Compared to the NBA’s 82-game slog and seven-game-series playoffs, the college basketball season is incredibly short. A few unlucky bounces can be the difference between a bubble team going dancing or going home; one cold shooting night in March can end a title contender’s season at the hands of a Cinderella mid-major.
Even at the single-game level, variance in college basketball reigns supreme. Each game’s outcome is a product of two factors: skill and luck. If basketball games were entirely reliant on skill, the best teams would win 100 percent of the time, and there would be no reason to watch the sport. Luck in basketball — specifically when it comes to shooting variance — is what gives underdogs a shot to pull off major upsets and lends the sport its excitement: what is the most exciting shot in basketball (a half-court buzzer-beater), if not a wildly improbable outcome?
The goal of college basketball ranking systems is to serve as a technological sifter: a perfect ranking system would be able to look at a series of game results, filter out all of the luck-based variance, and return an objective list of the best and worst teams in the sport.
No algorithm can do this perfectly — frankly, very few even come close — but they can still offer valuable insights into team performance. Here, we’ll look at the differences between two algorithms to determine which teams have overperformed and underperformed shooting the ball this season.
The first data point we’ll use is KenPom’s Adjusted Efficiency Margin, the gold standard in college basketball team ratings. Adjusted Efficiency Margin estimates how much a team would outscore the average D-1 basketball team by over the course of 100 possessions. The formula is complicated, but its inputs are fairly intuitive: beating good teams by lots of points boosts AdjEM, while losing to bad teams tanks it.
The second algorithm is ShotQuality’s own Adjusted Shot Quality. While KenPom ranks teams based on a team’s actual performance, Adjusted SQ ranks teams by expected performance based on quality of shots taken. It computes each team’s expected points per possession based on their shot selection using a unique algorithm, does the same thing for their opponents, adjusts each value for the level of competition, and then calculates the difference between those two values. The result, Adjusted SQ, is a measure of how many more points per possession a team is expected to score against an average opponent.
By applying these algorithms to every team in Division 1, we can compare a team’s expected performance (Adjusted SQ) to their actual performance (Adjusted EM). From there, we can compute a composite ranking for each team by averaging their Adjusted SQ ranking with their Adjusted EM ranking and dividing by two. For example, since Gonzaga ranks #1 in both AdjSQ and AdjEM, their composite ranking is (1+1)/2 = 1. No surprise there — the Bulldogs are the undisputed best team in college hoops at the moment. For other teams, there’s a bit more variance: Purdue ranks #2 in AdjSQ and #11 in AdjEM; splitting the difference gives them a composite rank of 6.5. Now that we have this list of college basketball teams sorted from best to worst, we can filter it however we like. For the purpose of this analysis, we’ll just look at the top 50 teams in composite rankings — in other words, the 50 best teams in college basketball.
From this list, we can select the five biggest overperformers and underperformers by subtracting AdjEM rank from AdjSQ rank. The teams at the top here will be the season’s biggest overperformers: squads whose on-court performance exceeds their expected performance based on shot quality. The reverse is true for teams at the bottom: these teams haven’t performed as well as expected, based on how much they dominate the quality of shots.
Let’s get into the two lists, starting with the biggest underperformers:
Indiana tops our list, as they rank 26th in Adjusted Shot Quality but just 47th in Adjusted Efficiency Margin. They’re currently on a four-game losing streak, tanking their KenPom ranking; during those four games, Indiana has shot just 25% from three. Using raw ShotQuality scores, Indiana’s actually been projected to win in two of those four losses — Northwestern and Michigan St. Some bad breaks for Indiana have them on the bubble, per KenPom, while their actual performance given average shooting luck should have them sitting around a 6- or 7-seed.
Seton Hall comes in second here. Their record is currently a middling 15-8, while ShotQuality projections would have them at a more impressive 17-6. Some of their poor luck is attributable to opponent free throw percentage: opponents have shot 78.7% at the line against Seton Hall, the second-highest opponent FT% allowed in the country. This stat — totally out of their control — has cost Seton Hall some close finishes.
Wisconsin ranks third on the list, with their elite ShotQuality rank of 17th translating to a more middling 31st place in Adjusted Efficiency Margin. However, what Wisconsin lacks in overall shot-making luck, they’ve more than made up for with their performance in close games: in five contests that ShotQuality deemed toss-ups (each team had a 50% chance to win), the Badgers are 4-1.
Connecticut and Mississippi State round out this list. While Connecticut ranks as the 8th-best team in the NCAA at defending the three according to ShotQuality, they’re only allowing the 192nd-lowest three-point percentage — a season-long streak of bad luck. Mississippi State is looking at an uphill climb to make the NCAA tournament with their 5-6 conference record; however, if they’d won their games against LSU and Mississippi like ShotQuality expected them to, they’d be sitting pretty at 7-4 in the SEC.
Now, let’s look at the biggest overperformers:
LSU sticks out like a sore thumb on this list, with an Adjusted Efficiency Margin rank of 16th but an Adjusted Shot Quality rank of just 49th. They’ve won three games more than SQ’s projections expected, mostly due to defensive overperformance: while their defense ranks just 17th in ShotQuality Points Per Possession Allowed, it ranks 1st in KenPom’s Defensive Adjusted Efficiency Margin.
Duke comes in a distant second — both ranking systems agree that they’re a great squad, but only KenPom ranks them in that elite top-10 tier. The Blue Devils have benefited from opponents shooting just 65.8% from the line against them, the 11th-lowest mark in the NCAA. Like LSU, they’ve also overperformed their SQ expected record by three games.
USC is up next on this list. ShotQuality and KenPom agree on the proficiency of USC’s two-point defense — they rank second in 2P% allowed and 4th in expected defense against shots at the rim — but a bit of shooting overperformance on offense this year compared to SQ expectations places them in this group of overperformers.
Arkansas and Memphis round out the rankings, each overperforming their AdjSQ rank by eight spots in AdjEM. Neither of these teams has exceeded SQ expectations too drastically — in fact, Memphis has actually been unlucky on defense this year according to ShotQuality. Instead, both teams have slightly overperformed their expected three-point shooting percentages, moving them up a notch in the KenPom rankings.
All data as of Wednesday, February 16.