — Change in log5 projection through two rounds.
Champ — Odds of producing the champion as of 3/20.

The successful conferences are surprising in my eyes. The Pac10 and SEC have clearly done the best — I don’t have to tell you that — and are the only conferences with three teams alive. The Pac10 has already exceeded its log5 expectations, so every win is gravy from here out. UNLV and Butler have represented the Mountain West and Horizon very well. It will be a successful tournament for Memphis if they get one more win.

The Big East suffered too many 1st round defeats and needs Pittsburgh to pick up for everybody else. Victory over UCLA drastically changes these numbers. Georgetown’s favored fairly comfortably to advance one more time before the probably UNC battle. It’s been a bit disappointing for the BEast this year, but it’s not a disaster.

The ACC stands alone as the only notable conference to severely underperform. They deserve some credit for having the bar set so high, but you can’t help but feel that they were overrated by both the committee and my numbers. Still UNC remains as the most likely to win the whole thing (if you can rationalize the rest of this paragraph).

Besides UNC’s 44% odds, the next three contenders for the title are Kansas (21%), Florida (13%), and Texas A&M (8%). Not surprisingly, they are the four teams I predicted would make the final four in my Yahoo! group picks.

I’ll be paying the most attention to Memphis-A&M, where I think A&M will prove superior in San Antonio, Oregon-UNLV, hoping for an upset, and Pitt-UCLA, where with any luck the protege will get the better of his mentor for once in this tournament. (In other words, it’s not looking too good Panther fans.)

Categories: Rankings, Dan'l B, Postseason

10 Responses to “Conference Perspective (the ACC sucks!)”

Big Willie Style wrote a comment on March 20, 2007

If you include the NIT, I bet the Big East is doing even better. I believe we have 3 of the 8 teams remaining in that tournament.

Pete wrote a comment on March 21, 2007

according to KenPom UNC has a 22.21 % chance of winning the tournament… not 44%… I think you messed up.

also, as of 3/20 log5, Florida alone has a 17 percent chance of winning the tournament — so how can the whole SEC only have a 14 %… etc?

Dan'l B wrote a comment on March 21, 2007

I used the ratings posted here, not Ken’s.

Maybe I don’t fully understand what Log5 is, but I thought it simply took its source data from win percentage?

Dan'l B wrote a comment on March 22, 2007

The log5 tables published on kenpom.com are based on win percentages derived from his adjusted ratings rather than raw win percentages. There are differences in the ratings here that have translated into different win percentages and slight differences in each teams’ probabilities.

The biggest reason for a difference between my system and others is that I weigh each game based on how recently they were played. The weights gradually decrease backwards through the season — games in early November have about 20% of the impact on the ratings that games in March have.

I have UNC rated a lot higher than they are most other places, and I suspect that I’m wrong about them. Over four future games that extra difference adds up to the 44% chance I give them and takes a lot away from the others.

My opinion is that log5 has some inaccuracies with the NCAAs because 1-game series make huge swings probable. It’s better suited for the professional playoff systems in MLB or the NBA.

Big Willie Style wrote a comment on March 22, 2007

For example, the Heat proved they were much better than the Mavs, and that was justified by a 7 game series. (=