clock menu more-arrow no yes

Filed under:

March Madness: Good Draw / Bad Draw

New, 15 comments

As we inch closer to Selection Sunday, it is time to use some math to see which potential draws might be good or not so good.

NCAA Basketball: NCAA Tournament-South Regional Practice Jamie Rhodes-USA TODAY Sports

I have always been a sucker when in comes to “bracketology.” In fact, for several years in a row I have gone through my own exercise of ranking the 68 teams and placing them into regions on the eve of Selection Sunday, the most holy day of the college sports calendar. That said, I honestly don’t see much value in bracketology starting in October or November, or frankly, even in February. Far too much can change and we really only have enough data to evaluate teams until they have played 30 games or so.

But, it isn’t February any more. It is now March. So, it’s time to starting taking a look at what the brackets might look like and how they might shake out.

One thing that I have learned over the years is that when it comes to predicting the results of the NCAA tournament, the variance that we see (re: upsets) follows the same trends that we see in the regular season when it comes to point spreads. In other words, by just looking at the point spreads, we can predict the rate at which (for example) the No. 12 seed will upset the No. 5 seed.

In case you don’t believe me, below I show the historical correlation between the win percentage as predicted by the Vegas line as compared to the actual rate that the higher (favored) seed wins.

Another thing that I have learned is that Kenpom efficiencies do an excellent job of predicting point spreads. If you put these two pieces of information together, it tells us that Kenpom efficiency data can give us insight into how much Madness to expect March and where it might occur.

Based on historical data, I can calculate both the mean and standard deviation of the adjusted efficiency margin for each seed, one to sixteen. Then, it is straightforward to plot the current adjusted efficiency margin of each team that is expected to make the field of sixty eight as a function of their projected seed in order to get a feel for their relative strength.

While the brackets will not be finalized until the evening of the 15th, there is already a lot of information that we can glean from the brackets they are being proposed by the various amateur “bracketolgists” out there. I find that the “Bracket Matrix” website is a good tool to get an overview of where teams might wind up on Selection Sunday. Essentially, this website provides a crowd-sourced average of the seed that each team in likely to receive. While note perfect, it provides a nice point of reference for an initial analysis of the tournament.

So, based on the most recent set of Kenpom and bracket matrix data, here is a snap shot of the projected top eight seeds, followed by a snap shot of the bottom eight seeds.

In both graphs, the large blue dot is the mean for that seed and the standard deviation is shown using error bars. The orange dots are the positions of the current estimated tournament field. We can draw a lot of initial conclusions from these data plots.

1) The top of the field is very weak

People have been saying this for months, and the data backs them up. Just looking at the top four seed lines, I count only three teams that are historically above average for their currently projected seed: Kansas, Duke, and MSU.

The projected No. 2 seeds and No. 4 seeds, in particular, look VERY bad. No matter where they wind up, teams like Maryland, Seton Hall, Florida State, Kentucky, and especially Auburn look ripe for a first or second round upset. The bulk of the No. 1 seeds and No. 3 seed are simply below average.

2) The rest of the field, however, looks fairly normal

From the No. 5 seed line down, in most cases there is a balance of teams above and below the mean. Notably, the No. 7 seeds all look pretty strong, while the No. 8 seeds and No. 10 seeds look weak. Also notable is that the No. 15 seeds and No. 16 seeds almost look a bit above average. Could there be another UMBC in that bunch? Would a potential Gonzaga-Eastern Washington match-up in Spokane get a little dicey? Maybe...

3) As for MSU, maybe a No. 3 seed isn’t so bad

While the chart above shows MSU as a No. 4 seed, I think that a lot of sites are starting to bump the Spartans up to the three line. I also think that it is certainly possible that MSU could win up on the two line if they were to win the Big Ten Tournament. But, it is possible that a No. 3 seed could provide a more clear path to a Final Four.

To understand why, let’s look at the possible brackets. If MSU is a No. 3 seed, they would match up against a No. 14 seed in the first round. This year, the No. 14 seeds look a bit weak, while the No. 15 seeds (the first round opponent for a No. 2 seed) look relatively strong.

As potential second round match-ups, the only potential above average No. 6 seed is Michigan, who will not be in the same half of MSU bracket (as MSU has already played them twice this year, unlike Minnesota last year). The other No. 6 seeds are below average. In contrast, the potential No. 7 seed teams (the likely opponent of a No. 2 seed) are all strong.

Moving onto the possible Sweet 16 match-ups, a No. 3 seed would most likely face a No. 2 seed, and as stated above the No. 2 seeds, with the exception of Dayton, are all very below average. I would take a match-up with Florida State in a heart-beat.

Furthermore, if we assume that Duke holds their position as a No. 3-seed as well, MSU could not be in their region. Based on Kenpom data, and MSU’s previous run-in with Devils this year, I am OK with that.

4) There are a few other teams that it would pay to avoid

Based on this analysis, I would not particularly like to face Arizona, Houston, West Virginia, or Texas Tech. All of those teams are on a path to get under-seeded, based on this data. That said, Ohio State and Michigan also looked to be poised to be dangerous as well. But, MSU is unlikely to cross path’s with either of those two teams.

That all said, there is two other question that bears asking:

5) How close to home will MSU be in the first two rounds?

MSU would prefer to play in Cleveland on Friday and Sunday. But, it is unclear if that pod will be available. Only two teams from each of the top sixteen will be placed there. Dayton will almost certainly wind up there and will almost certainly be seeded above MSU. That leaves only one slot.

As we look at the other teams in the top four seed lines, there is not quite as much competition as there is in most years. Kansas and Baylor are both likely headed for Omaha. Gonzaga and San Diego State are headed for Spokane and Sacramento. Duke and Maryland would slot into Greensboro, while Villanova and Seton Hall are likely headed to Albany. Florida State is a natural fit for Tampa, while Creighton will likely go to St. Louis.

If Oregon makes it to the No. 4 seed line, they will almost certainly be placed in Spokane. The lowest ranked remaining No. 4 seed will get sent to Sacramento. So, this leaves one slot each available in the cities of Cleveland, St. Louis, and Tampa. As we look at the teams remaining on the top four lines, MSU’s biggest competition for that last slot in Cleveland are Louisville and Kentucky.

Both Louisville and Kentucky are essentially equally close to St. Louis, so the highest ranked of those two will likely be slotted there. MSU is significantly closer to Cleveland, so the committee is going to want to send them there. Basically, as long as MSU ends the season ranked (in the eyes of the committee) higher than either Kentucky or Louisville, they should get that slot. If MSU is on the No. 3 seed line or above, this is virtually guaranteed. If not, they are likely bound for Tampa or Sacramento. If Auburn remains on the No. 4 seed line, they would prefer to go to Tampa. So, that could impact the overall calculus as well.

6) Which region is MSU likely to be placed into?

As a general rule, the committee is supposed to apply the “s-curve” in the bracketing of the teams, such that (for example) the strongest No. 1 seed is matched up against the weakest No. 2 seed in the same bracket as the strongest No. 3 seed and weakest No. 4 seed. However, the committee also tries to keep as many teams as possible close to home to minimize travel time for both the players and fans. In recent years, geography trumps the s-curve almost every time.

In this case, I think that the overall situation is still a bit too volatile to make any clear prediction. It seems clear that Kansas is going to the Midwest (Indianapolis), Baylor is going to the South (Houston) and either Gonzaga or San Diego State is staying out West (L.A.). The question is what to do about the East region (New York City).

Most, if not all bracketologists have San Diego State in the East, but I have my doubts. It just seems so natural to put both Gonzaga and San Diego State out West (with one of those teams as the No. 2 seed, likely the Aztecs), that I have a feeling that is how this will shake out. As for the East, Dayton seems to be the most likely candidate to take that last No. 1 seed. The No. 2 seed could then be the Big East champ, perhaps Seton Hall.

With the remaining No. 2 seeds, Florida State is a more natural fit in the South, which then leaves the Midwest No. 2 seed. If MSU were to win both the regular season Big Ten Title and the Big Ten Tournament, that No. 2 seed in Indy would look pretty good. The only problem is that Kansas might be waiting in the Regional Final. That would be a tough draw, based on Kenpom.

If MSU is a No. 3 seed, pretty much any region is in play. If Duke is also on the No. 3 seed line, they are likely to be in the East. So, I would guess any of the three other regions would be equally likely, but Midwest, again is perhaps most likely, due simply to geography.

While it is almost conference tournament time, there is still a lot that can happen in the final weeks of the season. I recommend that everyone sit back and enjoy the ride.