Seth Davis' College Basketball Rankings: Big Three Looms Large as Selection Sunday Draws Nigh

Buh bye Clemson and Dayton, and hello South Florida

Mar 03, 2024

Two from weeks from today, we’ll be talking about the 2024 NCAA tournament bracket. In the meantime, we’ve got two more Top 25 rankings to comb through. When we get to this late in the season, there’s not a lot of fluctuation from week to week, but there are always fresh results to finesse. I’m also including in my comments a big-picture breakdown of what I value when I dive into all the data, followed up by my usual point-by-point explanations.

A few weeks ago, I started including each team’s rankings in the four main metrics — the NET, KenPom, KPI, and BartTorvik. The NET is the NCAA’s official organizing tool and blends efficiency, strength of schedule, and results. KenPom is the most respected predictive metric (hi gamblers!) and includes data from the past five seasons. KPI places strict emphasis on this season’s results and schedule. And BartTorvik is similar to KenPom but has been filtered to include only this season’s games. If you want to play this game yourself, I highly recommend using Bracketologists.com, and especially their Compare Resumes feature.

Here, then, is the correct order of the Top 25 teams in men’s college basketball:

Dropped out: Clemson (20), Dayton (24)

Almost Famous: Boise State (24-34-29-34), Dayton (21-27-16-32), Michigan State (22-20-44-19), Nevada (41-41-26-41), Oklahoma (40-37-47-37), Texas (27-26), Villanova (26-24-57-25), Wake Forest (31-21-50-26)

SIX BIG PICTURE POINTS ABOUT HOW I VOTE

1. The great thing about being a poll voter is it’s entirely subjective, not just how you vote but why you vote. For starters, I think of my ballot as a poll, not an NCAA tournament bracket projection. The selection committee is instructed to value a game in mid-November the same as late February, but I am under no such constriction. I take into consideration full-season records in the four NET quadrants, but I place a heavy emphasis on a team’s performance over the last four weeks. BartTorvik’s data sorting filter, which allows me to see all the numbers between selected dates, is an extremely useful tool for this.

2. Metrics are useful, but they are not gospel. I mostly look to them to confirm what I already suspect, but I’m also searching for outliers. When I find one, it gives me a spot in the dirt to dig deeper.

3. A game’s location is just as important as its result. It’s hard to win on the road, in case you haven’t heard.

4. When it comes to factoring injuries, it matters to me how long the player is going to be out. If it’s only going to be a game or two, I’m more likely to forgive a bad loss. If it’s going to be a while, I’m more likely to rank the team as it is currently constructed, and then adjust accordingly when the player returns.

5. I try to honor my previous week’s rankings without being beholden to them. For example, last week I installed Houston at No. 1 even though the Cougars were ranked behind Purdue the previous week and the Boilermakers did not lose. But Houston had an excellent week with wins over Iowa State (home) and Baylor (road), so I vaulted the Cougars over both Purdue and UConn, which had lost big at Creighton.

6. I vote rationally, not sentimentally. Some of my fellow AP voters like to bend over backwards to include mid-majors, particularly in the last spot. I fell into this mindset in the past, but in recent years I’ve tried to keep it on the level. If you get my vote, you earned by the way you played, not because you won my heart.

NOTES ON THE VOTES

It doesn’t matter what order the top three teams are in, so long as you have the right three. Here’s an interesting parlor game: Would you rather have the Big Three to produce a national champ, or the field? Given that 15 of the last 22 NCAA champs were No. 1 seeds, I’d go with the Big Three, but this is the type of season where we should expect the unexpected.

I performed a minor tweak by moving Arizona up to No. 6 following its wins over Arizona State (road) and Oregon (home). The Wildcats have been ranked in the top four of the metrics all season. Their metrics rankings are better than Iowa State’s, but the Cyclones got the edge because of their Jan. 9 win over top-ranked Houston. Arizona has eight Quad 1 wins to Iowa State’s six, which would likely put the Wildcats ahead of Iowa State on the selection committee’s overall seed line. But the Wilcats have a Quad 3 loss at Oregon State while the Cyclones have none.
It’s interesting seeing Alabama and Washington State next to each other, because those are two teams whose rankings belie their metrics, albeit in opposite ways. Alabama’s metrics would dictate a top 10 ranking, but while the Tide only has one loss below Quad 1, their Quad 1 record is 3-8, and their best win was at home against Auburn. If you went only by Washington State’s metrics, they wouldn’t be ranked at all, but the Cougars swept Arizona and have a neutral court win over Boise State. Even Washington State’s lone Quad 3 loss, by eight points to Santa Clara on a neutral court, wasn’t terrible.
Moving down my ballot, Auburn and Duke are two more teams whose rankings are lower than their metrics would dictate. Auburn is a real head scratcher for me. The Tigers are 1-7 in Quad 1 games and undefeated everywhere else. Their best win was at home over Alabama. Duke’s metrics would appear to merit a higher ranking, and they may be better than I’m giving them credit for. If you filter BartTorvik’s data for the last month, Duke is ranked No. 2 in the country in efficiency margin, and it’s No. 14 in Wins Above Bubble. Yet, because the ACC is so weak (and no, it’s not because the coaches haven’t properly “gamed the system”), the Blue Devils only have one win against another team ranked on my ballot, and that was on a neutral court over Baylor. Plus, Duke also just lost sixth man Caleb Foster to a foot injury which is might keep him out the rest of the season.
Kentucky had a terrific week with wins at Mississippi State and at home over Arkansas, but both those opponents are unranked, and the Wildcats’ metrics are so-so. Plus, there weren’t many compelling reasons to move teams who were ranked ahead of them down. Kentucky moved up a spot to No. 16, which is higher than it’s ranked in any of the four main metrics.
Kansas has an injury issue with Kevin McCullar, which no doubt led to its loss at home to BYU. But the Cougars have had stellar metrics all season long, and they now have six Quad 1 wins, including at home over Iowa State, Baylor, San Diego State and Texas, and on the road over UCF. Plus, even with McCullar back, the Jayhawks lost at Baylor on Saturday. So I knew wherever Kansas landed, it needed to be behind BYU.
Gonzaga is a perfect illustration of the difference between a ballot and a bracket. The Zags re-entered both national polls last week (they were in my Almost Famous list), but according to all the mock brackets they were right on the bubble. That should change following their road wins at San Francisco and Saint Mary’s by a combined 31 points. KPI hasn’t liked Gonzaga all season (the WCC doesn’t help), but if you look at this team’s other metrics, you’d think it was closer to a No. 4 seed than the First Four. Over the last month, Gonzaga is No. 12 in adjusted efficiency margin on BartTorvik and No. 9 in WAB. I’ll predict that Gonzaga is going to out-perform the mock brackets on Selection Sunday. At any rate, they clearly deserve to be ranked, and even though they split the season series with Saint Mary’s, given that the margin was so much greater the second time (the Gaels won by four), I felt like the Zags should be a spot ahead.
Yes, Florida lost to South Carolina on Saturday, but the game was in Columbia, the margin was six points, and the Gators are higher in three of the four metrics. The Gamecocks have two Quad 3 losses while the Gators have none. So Florida stays one spot ahead.
I wrote last week that I was close to ranking South Florida despite its very low rankings in the four metrics. It is indeed a rarity that a team with those numbers would end up on my ballot. But the Bulls, who ended up No. 25 in the AP poll last week, won me over with their 15-point road win over Charlotte. That increased the Bulls’ lead atop the AAC to three games over second-place FAU. Plus, Memphis has won three straight to raise its NET ranking to No. 74, which means USF’s one-point home win over the Tigers now counts as a Quad 1 win. That means nothing to the selection committee, but it’s an illustration that this team is closer to owning real estate in at-largeville than its metrics would suggest. I would recommend the Bulls win the conference tournament and take suspense out of the equation, but if they get to the final and lose, I think they’ll be in the field of 68.
When I was scavenging for teams to add to my Almost Famous list, I was surprised that St. John’s had such a strong case. The Johnnies’ win at Butler on Wednesday was their second straight Quad 1 win, improving their record there to 5-9 and 4-2 in Quad 2. Now their main mission is to avoid losing one of their two Quad 4 games this week at DePaul, and at home against Georgetown. If St. John’s can avoid bad losses and get one more good win in the Big East tournament, then there’s a very good chance the Red Storm will hear their name called on Selection Sunday. Pretty good for a team with no lateral quickness.