"My rule was I wouldn't recruit a kid if he had grass in front of his house.
That's not my world. My world was a cracked sidewalk." —Al McGuire

Tuesday, February 21, 2023

Top-16 Reset


The NCAA revealed their Top-16 teams on Saturday, and the general reaction among bracketologists was one of confusion. There have been a number of comments about what the Selection Committee got wrong, where they were inconsistent, and where they flat out didn't make sense. But why was there so much confusion?

Matt Norlander on the CBS Sports Eye on College Basketball shared this from his interview with Selection Committee chairman Chris Reynolds: "They put about 22 hours of work into this top-16 this week...they got together on Monday, long day on Tuesday, a little more on Wednesday." This feels incredibly important for a few reasons. First, they were done selecting, seeding, and placing their top-16 by Wednesday. That means the Marquette win over Xavier, Tennessee win over Alabama, Indiana loss to Northwestern, and Maryland win over Purdue were not factored into the Top-16 at all. Norlander also mentioned "there was no other team in contention for a one-seed." We'll discuss this timing later, but this information seems incredibly important and might clearly explain the confusion from bracketologists.

In terms of results, Cracked Sidewalks had 15/16 teams correct (Our #12 UConn was out, our #18 Xavier was in). In terms of seeding, we had 11 exact and 5 within one seed line (including Xavier, though UConn was two off). However, to support the "decisions were made without factoring this week" theory, our bracket from Sunday was even more accurate. 15/16 were still correct, but 13/16 were exact seeds and the other 3 were within one seed line. Had we just gone with our bracket from before this past week's games, we would've been better than we were after the games. That supports that this past weeks games didn't matter. So what did we learn? Let's dig in.

1-Seeds: Alabama, Houston, Purdue, Kansas

We had all the teams right, but our order was off, with our #2 Kansas and #4 Houston flipped. Most likely, the Selection Committee voted on the 1-seeds and 2-seeds on Monday. At that time, Alabama had not yet lost to Tennessee and Purdue had not yet lost to Maryland. Despite that, it seems metrics were prioritized for Houston. The teams did fall in exact order of the RAP score on Monday night. Had they voted these on Friday, the order likely would've changed. But if they voted Monday and put heavy stock in Houston's predictive numbers, this makes sense.

2-Seeds: Texas, Arizona, Baylor, UCLA

We correctly placed 3/4 here, and said we could understand Arizona supplanting Tennessee, which is what happened in terms of teams, but again the order was wrong. This really seems to be down to timing. If the 2-seeds were voted in on Monday, Texas had the best RAP score and zero losses outside the field. That night, they took a loss to Texas Tech while Baylor got a win over the field against West Virginia. Two nights later, 3-seed Tennessee added a win over the top overall seed, Alabama. That win was a big part of why we put Tennessee on the 2-line. It seems UCLA's lack of quality wins hurt them and not factoring in results Monday or later impacted the order here. This will be considered in our new S-Curve.

3-Seeds: Tennessee, Virginia, Iowa State, Kansas State

This is the first line that shocked us, mainly because of the inclusion of Iowa State and Kansas State. Metrically, this line went perfectly with RAP Score of these teams, but Gonzaga, Marquette, and Indiana all had better RAP Scores than ISU or KSU. Metrically, there is a huge gap between Tennessee and Virginia to the rest of this line. The other thing that stands out is wins over the top two lines. Tennessee, Iowa State, and Kansas State all had three wins over teams on the top two lines while Virginia had one. What this tells us is that as we get further down the S-Curve, quality wins are more important than raw metric numbers. However, we also know that on Tuesday, Kansas State lost to Oklahoma, on Wednesday, Tennessee beat Alabama, Marquette beat Xavier, and Iowa State beat TCU. We do not believe these results were factored into the Top-16.

4-Seeds: Indiana, Marquette, Gonzaga, Xavier

The biggest surprise here, depending on who you ask, was either Indiana's inclusion (though we predicted them as a 4) or UConn's exclusion (that was surprising). What surprised us about Indiana was their placement ahead of Marquette and Gonzaga, During the week, both Marquette and Gonzaga improved their position with significant wins while Indiana hurt their own standing. Xavier added a loss that may not have hurt much, but opens the door for the teams behind them. Indiana does, however, have three true road wins over tourney teams, which is three times as many as Marquette, Gonzaga, and Xavier combined. We continue to see the importance of true road wins over tourney teams (something Marquette should bear in mind Tuesday night).

Presumed 5-Seeds: Creighton, Miami, St. Mary's, and UConn were mentioned in alphabetical order. Reynolds specifically said this was not the actual order of those teams. One thing that's notable is Creighton, St. Mary's, and UConn were the three highest NET and Predictive Metric Average teams not included. Further, of all the teams considered with five or more Q1+2 wins, Miami and St. Mary's had the best winning percentages. This shows a pretty straightforward mix of computer numbers and quadrant performance. So as confusing as this Selection Committee looks, it's probably more straight forward than people assume.


  • Road wins over tourney teams are gold: Whether it's the placement of K-State and Iowa State or the inclusion of Indiana, the teams that outperformed their metrics are the ones that got big wins on the road. Looking down the bracket, this is good news for teams like Kentucky, Wisconsin, and New Mexico.
  • Metrics alone are not enough: UConn and to a lesser extent St. Mary's not making the cut shows that just having gaudy predictive numbers won't be enough. Teams like North Carolina and Memphis shouldn't feel safe, but this reveal only makes it worse.
  • Championship Week (again) won't matter: If they made these decisions with 4-5 days of results not factored in, it's highly unlikely they'll allow the results of Championship week to matter. We saw it with Texas A&M's exclusion and Virginia Tech's seeding last year. Championship Week isn't worth considering when it comes to selection or seeding changes.
  • The Bubble shrunk: It looked a lot more wide open a week ago, but the weight of road wins over tourney teams really helps the competing Big 12 schools and high-majors like Mississippi State and Wisconsin while hurting teams like North Carolina and Arizona State. Ultimately, the final two spots came down to just four teams, with USC and Wisconsin edging out New Mexico and Penn State. The further we go into the season, the more clear the bubble becomes.

 Let's look at the updates.

Multibid Leagues

Big 10: 9

Big 12: 8

SEC: 8

ACC: 5

Big East: 5

Mountain West: 3

Pac-12: 3

American: 2

WCC: 2

No comments: