And so, it’s another season of bracketology in the books—and not too bad of one either. Here’s a look at my performance:
- 1 miss, Texas got in over Oklahoma, which I would have gotten right had I stuck to my first instinct.
- 45 teams seeded correctly (an improvement of 1 over 2025)
- 20 teams seeded within one seed line (an improvement of 4 over 2025 (16))
- 2 teams seeded within two seed lines, Vanderbilt and Utah State (an improvement from 6 in 2025)
This was my best performance since 2022 (1 miss, 45 seeded correctly, 21 within 1 line, 1 within 2 lines), so I feel like I’m on the right track in terms of navigating a changed environment.
Here are some things that stuck out to me, particularly in terms of consistency on the part of the Selection Committee, even if I think Division I Men’s Basketball Committee chair/Sun Belt Commissioner
Keith Gill’s transparency and frankness is an example that future chairs should follow to make the process more accessible to fans and the media. My North Star is this statement from Seth Davis on the Selection Show: Your Results-Based (backward-looking) metrics that determine selection and your Predictive (forward-looking) metrics that most determine your seeding.
- Three of my biggest misses resulted from inconsistency here. I based my placement of Vanderbilt (3 line), Virginia and St. John’s (4 line), and Utah State (7 line) on their predictive metrics. In the Commodores case, their average ranked 12th in predictive metrics and 8th in results-based metrics, which is a projected 2 seed.
- The Selection Committee clearly took the perception that they don’t pay attention to Sunday title games to heart—whether that be through very late scrubbing or building contingency brackets. That was evident in Michigan dropping from No. 2 overall to No. 3 overall and Purdue sitting on the 2 line due to the Big Ten Championship game’s result. But again, it feels like this was applied inconsistently. I would hope they didn’t drop Vanderbilt to the 5 line after the SEC Championship when the metrics clearly indicate the Commodores should have been higher—and those numbers would not have dropped that much after a loss to an Arkansas team that ended up on the 4 line, particularly after beating the final No. 1 seed the afternoon before.
Turning my attention to the Cut Line picture, and Miami (Ohio), Texas, Oklahoma, Auburn, and VCU, among others.
Remember the first part of Seth Davis’s statement, “Your Results-Based (backward-looking) metrics that determine selection.” Something that people either forget or don’t realize is that building the bracket is a multi-step process, and this is something a good bracketologist remembers and incorporates into their work.
- The first step is determining your 37 at-large teams. If a selected at-large team secures an auto bid, they move out of the at-large pool and a replacement is voted in.
- You can use this rule to determine where the Cut Line will be for any metric or ranking. Start counting from the top and every time you encounter a projected auto bid add 1 to the Cut Line total. This year, that number turns out to be either 44 or 45 depending on the group of metrics as you’ll see.
- Here’s what the Cut Line picture looks like when looking solely at the three Results-Based Metrics (KPI, Strength of Record (SOR), and Wins Above Bubble (WAB)):
- Here SMU is 45th, they are the 37th at-large team after accounting for 8 auto bids.
- Note where Miami (Ohio) and VCU are relative to the Cut Line. The RedHawks are 6 places above the Mustangs, and the Rams 8 places above them. So, if you assume that VCU would not have been an at-large with a loss to Dayton in the Atlantic 10 final, you might have very well been wrong.
- Turning our attention to our three SEC possibilities—Auburn, Oklahoma, and Texas, the Tigers should have been in based on their Results-Based metrics, but the Committee realized that including a 17-16 team might set a bad precedent. The Longhorns rank higher than the Sooners by 2 places, even though there are 3 spots between Texas and SMU, with both San Diego State and Virginia Tech between them. So, you can conclude there were some close votes and a lot of conversation about these teams.
- Note also that Dayton is 1 spot behind Oklahoma. This is another piece of evidence that VCU may not have necessarily been left out with a loss on Sunday.
Now let’s turn our attention to the second part of the process, seeding, where the three Predictive Metrics (ESPN’s BPI, KenPom, and Bart Torvik’s T-Rank) are more closely scrutinized.
- After teams are selected as an at-large or qualified via an auto bid, they are evaluated and ranked together.
- Again, start at the top and work your way down, adding 1 for each auto bid until you get to 37 at-large bids. VCU misses out on Predictive Metrics, so we only have 7 auto bids to account for. Our Cut Line, therefore, becomes 44.
- Note that SMU remains the Cut Line here, but there’s a whole lot more going on.
- 4 teams ranked above the Mustangs based on Predictive Metrics are out: Auburn, Indiana, Cincinnati, and Oklahoma.
- Note where the 4 at-large teams ranked below SMU are. TCU, Missouri, and UCF are in the general neighborhood of the Ponies, but Miami is not. In fact, the RedHawks are well out of the group. Their average rank is a whopping 15.4 places behind the closest at-large candidate, Cal, and 16.4 places behind that Dayton team that supposedly would have kept VCU out.
- So, you can see why Keith Gill said the RedHawks were not the last team in but were seeded in the First Four. A team playing Miami’s schedule that finished 28-1 against DI opponents, but playing just a single Quad 2 game (and none against Quad 1) is an unprecedented event that’s difficult to evaluate. And considering we only see a team finish the regular season undefeated approximately once a decade (Kentucky 2015, Saint Joseph’s 2004), the RedHawks’ selection really shouldn’t have caused the amount of drama it did.
One final point on the two types of metrics. You can also apply the general rule about using Predictive Metrics in seeding when it comes to auto bids (since the Results-Based metrics aren’t as necessary with the teams already included by rule).
My mistakes on this front are partially explained by this too.
- By Results-Based metrics, North Dakota State would have been seeded 13th and Troy 14th, but these were reversed in looking at Predictive Metrics.
- Both Queens and Siena should have been 15 seeds based on their results, with UMBC also rising to a 15 and Furman and Idaho slotted on line 16.
- But looking at their Predictive Metrics, Queens slotted as 15 and Siena a 16. However, that doesn’t explain why Tennessee State is a 15 instead of a 16 and why Penn and Idaho ended up swapped on the 14 and 15 lines, as I projected.
And there are our lessons learned for 2027. Tomorrow, I’ll be back with picks.
