Cracketology: Bracket Brokedown

Photo from gomarquette.com

Every year after the NCAA tournament bracket comes out, I look back to see how the Selection Committee did. Because their work reflects which teams are selected and where they are sent, they technically can't get it wrong. But this year, as soon as Selection Committee chairman Bubba Cunningham appeared with his Vice Chair Keith Gill, it was clear he needed cover for something the Selection Committee got wrong. We'll start with the most glaring error and work our way down from there.

West Virginia got snubbed

Buckle up, this is going to be a long one.

There is no other way to say it. The Mountaineers should have been in this field. This isn't just me saying it. On bracketmatrix.com, there were 111 brackets updated as of the Selection Show. All 111 had West Virginia in. No team has ever been left out that was included in all of their submissions. There are always additional brackets that come in from people who don't do regular updates throughout the year and just submit a bracket email at the end. Of the additional 109 brackets that did so, 104 had WVU in. Every serious bracketologist and the vast majority of people who just throw in results at the end included the Mountaineers.

Why should WVU have been a lock? The at-large field extends to team #46 on the S-Curve, North Carolina. In terms of selection, West Virginia was inside the top-46 (45 KPI, 42 SOR, 43 WAB) in all three resume metrics. North Carolina (55 KPI), Texas (58 KPI, 54 SOR), and Xavier (60 KPI) cannot claim the same, yet all three were in. As a result, West Virginia had a better resume average than any of those three included teams (as did Indiana, top-49 in all three metrics).

For years, we have been told the Selection Committee looks at who you played, where you played them, and what the result was. West Virginia won 6 Q1 games, 5 away from home (2 neutral, 3 true road games). Bubba Cunningham cited that West Virginia would be without Tucker DeVries for the Tournament as a reason they were left out. That is simply preposterous. DeVries only played 8 games this year. The Mountaineers were still 13-10 without him, including two of their best wins, at Kansas and over Iowa State. Those two wins are better than anything North Carolina has on their team sheet.

Photo from CBS Sports broadcast

Vice Chair Keith Gill even cited North Carolina's Q2 record as a reason for inclusion. This is ridiculous for two reasons. First, no one has ever used Q2 record as a reason to include a team before, particularly a team that was 1-12 in Q1 games. Second, there were ZERO tournament teams on North Carolina's Q2 column. Xavier at least beat UConn and Creighton in Q2, but UNC didn't add a single game of value. What the Selection Committee is effectively telling us is that a win over UCLA on a neutral court is worth more than winning at the Phog (UCLA & Kansas are on the same seed line) and beating ISU (a 3-seed) combined. Not to mention the wins over Gonzaga (N), Arizona (N), Cincinnati (A), and Utah (A), all also Q1 wins the Mountaineers earned, each of which equals the sum total of UNC's Q1 wins and combined obviously are better than UNC's sum resume.

It has been noted that Cunningham would not have been in the room when North Carolina discussed. I'm sure this is true, but it does not preclude Cunningham as chair from influencing this decision without being in the room. When the Selection Committee was initially discussing their procedures and then first got together in February, it was presumably Cunningham who explained what would be important or not important for selection. At that time, North Carolina was already 1-10 in Quadrant 1 games, but 5-0 in Q2. Graham Doeren has pointed out the influence a Chairman can have in this regard. Looking ahead, UNC only had one Q1 game left on their schedule, so they weren't going to significantly improve their Q1 record. But they would have the chance to improve their SOR/WAB simply by winning Q2 & Q3 games in the ACC. If Cunningham says "we like to prioritize the metrics over just raw results, especially the new WAB metric that was added this year. And teams in Q1 and Q2 are designed to reflect postseason caliber teams, with Q1A representing protected seeds, Q1B representing at-large caliber teams, Q2A representing upper seed (12-14) auto-bid teams, and Q2B representing lower seed (15-16) auto-bid teams. Because all of these could be tourney teams, we will weigh those games about the same." Anyone saying that Cunningham didn't matter or couldn't have mattered in this regard is being disingenous.

And while I'm zeroing in on North Carolina, Texas might be an even more egregious inclusion. In terms of resume metrics, the only top-46 metric they had was #45 WAB. Yes, they had 7 Q1 wins, but they balanced that with 10 losses. We're told that for North Carolina, Q2 record matters, yet Texas was 3-5 in Q2. Like North Carolina, they had zero wins over tournament teams in Q2. Texas tied a record for the most total losses for an at-large team (15) and had an abysmal NCSOS (#286) so they positively fail the "who/where/result" question because in terms of their controllable schedule, they played 8 low-major buy games (all wins), 3 teams not near the tournament (all wins), and 2 teams that were in consideration (both losses). They didn't beat a single non-conference opponent that was in at-large tournament consideration. No team has ever made the field at more than 1 game under .500 in Q1-3, yet Texas was 12-15, worse than the precedent by a magnitude of three. 

But of course, Big East fans will be happy Xavier was included. Why was Xavier included? I'm confident the answer is because there is no comparison under which you can include North Carolina and not Xavier. Both had just 1 win in Q1, but Xavier had the better win percentage there and two additional wins over the field. Xavier had a cleaner resume, with zero losses outside Q1+2. And like UNC, they had 8 of these apparently now coveted Q2 wins.

Photo by William Purnell | Imagn Images

So in order to snub West Virginia, they had to make up an excuse for Tucker DeVries, who hasn't played since the kids were opening St. Nick's stockings, shoehorn Xavier in because of the UNC comparables, and include a Texas team with unprecedentedly bad performance in meaningful games.

Quickly on Indiana, they had 4 Q1 wins, including 2 over protected seeds (at Michigan State, over Purdue), were undefeated in Q2 (and Q3 for that matter), and had better resume metrics than any of the three "Last Four In" teams we are discussing here. And don't get me started on Boise State or UC Irvine, it's very obvious that even with a majority of non-power conference members, the name on the front of the jersey is a whole lot more important than the substance of the resume. Texas & UNC over Boise is laughably bad, and while I can see leaving Irvine out (I did) there is no world under which they are not in the First Four out and should certainly be ahead of Ohio State. But again, name on the front of the jersey, unless you're Indiana who already fired their coach.

Missed Seedings

Three in particular stood out. The first one was on the 5-line, where Memphis shows up. The Selection Committee has 7 metrics on the team sheet. According to the NET, which is primarily a sorting tool for the quadrants, Memphis is ranked #51, which equates to a 13-seed. According to the results average, their 18.3 does reflect a 5-seed, but their predictive average puts them at 51.3, also a 13-seed. Logically, they should be somewhere between those ranges, which is why we had them as an 8-seed, giving them extra credit for their big wins and result averages despite the quality metrics saying they should be 5 seed lines lower. This miss was so bad that Memphis is a +3.5 point underdog at Action Network against 12-seed Colorado State.

Photo by Ben Solomon | AAC

What compounds this is our next missed seed. Louisville landed on the 8-line. We had them as a 6, which was right in line with their 23.3 overall metric average. But if the message from Memphis is that resume metrics matter more for seeding, then why was Louisville's 13.7 resume average, more than a full seed line in value better than Memphis', not get them a better seed than the Tigers, particularly when Louisville was better in every other metric on the team sheet? This is the definition of hypocrisy, having completely different criteria for teams. This is exacerbated by putting Louisville in the Lexington pod with Auburn. This means that if Louisville wins, the overall #1 seed will be playing a de facto road game in their opening weekend. It's less than an 80 mile drive from Louisville to Lexington. That will be a heavily pro-Louisville crowd. So not only are the Cardinals punished in terms of being given a seed worse than they deserved, Auburn is punished by being bracketed against a team better than they should see in the second round in a road environment. Typically, a team is never given a true seed lower than their lowest metric. Louisville's lowest metric on the team sheet was their #28 BPI, which was only that low because it includes a factor for preseason ranking, so Louisville's worst metric was that bad because the formula BEFORE ANY GAMES WERE PLAYED didn't think they were good. But even with that, they should be no lower than #28, which is the last 7-seed. There is no excuse for this.

The last error of note is Gonzaga. Quite simply, they should not have been an 8-seed. The design of the bracket is to give a favorable path to the best teams. That means a 1-seed should not have to worry about facing a top-10 predictive matchup before the Elite Eight, and certainly not in the round of 32. But that's the case for Houston if they advance to play the Zags. While the seeding is only one line off what we had, Gonzaga's predictive metrics were far too good to put a 1-seed in that position the first weekend.

Our Results

On the whole, it was a decent year. We correctly predicted 65/68 teams, but as virtually everyone had West Virginia in, it was more akin to 65/67. Of those 65, we had 49 on their exact seed line and 14 that were one line off. Our total Paymon score, which is used to judge entries on the Bracket Matrix, was 356 points, a 10-point improvement over last year. Rankings will be out later, but we'll mostly be on to other pursuits (like picking our own bracket) at that point.

My final argument is that it is past time for the Selection Committee to be broken up and reconfigured. Having partisan conference members play such a major role in who gets in and who is left out is a bad look. Right now, it is impossible to not look at the selection of UNC over West Virginia and not assume the Chair's presence influenced that decision. According to the New York Post, Cunningham earns a $67,905 bonus for making the field. Whether that played into the selection or not, the mere impression that it could is problematic.