February 27, 2007
Break Out The Slide Rules
A regular reader (and college professor) just figured out who's going to win the ACC Tournament. He took Pomeroy's ratings, applied them to the Log5 prediction method and PRESTO! - we have a champeen!
Actually, it just spits out the percent chance of each team winning, but going by the seedings if things ended today (and he put UNC as the 1 see, VT as the 2 and UVA as the 3), Carolina would be the overwhelming favorite. While their rating isn't that much higher than Duke's, they have a three times greater chance of winning the ACC Tourney. Check the results:
| Fri | Sat | Sun | Champion | |
| North Carolina | 1.00000 | 0.85970 | 0.63850 | 0.51400 |
| Duke | 0.93910 | 0.70350 | 0.26100 | 0.17110 |
| Maryland | 0.91280 | 0.68110 | 0.43280 | 0.14700 |
| Georgia Tech | 0.82730 | 0.45320 | 0.21980 | 0.05710 |
| Virginia Tech | 1.00000 | 0.51200 | 0.22560 | 0.05130 |
| Boston College | 1.00000 | 0.28660 | 0.05180 | 0.02060 |
| Virginia | 1.00000 | 0.29980 | 0.11270 | 0.01810 |
| Clemson | 0.54830 | 0.08260 | 0.02980 | 0.01300 |
| Florida St. | 0.45170 | 0.05760 | 0.01850 | 0.00720 |
| North Carolina St. | 0.17270 | 0.03480 | 0.00630 | 0.00050 |
| Wake Forest | 0.08720 | 0.01910 | 0.00280 | 0.00020 |
| Miami FL | 0.06090 | 0.00990 | 0.00050 | 0.00010 |
As I pointed out, it's interesting to see just how big a gap the Tar Heels have on the field. The best explanation for this is that according to the Pomeroy ratings, Duke is the second best team followed by Maryland and Georgia Tech. Well, all of those teams will have relatively difficult seeds. VT, UVA and Boston College will make up seeds 2-4 (again, if things ended today), but they have the fifth, sixth and eighth best power ratings.
It'll be interesting to rerun this once the real seeds are known. I have a hunch that either UVA or VT is going to be the top seed, not UNC.
If you want to see the whole calculation and give yourself a nasty flashback to your college math classes, check this PDF.
He also ran similar projections for the Big 12, CAA and Conference USA.
Another interesting way to use these numbers -- note for example that UNC has approximately the same chance of winning it all as VT has of getting to Saturday (or, 1.5x the chance of UVA getting to Saturday). Or, Duke, VT, and GT have about the same chance to make it to Sunday, but Duke has a 3x greater chance to win it all.
| baller wrote: |
| If the season ended today, then VT should be the #1 seed, right? Head to head they beat UNC 2x and UVA. |
Yes. I should have said that in my article. The calculations were done using Pomeroy's ratings as the tiebreakers (I think) instead of the official tiebreakers.
| 72hoo wrote: |
| The flaw in this analysis is that no effect is given to the teams having to play on Thursday. Duke, UMD and GA Tech may be 2nd, 3rd and 4th respectively, but they would have to win 4 games to win the tournament. VA Tech, BC & UVA may be 5th, 6th & 7th, but they would only have to win 3. That's a big difference by the time we get to Sunday. |
Do you mean the effect of fatigue? Because the analysis does factor in all four rounds.
| 72hoo wrote: |
| Right, the fatigue factor you can't model mathmatically. Or the depth factor, which is more important in a tournament setting then in any single game. |
You could try to model fatigue by finding old tournament data and looking at the actual results versus expected. In theory, that is. The problem would be in having accurate power ratings and collecting enough data. Not something I'd want to do, that's for sure, but it's probably doable.
My hunch would be that it's probably not as big a deal as most people would think. If Team A is playing their fourth day in a row against Team B who is playing their third day in a row, both will be somewhat tired. But both will also be bolstered by the excitement of playing in a Championship. You'd assume that Team B is the better team, because they had a bye in the tournament. Better teams are more-often-than-not also deeper. So Team B is usually going to win and the easy excuse is that Team A was tired, but more likely, it's just because the better team won.
The only ACC example is when State made their run to the finals in one of Herb Sendek's first years. They lost in the finals to a much better Duke team after giving up a pretty decent fight.
| tieguy wrote: |
| I think you're thinking of Herb's very first year, when as the 8 (9?) seed he won the Thursday night game, beat Duke on Friday, and an exhausted State lost to Carolina in the final. (There was a similar run late in his tenure, where he did lose to Duke, but I don't think it came from the 8/9 seed like it did Herb's first year.) |
I knew I should have looked that up. I thought it was his first year (Justin Gainey), but I wasn't 100% sure.
BrightCar Car Maintenance Software - Software To Manage Your Car Maintenance Schedule
