Upsets on the First Two Days of the Tournament

The first two days of the NCAA tournament provided a number of upsets. Two 15-seeds, one 13-seed, two 12-seeds and two 11-seeds all won, along with one 16-seed who may have won without the hindrance of the refs. Most commentators seem to be describing the upsets as being due to more general parity. That parity has been evident in the last few years with Butler's multiple trips to the Final Four, along with VCU's and George Mason's runs as well. However, an additional factor likely on display this year was the overall quality of the lower seeded teams. In particular, the usual bottom feeders that inhabit the 15 and 16 seed lines looked much better than usual. I think part of the reason was that many of these weaker conferences put their best team forward. In past years, many of these leagues would see their top team get knocked off in the conference tournament. The conferences ranked 20^th, 21^st and 23^rd – 27^th by RPI all sent their regular season champion. Of the higher ranked conferences only the Sun Belt sent a clearly inferior team that belonged on the 16-seed line, Western Kentucky. A couple of them sent teams that weren't their strongest, Detroit from the Horizon and Loyola from the MAAC, but those conferences sent teams that were at least in their top four teams. Add it all up and there was a logjam of teams that typically would have been 13 and 14 seeds getting pushed down to the 15-seed line. If you have enough good teams, some of them will pull off the upset.

Of course this might not be a sufficient excuse for Mizzou, as the one 15-seed that seemed like a typical 15-seed was Norfolk State.

2012 96-Team NCAA Basketball Field

As I did last year, and the year before I tried to come up with what my projection would be for the NCAA field if it was 96 teams instead of 68. One assumption I make is that they will take no losing teams. Two years ago this was a strong assumption, but this year it probably would not have made a difference. Unlike in last year's projection I will not assume that all of the NIT at-large teams will make it. While there were a lot of automatic bids to the NIT, like in years past, the different thing this year was that there were a lot of automatic bids who would have made the NIT without the automatic bids (Washington, Drexel, Nevada …). What that means is that we cannot assume that all of the NIT at-large bids would have made the NCAA tournament. In particular, it is hard to imagine Iowa making a 96-team NCAA tournament. However, being a large university with a large fan base who could host an NIT game made them a good choice for the NIT. I was also skeptical of Stanford, but they received a 3-seed so maybe my view of them is probably lower than that of the people making the decisions.

Clear additions to a 96-team field: Washington, Tennessee, Arizona, Seton Hall, St. Joe's, Mississippi, Miami, Dayton, Oregon, Drexel, Oral Roberts, Northwestern, Mississippi St., Akron, Nevada, Marshall

Likely additions: La Salle, Middle Tennessee, UMass, Central Florida

Bubble teams: Cleveland St., Northern Iowa, LSU, Valparaiso (The Horizon was a pretty highly rated conference so I think Cleveland St. and Valpo would have been selected)

Last 4 in: UMass, Bucknell, Stanford, Minnesota

Last 4 out: George Mason, Weber St., Princeton, Buffalo

I ended up taking all of the at-large teams from the NIT except Iowa (RPI in the 120s) plus two additional auto bids from the NIT that had 7 or 8 seeds (Valparaiso and Bucknell)

Projected 2011 NCAA Basketball Field

As I did last year, I tried to come up with what my projection would be for the NCAA field if it was 96 teams instead of 68. One assumption I make is that they will take no losing teams. Last year this was a strong assumption, but this year it probably would not have made a difference. I will assume that all of the at-large NIT teams would have made it, plus Missouri St. Since there were 13 automatic bids for the NIT that still leaves eight additional spots to come up with. Wisconsin-Milwaukee was an automatic bid and a five-seed, while Harvard was an at-large and a six-seed. However, making Milwaukee a five-seed meant that they were playing Northwestern which means a lot less travel. I will thus not assume that they would have been selected (see below).

My final eight in order of selection:

Marshall: should be in

Maryland: should be in

Southern Miss: should be in

Minnesota: overall ok record but finished the year 1-9

College of Charleston: good RPI, good record and won regular season conference title

Hofstra: worse RPI than CAA-mate Drexel, but finished higher in conference

Wisconsin-Milwaukee: a UAB-type pick, regular season champion from a highly-rated conference (at least at the 96-team level). The high NIT seed is indicative of strong assessments of this team.

Baylor: Not a good record but from a big-time conference

Last 4 out: Central Florida, Drexel, Iona and Mississippi St.

Next 4 out: Valparaiso, JMU, Kent St. and Tulsa

Why was UAB Selected?

I was preparing my annual 96-team tournament field. In trying to determine the last couple of teams, I kept coming back to the question of why the selection committee picked UAB. I think UAB had three things going for it:

1. High RPI
   
2. Regular-season conference champion
   
3. Played in a highly rated conference
   

In particular, the fact that it won the conference championship of Conference USA, which overall had a very good RPI, was very important to the team's selection. Also, the conference did not have any obvious other at-large teams to select. I think everyone has only focused on the high RPI, but I think the latter two elements actually mattered more and will be important to my analysis of an extended 96-team tournament.

Clemson and UAB are the same

Clemson and UAB play tonight in one of the First Four games of the tournament. As I said in my previous post, there has been a lot of whining about UAB's selection. However, there has been almost nothing said about Clemson's inclusion, even though when you look at their profiles they look very similar. UAB has the better RPI and the better record. They both beat one marginal tournament team (VCU for UAB, Florida St. for Clemson). Clemson beat 6 NIT or NIT-bubble teams, UAB beat 7. Clemson had more losses to good teams, but neither of them had any real bad losses. If you remove Clemson's three losses to UNC, they lost the same number of games and the profile of teams they lost to look amazingly similar. Heck, they both lost to Duke on the road.

It seems like the only reason to include Clemson is that they beat Virginia Tech and Virginia Tech beat Duke. Virginia Tech's win over Duke might be reason to include them in the tournament, but it seems like a bad reason to include Clemson.

Must be that Clemson plays in the ACC and UAB is in Conference USA.

2011 Bracket Announcement and ESPN

The wailing and gnashing of teeth from ESPN over the inclusion of UAB and VCU over Colorado and Virginia Tech is quite annoying.

Hubert Davis talked incessantly about the Eye Test, which near as I could figure meant: (i) Teams he saw play, (ii) Teams from big conferences and (iii) Teams with big-name recruits. Pretty much teams from the BCS conferences.

Digger Phelps talked about getting things done in the conference season. Of course the big-conference teams will be the only ones who can beat top teams in the conference portion of the season. Also it seems a little strange to talk about teams that were denied after getting the job done in the conference season when they were 8-8 in conference (Colorado).

I am half-convinced that Jay Bilas would be perfectly happy if they just took the 68 best teams from the BCS conferences. It is particularly annoying when he complains about how worthless RPI is and then talks about how poor a team's record is against the RPI Top 50.

Dick Vitale must have three or four rants prepared and then picks the one that makes the most sense given the bracket that is announced. In the past he has railed against the exclusion of the small schools. This year he complained about the exclusion of Colorado and VPI.

The whole program had a tendency to make "My father can beat up your little brother" arguments, i.e. focusing on the positives of Colorado and the negatives of VCU and UAB. They posted a graphic on VCU that only had the negatives, such as their worst loss against Georgia St. Meanwhile there was no mention of the fact that in addition to playing a weak non-conference schedule, Colorado played most of those games at home.

This all should not necessarily mean that the committee got it right. I personally would have taken St. Mary's, VPI and Colorado over UAB, Clemson and Georgia, but this number of changes is not much different than my usual preferences.

What a Hypothetical 96-Team Field for the NCAA Tournament Would Look Like

The NCAA seems pretty insistent on expanding the NCAA Tournament to 96-Teams. Just like with last year's tournament, I wanted to see what a 96-team field this year would look like. I assume that the teams who made the actual tournament (65), plus all of the at-large teams for the NIT (24), plus any automatic bids for the NIT who had a higher seed than the lowest at-large team (1-Kent St.) would all make it. That leaves six slots. I will primarily use RPI as a measure of the best teams, though I will actually examine the records of the teams and generally teams from big conferences get in with lower RPI's. A dubious assumption I will also make is to also assume that any team with a losing record will not get in. Using those criteria I would say that the additional six teams would be Arizona, VCU, Marshall, Miami, SLU and Alabama. Based on my selections every BCS conference team with a strictly winning record, except an ineligible USC, would have made the tournament.

Here is a full bracket for a 96-team tournament from the Washington Post. The only differences between mine and his, other than 1-bid conferences, are that he includes Georgia, South Carolina, Charlotte and La Tech, and I include Ohio, Houston, Alabama and Nevada. I had excluded Georgia and South Carolina because they had losing records. Ohio and Houston won automatic bids; Charlotte continued its tail-spin and Nevada beat LaTech after the other story was written. So the differences can be explained by either one assumption (including losing teams) or when the selections were made.

Playoff Creep II

Mid-Majority had a post a week ago discussing expansion of the NCAA Tournament to 96 Teams (H/T CAAHoops). The post talks about how it will not cheapen the regular season. As I explained before, the problem is that the majority of the additional bids will still go to the BCS Conference teams. I do not see how expanding the tournament will make the regular season be more important when all it will take is a winning record for the big conference teams. Also it says that the seeds will matter. The seed difference for 7 and 8 versus 9 and 10 will matter a lot more. However, the upper seeds will matter a lot less. The difference between a 1-seed and a 5-seed is pretty important right now, as a 5-seed faces a formidable opponent in the first game, while the 1-seed does not. But with an extra round the 1-seeds will be facing much better competition. The 5-seeds could face a 12-seed or a 21-seed, while the 1-seed will face either a 16-seed or a 17-seed. Not much of a difference there.

I think the better first step is to start awarding the at-large spots to teams from more conferences. Whether there are 65 or 96 teams, the key to making the regular season matter is to have a closer to uniform distribution of at-large bids.

Playoff Creep

One of my biggest concerns about a college-football playoff is the problem of playoff creep, i.e. that the number of teams would increase over time. A four-team or an eight-team playoff might end up with the best team winning fairly often, but a 16, 24 or 32 team playoff could have a very low correlation between the best teams and the championship teams. However, there is always going to be a debate about who should be the last team in, so there will always be pressure to include more teams. As further evidence of my point, the NCAA basketball tournament is considering increasing the number of teams from 65 to 96. As a fan of William and Mary, I probably should applaud this as it may be the only chance for my team to finally get to the NCAA tournament. Well, maybe they won't need help this year. Regardless, I think this proposal would have a huge negative effect on the importance of the regular season.

The four key effects on the importance of the regular season:

1) The big conference teams will be able to get in with a lower number of wins. Assume that all of the teams that made the NCAA tournament plus all of the NIT at-large teams plus NIT automatic bid team Davidson would have made a 96-team tournament in 2008. Then add in three additional teams, which using the NCAA's methodology, I guess to be Cincinnati, Houston and Nevada. That would mean the seven BCS conferences would get the following number of teams into the playoffs:

ACC: 9 of 12

Big 10: 9 of 11

Big 12: 9 of 12

Big East: 11 of 16

Pac 10: 7 of 10

SEC: 7 of 12

Only five BCS conference teams with a winning record would have missed out. Three of those teams were in the relatively weaker SEC (Ole Miss, Vanderbilt and Alabama), and the other two were barely above .500, Seton Hall (16-15) and NC State (16-14).

2) The potential 1, 2 and 3 seeds will have less to play for. The key gains for being a 1 seed as opposed to a 2 seed are that you avoid playing the likely other best teams in the country until the Final Four (other 1 seeds) and that you get an easier game in the first round. There has never been a 16 seed to beat a 1 seed and only four times has a 15 beaten a 2. Part of the reason is that the 1 and 2 seeds should be the best teams in the country, but another reason is that the 15 and 16 seeds are substantially worse than even the 14 and 13 seeds. There is a tremendous drop-off in quality of teams moving from 13 to 14 to 15 to 16. The last upset by a 15 seed was in 2001, and one reason for that is because the committee has gotten better at seeding the low seeds. The top teams get a real advantage by getting a top seed, playing a much weaker opponent and it gives them a little incentive to play well during the season. However, if the tournament expands to 96 teams, the 1 seeds will play the winner of the 16-17 game, which will now be a much better team. Last year, it would have included teams such as Kansas State, Kentucky and Tulsa. The old 15 and 16 seeds would now be 23 and 24 seeds and likely out of the tournament by the time the top teams play. The difference between being a 1 or 6 seed would drop considerably.

3) The mid majors would actually increase their at large bids. My projections of last year's tournament would have added at-large bids for the WCC, CAA, MAAC, Southern and MVC (2 teams) conferences. Those conferences had no at large bids last year. The mid-majors' regular season might actually influence their likelihood of playing in the tournament.

4) There would be a big difference between being an 8 seed and being a 9 seed, since the 8 seeds would have a bye and the 9 seeds would not. Teams at this borderline would have an incentive to win their last games to avoid an extra game.

If the key factor is making the regular season meaningful, reasons 1 and 2 argue against increasing, while 3 and 4 would argue in favor of increasing. It seems to me that the first two reasons are more important than the last two. Reason 4 only applies to a particular type of middling team. Reason 3 could be lessened if the committee would select a few more mid-majors for a 64-team tournament. In particular, the large conferences, which get the most attention, would have a particularly reduced importance to the regular season.