I have been tinkering with some more numbers related to seeds and Kenpom rankings under the general theory that the Kenpom rankings should be a pretty good indicator of relative strength and weakness of teams in the upcoming NCAA tournament. I have run a few simulations over the past 1-2 weeks and reported out here which brackets look good or bad.
But, today it occurred to me that there is perhaps an easier way to visualize this. Here is what I did. I have Kenpom data back to 2002 and I can use that to specify the usual average and standard deviation of the Kenpom adjusted efficiency margin as a function of seed.
I then went to the bracket matrix website site and pulled the current consensus projected seeds for the entire field. I then plotted the data on the same chart. For the Top 8 seeds, this is the current snapshot. The blue circles are the Kenpom average for that seed and the error bars represent one standard deviation from that mean.
Here are current takeaways:
1) Things look a bit top heavy this year. The Top 3 games (UVA, Gonzaga, and Duke) are all very strong 1s, historically speaking. All the projected 2-seeds are well above average, as are half of the 3-seeds.
2) That said, it would be appear to be very beneficial to be in the same bracket as the weakest 1-seed, who is most likely to be the SEC champ, I would guess. The good news for MSU is that this is likely going to be a site that is a good geographic match for MSU, so the odds are a bit better that this will happen.
3) As for the 3-seeds, it looks like Houston and LSU are the teams to hope for. If MSU can hold onto a 2-seed, our odds are decent to get one of the two in our region, because the committee will almost certainly separate MSU, UofM, and Purdue into different regions. It would not appear to be beneficial to draw Texas Tech, based on this analysis, and if VA Tech jumps up to the 3-seed line, that also seems like a bad draw.
4) Other than VA Tech, the other 4-seeds look relatively weak, while the 5-seeds look fairly average to strong. Marquette in particular looks like an interesting candidate to lose in the 1st round.
5) The 6-seeds and particularly the 7-seeds look strong. If MSU does get a 2-seed, Cincinnati would be best to see in our region. Auburn looks a little scary. Meanwhile, the 8-seeds look a bit weak.
As for the bottom half of the brackets:
6) The 9-seeds look very average
7) The 10-seeds look a little above average, but ironically, MSU has already played (and beaten) 3 of them (Texas, Florida, and Ohio State). I would hope the committee would know this and pair us with NC State, if it comes to that
8) The 11-seeds look average to weak, with St. John's being the weakest. If MSU slips to a 3-seed, this could come into play.
9) The 12- and 13- and 14-seed all look pretty average right now, with Clemson and New Mexico State being two teams to watch out for.
10) The 15-seeds all look above average, with Northern Kentucky looking the best. That team has been paired with MSU in some brackets and seems likely to be MSU's opponent if MSU stays in Columbus. Loyola Chicago (or Montana) seems more likely if MSU winds up in Des Moines.
11) Colgate looks like a really strong 16-seed. You never know....
Now, that all said, college basketball remains very chaotic and unpredictable. I have not gone through enough past brackets to know how well this type of analysis might yield some hints as to how a bracket might shake out. That said, Kenpom data is a pretty good predictor of the odds of certain results and on average, I would expect this type of analysis to be right more than it is wrong. But, the variance is going to be large. This analysis might only be "right" 55% or 60% of the time (if that).
But, I think this is fun, so... enjoy!
But, today it occurred to me that there is perhaps an easier way to visualize this. Here is what I did. I have Kenpom data back to 2002 and I can use that to specify the usual average and standard deviation of the Kenpom adjusted efficiency margin as a function of seed.
I then went to the bracket matrix website site and pulled the current consensus projected seeds for the entire field. I then plotted the data on the same chart. For the Top 8 seeds, this is the current snapshot. The blue circles are the Kenpom average for that seed and the error bars represent one standard deviation from that mean.
Here are current takeaways:
1) Things look a bit top heavy this year. The Top 3 games (UVA, Gonzaga, and Duke) are all very strong 1s, historically speaking. All the projected 2-seeds are well above average, as are half of the 3-seeds.
2) That said, it would be appear to be very beneficial to be in the same bracket as the weakest 1-seed, who is most likely to be the SEC champ, I would guess. The good news for MSU is that this is likely going to be a site that is a good geographic match for MSU, so the odds are a bit better that this will happen.
3) As for the 3-seeds, it looks like Houston and LSU are the teams to hope for. If MSU can hold onto a 2-seed, our odds are decent to get one of the two in our region, because the committee will almost certainly separate MSU, UofM, and Purdue into different regions. It would not appear to be beneficial to draw Texas Tech, based on this analysis, and if VA Tech jumps up to the 3-seed line, that also seems like a bad draw.
4) Other than VA Tech, the other 4-seeds look relatively weak, while the 5-seeds look fairly average to strong. Marquette in particular looks like an interesting candidate to lose in the 1st round.
5) The 6-seeds and particularly the 7-seeds look strong. If MSU does get a 2-seed, Cincinnati would be best to see in our region. Auburn looks a little scary. Meanwhile, the 8-seeds look a bit weak.
As for the bottom half of the brackets:
6) The 9-seeds look very average
7) The 10-seeds look a little above average, but ironically, MSU has already played (and beaten) 3 of them (Texas, Florida, and Ohio State). I would hope the committee would know this and pair us with NC State, if it comes to that
8) The 11-seeds look average to weak, with St. John's being the weakest. If MSU slips to a 3-seed, this could come into play.
9) The 12- and 13- and 14-seed all look pretty average right now, with Clemson and New Mexico State being two teams to watch out for.
10) The 15-seeds all look above average, with Northern Kentucky looking the best. That team has been paired with MSU in some brackets and seems likely to be MSU's opponent if MSU stays in Columbus. Loyola Chicago (or Montana) seems more likely if MSU winds up in Des Moines.
11) Colgate looks like a really strong 16-seed. You never know....
Now, that all said, college basketball remains very chaotic and unpredictable. I have not gone through enough past brackets to know how well this type of analysis might yield some hints as to how a bracket might shake out. That said, Kenpom data is a pretty good predictor of the odds of certain results and on average, I would expect this type of analysis to be right more than it is wrong. But, the variance is going to be large. This analysis might only be "right" 55% or 60% of the time (if that).
But, I think this is fun, so... enjoy!
