Predicting The Top 4 Seed Lines

Bracketology season is nearly at an end, with the official bracket reveal coming at 6 p.m. eastern this afternoon. While most of the national conversation now centers around the bubble teams, the most relevant questions for Duke fans are what teams will be the Blue Devils’ biggest competition in their bracket.

With that in mind, I’m throwing my hat into the proverbial ring to predict that small portion of the bracket—namely, the Top 4 seeds in each region. I’ll do so roughly following the NCAA’s own

procedures and insights into the process from the NCAA’s mock selection process a few weeks ago. I’ll do so assuming the favorites in the most impactful games remaining this afternoon hold serve (Michigan over Purdue; Vanderbilt over Arkansas).

Step 1: The consensus rankings

I’ll begin with the best approximation we have of the committee’s consensus “true seed” list: the consensus from BracketMatrix.com. That would give us the following 1-16 ranking:

1. Duke
2. Michigan
3. Arizona
4. Florida
5. UConn
6. Houston
7. Iowa State
8. Michigan State
9. Purdue
10. Illinois
11. Gonzaga
12. Nebraska
13. Virginia
14. Vanderbilt
15. Alabama
16. Kansas

Step 2: “Scrubbing” the true seed list

According to the NCAA’s procedures, the next step is to compare each pair of adjacent teams on the seed line and make any necessary adjustments. Again, operating under the initial assumption of the above consensus true seed list (which is by no means accurate), I would make the following adjustments:

Houston over UConn: Houston and UConn are similar in the “result-based metrics” but Cougars are a tier above the Huskies in the “predictive metrics”. Houston also has 3 more Q1 wins than UConn and has qualitatively looked much better over the past few weeks, including last night—Houston competed with Arizona until the final minute, while UConn was non-competitive against St. John’s.

Virginia over Nebraska: Virginia has one fewer loss than Nebraska and just one fewer Q1 win despite three less opportunities. The Cavaliers also have superior result-based metrics compared to the Cornhuskers, while the pair have similar predictive metrics. The Cornhuskers do have better quality Q1 wins, including an impressive road victory over Illinois, but this might be negated by a one-and-done performance in the Big 10 tournament.

Arkansas over Kansas: Arkansas would be No. 17 on the consensus list, and I have them jumping Kansas for the final No. 4 seed. These two teams are in the same tier in the result-based metrics with Arkansas holding the slight edge in the predictive metrics—which apparently are more relevant to the seeding process. I think Arkansas having two fewer losses and a better conference tournament run, combined with Kansas’ mercurial performance all year, wins out.

This “scrubbing” gives us a new true seed line of:

1. Duke
2. Michigan
3. Arizona
4. Florida
5. Houston
6. UConn
7. Iowa State
8. Michigan State
9. Purdue
10. Illinois
11. Gonzaga
12. Virginia
13. Nebraska
14. Vanderbilt
15. Alabama
16. Arkansas

Step 3: Build the bracket

Now we assign teams to each bracket following the NCAA’s procedures, line by line. The No. 1 seeds end up being straightforward, as each squad fits naturally into a region: Duke to the East, Michigan to the Midwest, Arizona to the West, and Florida to the South.

Then, contrary to popular belief that just won’t die, the No. 2 seeds are placed based on location preference, not the S-Curve. No. 5 overall Houston would get first location preference and end up in the South. UConn would be next and end up in the East. Iowa State then ends up in the Midwest, with Michigan State drawing the short straw and being sent West.

The same process occurs on the 3 and 4 lines, sending teams into brackets based on geographic location with priority to the higher seeds. There are, however, some extra principles to consider at this stage. One such principle is that (quoted verbatim), “Each of the first four teams selected from a conference shall be placed in different regions if they are seeded on the first four lines.” Purdue, therefore, can’t get sent to the Midwest with fellow Big Ten team Michigan; the East is the next best location. Illinois faces a similar conundrum and therefore ends up in the South. Gonzaga is a natural fit for the West, leaving Virginia in the Midwest.

On the four line, things get complicated. Vanderbilt, Alabama, and Arkansas are the 2nd-4th SEC teams, so none of them can end up in Florida’s South region. Vanderbilt is closest to Chicago, so they go to the Midwest. Alabama gets the next priority and ends up in the East, then Arkansas gets sent out West. Nebraska is stuck in the South by default.

Step 4: Balance the brackets

We now have the following four regions, where I’ve included each team’s true seed and the sum of those seeds:

East (31): Duke (1), UConn (6), Purdue (9), Alabama (15)

Midwest (35): Michigan (2), Iowa State (7), Virginia (12), Vanderbilt (14)

West (38): Arizona (3), Michigan State (8), Gonzaga (11), Arkansas (16)

South (32): Florida (4), Houston (5), Illinois (10), Nebraska (13)

There’s an obvious problem here: a major imbalance between the difficulty of the East and West regions! As the committee builds the brackets it can adjust for this scenario (quoted verbatim): “Generally, no more than five points should separate the lowest and highest total” seeds in each region.

This is where Duke being the No. 1 overall seed helps make its bracket easier. There are two straightforward solutions: Swapping UConn and Michigan State on the No. 2 line or Purdue and Gonzaga on the No. 3 line. Either would cause a Big 10 pile up that the committee would want to avoid, and there’s no clear guidelines on how they’d deal with that; however, there’d be a major uproar if No. 1 overall seed Duke had objectively the hardest bracket by this large of a margin. I bet they’d swap the 3 seeds, preferring to punish a team further down the bracket rather than send UConn out West. That would leave:

East (33): Duke (1), UConn (6), Gonzaga (11), Alabama (15)

Midwest (35): Michigan (2), Iowa State (7), Virginia (12), Vanderbilt (14)

West (36): Arizona (3), Michigan State (8), Purdue (9), Arkansas (16)

South (32): Florida (4), Houston (5), Illinois (10), Nebraska (13)

Final predictions:

This exercise started from the assumption that the “consensus” from BracketMatrix is a reasonable assumption of the starting consensus in the committee room, but even some minor deviations would change this exercise significantly. For example, much of the Big 10 conundrum would be eased if Vanderbilt moves up to a No. 3 seed following an SEC championship this afternoon.

I actually think that’s a likely outcome, perhaps even if Vanderbilt loses—it would be easily justifiable and make the bracketing process much easier. Swapping Vanderbilt with, say, Illinois would leave two Big Ten teams on the 3 line and two SEC teams on the 4 line to facilitate things. Playing out some of the corresponding adjustments to the true seed line, I think the final bracket is more likely to look like:

East: Duke, UConn, Vanderbilt, Nebraska

Midwest: Michigan, Iowa State, Virginia, Alabama

West: Arizona, Michigan State, Gonzaga, Arkansas

South: Florida, Houston, Purdue, Illinois

Go to the DBR Boards to find Blue Healer Auctions | Drop us a line