The NCAA tournament bracket’s unveiling is always a cathartic event, especially given the current times we live in. Yet, it took less than five minutes for it to be a sobering reveal for Missouri.
To nobody’s surprise, MU slid down the overall seed list, but the hope had been it would halt somewhere in the vicinity of a No. 7 seed. Under those circumstances, the Tigers would avoid a potential second-round date with a death machine, whether it’s Gonzaga, Baylor, or Michigan. That did not come to pass.
By now, you know the selection committee plopped MU on the No. 9 line in the West Region, home to a group of Bulldogs from eastern Washington. Later on, the overall seed list showed coach Cuonzo Martin sitting at 33rd, essentially the highest-rated 9-seed, and at least five spots removed from relative safety.
Stumbling to a 4-6 finish put a damper on seeding, but for more than a month, we labored under the idea that MU’s impressive cache of Quadrant wins would act as insurance. Why? Well, the committee gave us a strong hint during an early bracket reveal that it valued so-called elite victories. For their part, the Tigers owned four of them in wins against four teams — Illinois, Alabama, Arkansas and Tennessee — situated in the top 15 of the NET rankings.
The notion that MU might fight its way back toward a protected seed ended a while ago. (A loss at Georgia probably put the kibosh on it.) And yes, predictive metrics — KenPom, BPI and Sagarin, painted a dubious picture of the Tigers’ overall body of work.
But Cuonzo Martin’s program met another committee brief: schedule ambitiously in non-conference, particularly when it comes to home and road games against top-drawer teams. Not only did MU follow that advice, but it knocked off Oregon and Wichita State. And even within SEC play, it picked up top-30 NET wins at Arkansas, Tennessee and Florida.
So, even if the analytic profile wasn’t blemish-free, there were enough on-court results to respectable seed. Bracketologists mostly concurred, consistently slotting MU as a steady No. 7 seed. They were off by a considerable margin.
So, what gives?
In subsequent interviews, committee chair Mitch Barnhart seemingly confirmed that the body’s methodology flipped. Strength of record helped it set the 68-team field, but predictive metrics like NET weighed heavily in seeds’ pecking order. Doing so explains why MU fell five seed lines in four weeks.
Put simply, the Tigers’ lackluster No. 47 NET rating served as the worst tiebreaker when MU’s resumé competed with other teams for seeding position. It turns out, Martin’s crew didn’t have robust enough coverage in its policy.
I understand how jarring that must seem. And the hackles it raised are also familiar. How much sway should metrics have? Shouldn’t on-court results be the ultimate judge? And what’s the point of lining up difficult opponents if the committee seemingly discounts it when it meets in a conference room at the JW Marriott Hotel in Indianapolis?
I’ll leave it to Sam to address those rhetorical matters. For this piece, though, I thought I’d try to clear up the confusion by peeking under the hood of Mizzou’s efficiency metric — and how it produced a frustrating outcome.
The approach: calculate the adjusted efficiency margin of each team slotted between the No. 6 and No. 9 lines in the seed list. In this case, I looked at how they fared in their Quad 1 victories and defeats. While NET ratings are listed, I used KenPom’s ratings to calculate the adjusted efficiency margins.
Take a look.
Calculating basic correlations, which are also unweighted, tells us that in this neighborhood, Quad 1 wins (r=0.42) and NET rating (r=0.38) were moderately related to a team’s overall seeding but who a team beat might have carried slightly more weight. That’s good for MU.
But look at the far-right column, which shows the Tigers’ adjusted efficiency in those games. Squarely in the middle of the pack, right? But MU also has more Quad 1 wins and played better than seven other teams, including six seeded in front of them.
Is the committee looking at each resumé with this level of granularity? No, but think of those possessions as an ingredient going into MU’s net rating. The Tigers did beat a bunch of good teams — their average opponent ranked fourth out of 16 teams — but they didn’t bludgeon any of them. (Its best result might have been a 13-point victory at Arkansas.) And it’s behind fellow SEC teams LSU and Florida.
Those victories helped MU rank 28th in strength of record (SOR) and eighth-best among teams seeded between No. 6 to No. 9 in the field. On those metrics alone, MU could build a sturdy case for a No. 7 seed.
Instead, the Tigers wound up seeded behind teams that had fewer Quad 1 wins, a lower SOR, and played worse in the quality wins they did piece together: Texas Tech, UConn, and Oklahoma. And you could make a compelling case that MU should have been ahead of Florida, which has two fewer Quad 1 wins, a worse SOR, and lost to the Tigers head-to-head. The same could be said of Oregon, which does have a better SOR but only two close Quad 1 wins.
Debating the strengths and weaknesses of each team’s body of work, though, can lead you into a thicket. That’s why NET, which is a rip-off of KenPom, was instituted. If used properly, it would lend context to the discussion: Once we account for every possession a team played and try to account for the strength of schedule, who is better than whom?
So far, we haven’t factored in how losses might play a role. Let’s do that now.
When we sort teams by their adjusted efficiency in losses, MU’s sitting near the bottom. In nine defeats, the Tigers’ posted a minus-15 net rating, and even when we account for the quality of the opponent, MU’s efficiency is still equivalent to the No. 158 team in KenPom’s ratings.
Now, there’s no direct correlation between adjusted efficiency margin in losses (r=-0.07) and overall seeding. However, there’s a moderately negative correlation (r=-0.56) with NET rating, meaning that as efficiency in losses improves (the value gets higher), a team’s net rating also improves (the actual numeral gets smaller).
Mizzou’s the inverse.
Dropping four games by 10-plus points, including three by 15 or more, certainly didn’t help MU. But again, it’s who beat the Tigers that compounded the issue. The average KenPom rating of teams that topped the Tigers checked in at 46th, or fifth worst out of 16 teams. The Tigers were only ahead of Georgia Tech, Oregon, Loyola Chicago, and St. Bonaventure, and the latter two are disadvantaged because of conference affiliation.
Again, though, the question is how much of a penalty MU should face for those woeful performances. Notice how Florida (3.37) and LSU (3.92) were only slightly less terrible in losses than MU? The majority of their losses came in conference play, and each had at least two losses to SEC teams sitting lower than 60th in KenPom. LSU dropped four games by 13-plus points, and Florida had seven defeats by double-digits.
While LSU is five spots ahead of MU in strength of record, the Gators are eight places behind MU. Meanwhile, the Tigers still own more Quad 1 wins, and four of them are better than any outcome on the respective team sheet for their conference peers. Yet LSU and Florida both wound up seeded higher.
Again, the committee would never get this deep into the weeds, but they see the ultimate byproduct — a team’s net rating. This season, efficiency margins in losses (r=-0.56) had a better correlation with NET rating than its performance in Quad 1 wins (r=-0.35), meaning that the Tigers’ bad losses (again) potentially dragged down the program’s profile to a point where it was jumped in seeding.
The relative strength of the SEC as a whole might have contributed, too. Given that NET ratings mimic predictive models, a team saddled with a bunch of losses is insulated if the margins are close and come against highly-rated opposition.
For example, Texas Tech collected only four Quad 1 wins and had the second-worst adjusted efficiency in those victories of any team in our sample. However, the Red Raiders played in a rugged Big 12, and its close defeats came against solid opponents (average rating: 20) and produced an efficiency margin (13.36) that would rank 71st in KenPom. Those wins and enough “good” losses helped coach Chris Beard’s group end up 17th in NET and score a No. 6 seed.
Throughout the season, we talked at length about how every possession matters. You don’t get to pick and choose what’s included in the NET rating. It’s a stew, and every result — tasty or disgusting — gets tossed in the pot. So, you could argue that MU’s outcome is (somewhat) deserved. The Tigers’ quality wins and putrid road performances all factored into its NET rating.
To me, though, the question is how much sway NET should wield in seeding decisions. Clearly, I’m a fan of analytics and predictive metrics. They help focus and sharpen conversations and, at their best, ground subjective discussions around an objective set of data. That being said, they shouldn’t replace reasoned, human judgment, and part of me thinks that happened here.
Now, you can’t discount the pandemic’s influence. It put pumped shotgun shells into the schedules of some teams. Long before March, Barnhart also signaled that the committee was simply going to evaluate what was on paper. No debating hypotheticals or weighing out which teams were more impacted by COVID-related stoppages.
Somehow, it feels like they went beyond that, outsourcing the deliberations to whatever top-line number NET spat out. I’m sure the committee would disagree. But it’s hard to be sympathetic to their case when the tournament’s No. 6 seeds are lined up Nos. 17-20 in the ratings — a group featuring 11 combined Quad 1 wins to MU’s seven alone. Under those circumstances, you might scoff when Barnhart says results matter.
The Monday after Selection Sunday features another tradition: complaining. Fans groan about their team getting left on the stoop. They (usually) gripe about the region where the committee shipped their squad. And yes, they whine about the disrespect shown to them in seeding. (Editor’s note: I was one of those people)
Maybe Mizzou’s just a typical case of the latter. Or a weird outlier in a season that unfolded in unprecedented circumstances. Regardless, it’s easy to see why it might feel aggrieved.