clock menu more-arrow no yes mobile

Filed under:

I Don't Really Care What You Do with Your Slide Rule

Earlier this week, Dave Sez broke out the log5 predictions for the ACC, the latest fad in basketball number crunching. This in turn prompted Chad Orzel to complain about the overstatisticification of basketball blogging. I'm not particularly interested in log5 type analysis myself, but I am interested in things that give the Tar Heels better than 50-50 odds to win the ACC championship. That seemed high to me even at the time, not considering the subsequent loss to Georgia Tech. But the numbers had been crunched assuming UNC would be the top seed - far from a sure thing, although Clemson currently leads Virginia tech by five - so I wondered, how does the seeding effect the numbers?

So I ran an experiment. I punched the log5 formulas into an Excel spreadsheet using the same initial data Dave's reader had. I switched Duke and Maryland's position, as the Blue Devils can't finish higher than sixth, and then ran the numbers a couple of different ways.

As Dave Sez had it, but with Maryland and Duke swapped:

North Carolina (1)   0.5215
Maryland (5)         0.1046
Virginia Tech (2)    0.0458
Boston College (4)   0.0225
Virginia (3)         0.0154
If all the favorites win out:
North Carolina (2)   0.5082
Maryland (5)         0.1543
Virginia Tech (1)    0.0529
Boston College (3)   0.0210
Virginia (4)         0.0189
If the unthinkable leads to a Blue Devil victory:
North Carolina (5)   0.4800
Maryland (4)         0.0979
Virginia Tech (2)    0.0502
Boston College (1)   0.0309
Virginia (3)         0.0171
The wacky complications that put the Heels third:
Virginia Tech (1)    0.0556
North Carolina (3)   0.4922
Maryland (5)         0.1502
Boston College (4)   0.0358
Virginia (2)         0.0184
Upsets all around:
North Carolina (5)   0.4885
Maryland (4)         0.1019
Virginia Tech (2)    0.0493
Boston College (3)   0.0297
Virginia (1)         0.0176
So it turns out seeding doesn't have a significant effect on the odds of winning - the numbers are much more heavily influenced by Ken pomeroy's formulas. The odds are only as good as one's faith that UNC was the best team in the country last week, and I'm sure not about to buy that. Perhaps the Heels shouldn't lay down against Duke today, you think?