(There were a few other threads on this, but I decided that I would create one more. You're welcome.)
OK, I sort of went down a big rabbit hole on something. So much so that I think I need to type it as one of my "blog post" pieces. But, here is the short version, for now:
Based on my analysis of basketball and NCAA tournament for well over a decade now, I have come to a few clear conclusions. They are:
1) The most reliable predictor of the odds that a team wins or loses is the Vegas spread
2) This statement applies equally to the regular season as well as the tournament, as it perfectly explains the rate that (for example) a 15-seed will upset a 2-seed.
3) Kenpom efficiency data is a very good tool to estimate the Vegas spread and/or the odds that in a given match-up, team A will beat team B.
So, if I put all this together, I can (and did) put together a spreadsheet to simulate any arbitrary NCAA tournament bracket. It will essentially tell you the odds that any seed will make the Final Four. My basic analysis says that in a perfectly statistically average bracket (from a Kenpom point of view) the Top 4 seeds will have the following odds to make it to the Final Four:
1-seed: 36%
2-seed: 21%
3-seed: 12%
4-seed: 9%
When I compare my simulation to the actual results over the past 40 NCAA tournaments, it matches pretty well. The actual data says the 1-seed rate is slightly higher (41%) but the other three values are within 1%.
So that's the background. Now, when I plug in the current ESPN projected bracket with Kentucky as the 1-seed and MSU as the 2-seed, and so on, and use the current Kenpom efficiency data, the result strongly supports the ideal that has been discussed today that "MSU wants Kentucky". This looks to be a very favorable bracket for MSU.
In detail, my math suggests that MSU would have a 42% chance to make the Final Four, which is double the normal odds for a 2-seed. UK's odds are only 26%. As I look at the bracket, the reasons for this become a bit more clear. MSU's current Kenpom metrics are better than UK, and well above average for a 2-seed. On the other hand, UK would be a slightly below average 1-seed.
But, there are some subtleties that actually make this bracket even more favorable than it appears on its face. Specially, the 7- and 10-seeds (Washington and VCU) are both "below average" (meaning their current Kenpom margin is lower than the historical average for those seeds.) Also, Houston is a "below average" 3-seed. So, it would seem that MSU would have a slightly easier path to the Regional Final.
On the other side, Virginia Tech as a 5-seed looks very undervalued based on current Kenpom ranking. Even crazier is that Bucknell is a strong 16-seed. Kentucky's path in this bracket in more difficult than a typical 1-seed.
Perhaps the weirdest part of all of this is that if I swap MSU and UK, MSU's odds as the 1-seed actually go down by a few percentage points. As a general rule, you always want the 1-seed over the 2-seed, but based on this analysis, MSU is actually better off as the 2. Crazy.
Now, please feel free to add any caveat you would like about the impact of Ward's injury, the role of individual match-ups, etc. That is all valid. But, I cannot really calculate that, so here we are. So, I am going to continue to play with this new toy that I created and I will try to update the board when I have something interesting to say.
Enjoy!
OK, I sort of went down a big rabbit hole on something. So much so that I think I need to type it as one of my "blog post" pieces. But, here is the short version, for now:
Based on my analysis of basketball and NCAA tournament for well over a decade now, I have come to a few clear conclusions. They are:
1) The most reliable predictor of the odds that a team wins or loses is the Vegas spread
2) This statement applies equally to the regular season as well as the tournament, as it perfectly explains the rate that (for example) a 15-seed will upset a 2-seed.
3) Kenpom efficiency data is a very good tool to estimate the Vegas spread and/or the odds that in a given match-up, team A will beat team B.
So, if I put all this together, I can (and did) put together a spreadsheet to simulate any arbitrary NCAA tournament bracket. It will essentially tell you the odds that any seed will make the Final Four. My basic analysis says that in a perfectly statistically average bracket (from a Kenpom point of view) the Top 4 seeds will have the following odds to make it to the Final Four:
1-seed: 36%
2-seed: 21%
3-seed: 12%
4-seed: 9%
When I compare my simulation to the actual results over the past 40 NCAA tournaments, it matches pretty well. The actual data says the 1-seed rate is slightly higher (41%) but the other three values are within 1%.
So that's the background. Now, when I plug in the current ESPN projected bracket with Kentucky as the 1-seed and MSU as the 2-seed, and so on, and use the current Kenpom efficiency data, the result strongly supports the ideal that has been discussed today that "MSU wants Kentucky". This looks to be a very favorable bracket for MSU.
In detail, my math suggests that MSU would have a 42% chance to make the Final Four, which is double the normal odds for a 2-seed. UK's odds are only 26%. As I look at the bracket, the reasons for this become a bit more clear. MSU's current Kenpom metrics are better than UK, and well above average for a 2-seed. On the other hand, UK would be a slightly below average 1-seed.
But, there are some subtleties that actually make this bracket even more favorable than it appears on its face. Specially, the 7- and 10-seeds (Washington and VCU) are both "below average" (meaning their current Kenpom margin is lower than the historical average for those seeds.) Also, Houston is a "below average" 3-seed. So, it would seem that MSU would have a slightly easier path to the Regional Final.
On the other side, Virginia Tech as a 5-seed looks very undervalued based on current Kenpom ranking. Even crazier is that Bucknell is a strong 16-seed. Kentucky's path in this bracket in more difficult than a typical 1-seed.
Perhaps the weirdest part of all of this is that if I swap MSU and UK, MSU's odds as the 1-seed actually go down by a few percentage points. As a general rule, you always want the 1-seed over the 2-seed, but based on this analysis, MSU is actually better off as the 2. Crazy.
Now, please feel free to add any caveat you would like about the impact of Ward's injury, the role of individual match-ups, etc. That is all valid. But, I cannot really calculate that, so here we are. So, I am going to continue to play with this new toy that I created and I will try to update the board when I have something interesting to say.
Enjoy!
