clock menu more-arrow no yes

Filed under:

Bracketology 2021: Recapping My Performance (And the Committee’s Too)

Plus, a look at the regions and pods based on the NET and seed list.

NCAA Basketball: SEC Conference Tournament-Lousiana State vs Arkansas
Both Arkansas and LSU figure into bracketing errors for 2021, from my perspective anyway.
Christopher Hanewinckel-USA TODAY Sports

I did merely an okay job of projecting this year’s field, at least based on the Bracket Matrix’s rankings, where I rank 140th out of the more than 200 brackets submitted. (I no longer feel quite so bad about wasting my winters away on this exercise.) My Paymon score of 345 was 27 points off the leading bracket’s, projected by old friend Patrick Stevens of The Washington Post, and 21 points lower than the Matrix’s consensus of 366. But it was also 102 points better than the worst performing bracket.

Yes, I correctly selected 67 of 68 teams, but let’s be real here. Of that total, 31 (usually 32) are automatic bids, so they shouldn’t count at all. Plus, by the time Selection Sunday rolls around somewhere between 30 and 34 of the 36 or 37 at-larges are virtual locks. So, I’m proposing a new measure for team selection success. I’ll call it Cut Line Percentage for now, and this is merely the percentage of the final eight at-large teams placed in the field. This encompasses the First Four and the last four byes. In my final bracket, that group looked like this:

Last Four Byes: VCU, Michigan State, Maryland, UCLA
Last Four IN: Drake, Syracuse, Louisville, Wichita State

The Committee’s last eight in looked like this:

Last Four Byes: VCU, Rutgers, Syracuse, Utah State
Last Four IN: Michigan State, UCLA, Wichita State, Drake

While I missed two teams when looking at the two groups head-to-head, with the Committee seeding Maryland higher and my projection placing Rutgers higher, there’s only one difference between my field and the real thing. That’s Utah State over Louisville. So, of the last eight teams I projected in, I missed one. That makes my Cut Line Percentage 87.5 (7 of 8).

Seeding Misses

Of the 67 correct teams in the field I seeded, I placed 42 of them correctly, that’s a decrease of one from 2019, but still four better than either 2017 or 2018. For the second consecutive tournament, I missed 18 teams by one line most of which were in the seven-to-10 range:

  • Underseeded by me (10): Houston, West Virginia, Colorado, BYU, Clemson, Loyola Chicago, Maryland, UC Santa Barbara, Ohio, Drexel
  • Overseeded by me (8): Texas, Oklahoma State, USC, Oregon, Wisconsin, Rutgers, Michigan State, Cleveland State

Then there were seven teams I missed by two lines:

  • Underseeded by me (3): Texas Tech, Florida (d’oh!), North Carolina
  • Overseeded by me (4): LSU, Georgia Tech, Missouri, Colgate

Thankfully, I didn’t miss any teams by three lines, which happened in 2016, 2017, and 2018.

How Did the Committee Do?

Overall, I thought the Selection Committee did a decent job, particularly given the unique circumstances of this season, but some decisions had me wondering.

  • West Virginia over Oklahoma State for a spot on the three line was an odd choice, but even weirder when you notice there are five spots between them in the Seed List, with Texas one spot lower the Mountaineers. Based on the Committee’s bracket preview show from February 13th it looked like they were again valuing quality wins more than anything else, at least when it came to the top of the bracket. By that measure, the Cowboys should have been seeded higher than either the Mountaineers or Longhorns, thanks to 10 quad 1 wins (including two over WVU, including one in Morgantown without Cade Cunningham and Issac Likekele), more than West Virginia’s seven and Texas’s eight. Perhaps the Committee put more faith in the Cowboys’ two losses to TCU than their wins.
  • Further down the bracket, this was something friend of the blog Matt Harris pointed out on Twitter on Sunday night regarding Missouri’s nine seed.

As the Committee worked its way down the bracket, the NET seemed to take precedence over the quality wins metrics, particularly when considering teams whose results were all over the place like Missouri, 7-6 in Quad 1 games but with a NET of 47, and Michigan State, with three top 10 wins and a NET of 70, which was fueled by a few blowout losses on the Spartans’ part. The reliance on the NET seems to explain why Syracuse and Utah State got the nod over Louisville despite similar profiles—the Orange’s ranking on Sunday was 16 spots higher than the Cardinals’, the Aggies’ 17.

It would just be nice to have some consistency regarding selection principles throughout the bracket.

Bracketing by the Seed List

With all NCAA Tournament games scheduled for Indiana in 2021, the Committee noted that it would use the seed list to build the bracket this year, keeping in mind that the usual restrictions related to regular-season matchups would apply. Well, how well did they do?

Based on the Committee’s Principles and Procedures document (page 5, point 5):

The sum totals of the top four seeds meet the requirement:

  • South: 31
  • East: 33
  • West: 36
  • Midwest: 36

Remember that the sum top four seeds in a region will equal 34 in a truly balanced bracket. Here, the South and East are slightly stronger regions than the West and Midwest. That is Gonzaga and Illinois will have to get through slightly weaker 4 and 2 or 3 seeds to reach the Final Four than Michigan, or especially Baylor.

Expanding to the entire regional brackets:

  • South: 528 (33 avg.)
  • Midwest: 529 (33.063 avg.)
  • East: 531 (33.1875)
  • West: 535 (33.4375 avg.)

This time, the West appears to be slightly weaker than the East, Midwest, and South overall.

Matchups and the Seed List

Based on the seed list, this is how the First Round matchups within each pairing rank from easiest to most difficult (largest gap between raw seed numbers to lowest)

  • 1 vs. 16: Gonzaga (First Four), Baylor, Illinois, Michigan (First Four)
  • 2 vs. 15: Alabama, Ohio State, Houston and Iowa (tie)
  • 3 vs. 14: Arkansas, Kansas and West Virginia (tie), Texas
  • 4 vs. 13: Florida State, Oklahoma State and Purdue (tie), Virginia
  • 5 vs. 12: Creighton, Villanova, Colorado and Tennessee (tie)
  • 6 vs. 11: USC (First Four), BYU (First Four), Texas Tech, San Diego State
  • 7 vs. 10: Oregon, Clemson, Florida and UConn (tie)
  • 8 vs. 9: LSU, Loyola Chicago and North Carolina (tie), Oklahoma

The numbers range from a 66.5-place difference (Gonzaga vs. Appalachian State or Norfolk State) to just one (Oklahoma and Missouri in the 8/9 game in Gonzaga’s pod).

Expanding this to look at the 16 chalkiest Second Round matchups from easiest to most difficult:

  • 1 vs. 8: Gonzaga/Oklahoma, Baylor/North Carolina, Illinois/Loyola, Michigan/LSU
  • 2 vs. 7: Ohio State/Florida, Alabama/UConn, Houston/Clemson, Iowa/Oregon
  • 3 vs. 6: West Virginia/San Diego State, Arkansas/Texas Tech, Texas/BYU, Kansas/USC
  • 4 vs. 5: Florida State/Colorado, Purdue/Villanova and Oklahoma State/Tennessee, Virginia/Creighton

While there’s a difference of one spot between the Cavaliers and Bluejays, Gonzaga has a 31-spot advantage over Oklahoma.

Then There’s the NET

If you’re looking for more potential upset insight, examining the bracket by NET is another tactic to try.

Based on the NET, each region’s level of difficulty looks a bit different. First looking at the sum of the rankings of the top four seeds in each. Remember, that the lower the total, the stronger the region.

  • West: 30 (36 by seed list)
  • South: 46 (31 by seed list)
  • East: 57 (33 by seed list)
  • Midwest: 61 (36 by seed list)

By this metric, Gonzaga’s path to the Final Four is far more difficult than Baylor’s and, especially, Michigan and Illinois’s.

However, things change a bit when looking at the entire region

  • South: 699 (528 by seed list)
  • East: 809 (531 by seed list)
  • West: 838.5 (535 by seed list
  • Midwest: 848 (529 by seed list)

Here, the South is far stronger than any other region, and there’s a specific reason for this—the presence of Colgate, ninth in the NET, as the 14 seed. In the other three regions, the 14s are ranked 114th, 121st, and 74th.

Matchups and the NET

The Raiders’ ranking makes the next couple of groupings extremely weird. First, the ranking of first-round matchup from easiest to most difficult for the higher seeds (largest gap between teams’ NET rankings to lowest). I’ve included these gaps for reference, if the number is negative, the lower seed has the NET advantage. I’ve bolded these instances.

  • 1 vs. 16: Gonzaga (189, First Four), Michigan (179, First Four), Baylor (160), Illinois (132)
  • 2 vs. 15: Houston (155), Ohio State (150), Alabama (131), Iowa (101)
  • 3 vs. 14: Kansas (103), West Virginia (97), Texas (53), Arkansas (-5)
  • 4 vs. 13: Virginia (75), Florida State (59), Oklahoma State (57), Purdue (37)
  • 5 vs. 12: Tennessee (78), Colorado (49), Villanova (38), Creighton (28)
  • 6 vs. 11: USC (39.5, First Four), BYU (38, First Four), Texas Tech (23), San Diego State (22)
  • 7 vs. 10: Florida (17), UConn (5), Oregon (4), Clemson (-3)
  • 8 vs. 9: Loyola Chicago (24), Oklahoma (11), North Carolina and LSU (-5, tie)

Yes, third-seeded Arkansas is an underdog by NET in its matchup with Colgate, joining one seven and two eight seeds.

Curiously, the all-power 5/12 games, Tennessee-Oregon State and Colorado-Georgetown, feature larger NET gaps than the power-vs.-mid pair of Villanova/Winthrop and Creighton/UCSB.

Now, let’s look at the chalkiest Second Round matchups by NET. Again, if the number is negative, the lower seed has the NET advantage. I’ve bolded these instances. I’ve also included an overall ranking, since these don’t flow naturally from seed group to seed group.

  • 1 vs. 8: Gonzaga/Oklahoma (35, 2nd), Baylor/North Carolina (30, 3rd), Michigan/LSU (24, 5th) Illinois/Loyola (7, 10th)
  • 2 vs. 7: Houston/Clemson (36, 1st), Iowa/Oregon (27, 4th), Ohio State/Florida (23, tie 6th), Alabama/UConn (23, tie 6th)
  • 3 vs. 6: Kansas/USC (8, 9th), West Virginia/San Diego State (6, 11th), Arkansas/Texas Tech (2, 12th), Texas/BYU (-1, 13th)
  • 4 vs. 5: Virginia/Creighton (14, 8th), Purdue/Villanova (-5, 14th), Florida State/Colorado (-10, 15th), Oklahoma State/Tennessee (-16, 16th)

So, three potential 4/5 matchups would see the No. 5 seed favored by NET, while this is the case for one No. 6 against a three. Also, Illinois’s second-round matchup with Loyola is more comparable to a 3/6 game than a 1/8.

Of the 16 pods, the top seed (1, 2, 3, 4) is the strongest team based on NET ranking:

  • 1 seed: 4 times
  • 2 seed: 4 times
  • 3 seed: 1 time (Kansas in the West, 6 seeds in the East and Midwest, 14th-seeded Colgate in the South)
  • 4 seed: 1 time (Virginia in the West, 5 seeds in the other three regions)

Since the NET is all over the place, use this information wisely when filling out your picks. Speaking of that, I’ll deliver mine at this time tomorrow (Wednesday).