"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, March 16, 2021

Breaking Down the Bracket

So this is going to look at how Cracked Sidewalks did in terms of comparing our projections with the actual Selection Committee field, as well as where the Committee got it right and where they got it wrong.

The Results

Of the 203 bracketologists on bracketmatrix.com, Cracked Sidewalks ranked 58th. We correctly picked 66/68 teams. In terms of seeding, we had 49 teams placed on the correct seed line and 61 teams placed within one line of their actual seed. Cracked Sidewalks' score of 357 is the best we've had and was better than many of the better known bracketologists, such as Joe Lunardi at ESPN and Jerry Palm at CBS.

While 66/68 seems good, the top 178 brackets all had 66+ teams correct, so this is a place we're disappointed. Placing 49 exact and 61 within a line (which is the distance teams get moved due to matchups) is a strength. We're proud of those numbers. With NET being a new tool, this is why many of us wanted a bracket last year for a tournament we knew wouldn't happen. It would've given more perspective on the Selection Committee thought process.

The Bubble

The biggest one we got wrong was having Syracuse in and Louisville out. I would argue this was a mistake by the Committee, but I can see their rationale. Syracuse had better computer numbers (largely due to Louisville's 40+ point losses to UNC & Wisconsin) and similar records in the various quadrants. I don't think 'Cuse should've avoided the First Four. The other one we got wrong was swapping Mountain West teams. I suspected they would get two in and went with Colorado State over Utah State because they had better top-end wins and no bad losses. Ultimately, the Selection Committee prioritized the predictive metrics, where USU was superior virtually across the board. It seems the predictive metrics and the NET, where both Syracuse and USU were top-40 while Louisville and CSU were sub-50. Lesson learned.

Chris Mack & Louisville were the biggest snub

Photo by Joe Robbins | Getty Images

There were quibbles about the inclusion of #70 Michigan State and #72 Wichita State based on their poor NETs. While I understand the complaint, I'm pretty sure the argument in favor of them was their top-5 wins. Teams in the top-5 had 13 combined losses. Every team with a top-5 win and finished with a winning record is in the field. That includes MSU and WSU.

The other thing they got wrong was Drake. In the Committee's S-Curve, Drake was listed behind Oregon State and Georgetown. This was simply wrong. Oregon State and Georgetown, had they lost, would not have earned at-large bids. They weren't close. So how is Drake behind them in the Curve? There's no rational explanation.

Our Seed Misses

While it was a one-line disparity, we disagree with West Virginia as a 3-seed and Oklahoma State as a 4-seed. The Cowboys may not have had the computer numbers, but they had more Q1 wins than anyone other than 1-seed Illinois. While their NET and predictive metrics (BPI/kenpom/Sagarin) are lower than WVU (29 to 24 and 30.7 to 24.3, respectively) their resume strength metrics were notably better (7.0 to 13.5) and their work at the top of the resume, which was clearly superior (10-6 vs 7-7 in Q1).

We disagree with Loyola-Chicago on the 8-line. It's understandable that their body of work didn't merit the top-3 seed their NET and predictive metrics would've indicated, the Rambler had a better NET and better Q1 winning percentage than any team on the 6-line. They should've been a 6, swapped with San Diego State, who didn't have a Q1 win until this past Saturday.

Florida (7 instead of 9) and Wisconsin (9 instead of 7), were 2-line misses as well. Teams from the 7 to 9 line accounted for 8 of our 17 correctly picked but incorrectly seeded teams. It was a logjam and there wasn't a lot to differentiate them, so we're okay with those.

The Selection Committee confused Colgate's seed with Keegan Records' jersey number

Photo courtesy of Colgate Athletics


Our other 2-line miss was Colgate, who was a 14-seed we had as a 12. When I look at resumes, I look primarily at record by quadrant, NET, resume based metric averages, and predictive metric averages. Colgate is equal or better than any 12-seed in Quadrant 2 & 4 records, NET rank, and resume metrics, That means they were the best team in 4 of the 7 factors I find tend to best determine seeding. They were better than each 12-seed in at least one of the other two categories (Quadrant 3 and/or predictive metrics). So Colgate was better than any 12 seed in at least 5/7 categories I look at.

Q1 WP Q2 WP Q3 WP Q4 WP NET Resume Predict
Colgate NA 1.000 .857 1.000 9 50.0 65.0
Oregon St .500 .333 .800 .714 91 66.5 84.7
Georgetown .250 .833 .667 1.000 64 51.5 61.0
Winthrop NA 1.000 1.000 .950 55 57.0 88.3
UCSB NA .333 1.000 1.000 54 53.0 75.0

Colgate is equal to or better than Oregon State in 6/7 categories, Georgetown in 5/7 categories, Winthrop in 5/7 categories, and UCSB in 5/7 categories. Sure, you can dig into Strength of Schedule, Average NET W/L, and other factors, but that's digging pretty deep. It's also worth noting the two high-major teams both edged Colgate in the Q1 field only because Colgate didn't play a single Q1 game, just like Winthrop and UCSB, also on this line. No, Colgate shouldn't have been a top-3 seed like their NET indicated, but they deserved better than a 14.

Regional Thoughts

West: Seeding was generally fine, but I'm not a fan of Gonzaga being in a region where they already beat the 2, 3, and 4 seeds by double-digits. I understand that seeding principles meant they had to have a Big 10 team on the 2-line and a Big 12 team on the 3-line, but they could've placed Ohio State and Texas in this region for a bit more interest. Yes, it gives them a slightly tougher 2 and slightly weaker 3, but that balances. More effort should've been put into this and now I find myself cheering for Creighton and USC just so Gonzaga gets to see some new opponents.

South: This region is generally fine. No real complaints, though this is where switching Iowa and Ohio State or Purdue and Virginia to the West would've made for more compelling stories.

East: In 9 of the last 10 tournaments, a 1-seed bowed out by the Sweet 16, and Michigan sure seems like the pick for that this year. While they should coast through their first game, the Wolverines have lost 3 of their last 5 and may be without their most important player, Isaiah Livers. LSU and St. Bonaventure would both be challenging opponents that offer vastly different styles of play to prepare for, and a potential Sweet 16 game with Florida State is the kind of team that is a nightmare matchup for a team that performs worse against teams that play up-tempo and do their damage inside like the Seminoles. This should be a fun region and I could see a path for Michigan, Alabama, Texas, FSU, BYU, or even UConn to come out if things break right for them.

Midwest: I feel bad for Illinois. Two of the teams that we feel were glaring seed errors, Loyola-Chicago and Oklahoma State, are potential first weekend matchups for them. When the Selection Committee incorrectly seeds teams, it doesn't just hurt the teams they incorrectly seeded, but also the teams they have to play. If the Illini manage to escape what is the toughest first weekend path of any 1-seed and seeding holds, they would have to take on Houston, who based on our February research into the "Gonzaga vs the field" question is, as of Selection Sunday, one of just four teams (Gonzaga, Michigan, and Illinois are the others) that should be considered viable National Championship contenders. Though we could add UConn in the East, since they are the most likely program to buck the trends we uncovered. In addition, it's interesting that the Big 10 tournament clearly was the impetus to move Illinois ahead of Michigan on the 1-line, but Oklahoma State didn't garner similar movement from the 4 to the 3, which means that while Championship Week helped the Illini directly, it harmed them indirectly.

Final Thoughts

I'm pleased with the results this year. Clearly the Selection Committee prioritized NET and metrics when it came to the bubble as everyone in the top-40 was included. While there were a couple of outliers, big wins ultimately were the trump card for teams outside the top-40. Correctly placing all 16 teams in the protected seeds and only swapping two of them in terms of seed line is a nice feather in the cap, though I'm more proud to have correctly placed 13/14 teams on the 14 to 16 lines, because that's where I feel you demonstrate attention to detail in understanding the entirety of the field.

No comments: