Big Ten Tournament Seeding Percentages
This diary is based on all the possible outcomes to the last six Big Ten basketball games and how the outcomes affect the tournament seeding. I went through all 64 possible combinations and weighted each outcome based on the percentages that kenpom.com had projected for each game. Here are some of the results:
Most Likely Outcome for Seeding:
1 |
Ohio State |
2 |
Purdue |
3 |
Wisconsin |
4 |
Illinois |
5 |
Michigan |
6 |
Michigan State |
7 |
Penn State |
8 |
Minnesota |
9 |
Northwestern |
10 |
Iowa |
11 |
Indiana |
This outcome has a 24.9% likelihood of occurring.
Least Likely Outcome for Seeding:
1 |
Ohio State |
2 |
Wisconsin |
3 |
Purdue |
4 |
Michigan State |
5 |
Penn State |
6 |
Illinois |
7 |
Michigan |
8 |
Northwestern |
9 |
Minnesota |
10 |
Iowa |
11 |
Indiana |
This outcome has a 0.04% likelihood of occurring.
Here is the list of each team with their probabilities of seeding. The order is based on the most likely outcome listed above:
Ohio State |
|
1 |
82.15% |
2 |
17.85% |
Purdue |
|
1 |
17.85% |
2 |
79.66% |
3 |
2.49% |
Wisconsin |
|
2 |
2.49% |
3 |
97.51% |
Illinois |
|
4 |
38.03% |
5 |
51.16% |
6 |
7.76% |
7 |
3.05% |
Michigan |
|
4 |
27.97% |
5 |
38.03% |
6 |
22.78% |
7 |
11.22% |
Michigan State |
|
4 |
34.00% |
5 |
6.19% |
6 |
41.08% |
7 |
18.73% |
Penn State |
|
5 |
4.62% |
6 |
28.38% |
7 |
67.00% |
Minnesota |
|
8 |
67.00% |
9 |
33.00% |
Northwestern |
|
8 |
33.00% |
9 |
67.00% |
Iowa |
|
10 |
72.31% |
11 |
27.69% |
Indiana |
|
10 |
27.69% |
11 |
72.31% |
Interesting things to note:
- Michigan State is more likely to get a 7 seed than a 5 seed
- Illinois is the most likely team to get a 4 seed and a 5 seed
- Purdue has a decent shot (17.85%) at getting the 1 seed (they need win against Iowa and have Wisconsin beat Ohio State)
- Michigan has a 66% chance at a first round bye
- Michigan’s most likely 1st round opponent is Illinois
- Michigan has a 6.2% chance at a rematch with Michigan State in the 4/5 game
Obviously this will change after tonight’s game, but the snapshot in time is a nice reference to have. Let me know if you want to know any more specifics about the stats.
What is the data behind the probabilities? What win% probability is each team given in a matchup? Is it just 50-50?
The probabilities are based on what kenpom has for each game. For instance, when I did this, Michigan has a 66% chance to beat MSU on Saturday and OSU has a 79% chance to beat Wisconsin on Sunday.
*NINJA-ed*
The likely percentage of my TV breaking if we lose to sparty...
That MSU 5-seed percentage is fascinating. All the other percentages are binomial or have a normal distribution shape.
What makes it so hard for MSU to get a 5 seed?
I'm sure it's just because of how they fall with the tie breakers. They lose tie breakers with both us and ill meaning if they lose to us on sat, they need ill to lose in order to get the 5 seed which is highly unlikely.
They also lose a tiebreaker with Penn State. In the situations where that arises, they have to go to the second to last "head-to-head" tiebreaker, record vs Iowa, where Penn State wins with a 1-0 record to MSUs 1-1.
Thanks. I knew it was how the tiebreakers fell, I just didn't know the tiebreak scenarios were.
Isn't MSU 1-1 head-to-head against Illinois? What are the tiebreakers that cause MSU to be 6th if Illinois, Michigan and MSU are all 9-9?
If Michigan, Michigan State and Illinois are tied at 9-9, you have to take each teams record against the other two. Michigan and Illinois would have a 2-1 record against the other two and Michigan State would have a 1-3, because they would have lost to Michigan twice. Becuase of their inferior record against the other two teams, MSU would be the 3rd place 9-9 team, Illinois would be the 1st 9-9, based on their head-to-head victory over Michigan and Michigan would be the 2nd 9-9 team.
As a huge statistics nerd, I would love to see how Kenpom computes these percentages. Factor analysis, perhaps?
I'm interested as well. I wonder what kind of database he has to keep all the necessary data for his calculations?
Thanks for putting this together.
Comments