(Apologies if this belongs elsewhere, but I haven't seen this analysis done yet).
At 11-3, with a half-game lead on Staee and four games remaining, Michigan is obviously in the driver's seat for the B1G basketball title. Using the game predictions from KenPom's site, I've done a quick probability analysis to see what the odds are that there's a banner to be hung.
First, Michigan's expected record, along with a percent chance:
15-3 | 36% |
14-4 | 43% |
13-5 | 18% |
12-6 | 3.1% |
11-7 | 0.18% |
(I used two significant figures, since there were two in the KenPom data; obviously, they won't add to exactly 100%).
Here's Staee's expected record:
14-4 | 16% |
13-5 | 49% |
12-6 | 32% |
11-7 | 3.7% |
Iowa:
13-5 | 15% |
12-6 | 40% |
11-7 | 33% |
10-8 or worse | 12% |
Wisky:
13-5 | 37% |
12-6 | 44% |
11-7 | 17% |
10-8 or worse | 2.4% |
Ohio:
12-6 | 33% |
11-7 | 44% |
10-8 or worse | 23% |
Nebraska:
12-6 | 5.7% |
11-7 | 26% |
10-8 or worse | 68% |
Put it all together, and you get the following possibilities (all chances here are conditional -- e.g., each line should add up to 100% within the limits of rounding and significant figures):
Record | Outright Title | Shared Title | No Title |
---|---|---|---|
15-3 | 100% | ||
14-4 | 84% | 16% | |
13-5 | 19% | 65% | 16% |
12-6 | 0.20% | 19% | 81% |
11-7 | < 0.01% | 0.20% | > 99% |
When you factor in the chances that Michigan achieves each of these records (from the first table), and add it all up, and there is a 75% chance of an outright title, a 19% chance of a shared title, and a 6% chance of being bannerless. (Coincidentally, I coded up a simulation using the same KenPom percentages, ran it 100 times, and got at least a share of the title exactly 94 times).
Long story short, even with a single loss, the odds are still in our favor to win the title outright, since KenPom doesn't think Staee is likely to run the table, and 2-2 down the stretch is likely to be enough to secure a share of the title. Like many of you, I never would have predicted this in December.
Go Blue!