by Danny Blumenthal
Embed from Getty ImagesDespite 36 points from star Luka Garza, Iowa was upset in the 2nd round of the NCAA Tournament. Photo by Jack Dempsey/Getty Images
March Madness has been especially mad this year. For the first time in the 64-team era, four teams seeded 13th or worse won in the opening round, and no tournament has had a higher sum of seeds (94) in the Sweet 16. But despite the abundance of upsets, no two underdogs have won the same way. Syracuse has ridden Buddy Boeheim’s hot shooting to back-to-back victories, while 14-seed Abilene Christian overwhelmed the Texas Longhorns with its swarming defense. This year’s most unexpected Cinderella, Oral Roberts, has relied almost entirely on two stars. Both Max Abmas and Kevin Obanor (who haven’t missed a single minute of the tournament) have scored 25 or more points in consecutive games, while no other Golden Eagle has recorded even ten.
For success in March, is it better for teams to adopt the star-focused structure of Syracuse and Oral Roberts, or should teams opt for Abilene Christian’s balanced approach? Anecdotally, a case could be made for both systems. It’s easy to remember superstars like Kemba Walker getting hot in the tournament and carrying their teams to the title. And given the smaller number of players on the floor and the fact that they play both offense and defense, top players have an outsized impact in basketball compared to other sports. Nevertheless, it may be easier for coaches to game-plan for a team with only one main contributor. In addition, if a team’s leader gets into foul trouble or is injured, the team may be thrown off its game and subsequently fall short of expectations. With potential benefits to both team-building approaches, HSAC decided to dive into the data to find whether one team style is better come March. Do teams with a star player typically exceed expectations?
The initial definition of a “star player” is simply one who scores a lot of points. In this case, players who averaged at least 20 points per game were deemed stars, since they almost always led their team in scoring. To determine whether teams which built around stars performed better in the tournament than those which didn’t, a t-test (used to identify differences in averages), was implemented. In the first test, every tournament team since 2002 was split up by whether they had a 20+ point scorer or not, and the outcome variable was Performance Above Seed Expectation (PASE; from barttorvik.com). This measures how far a team advanced in comparison to previous teams on that same seed line. The results are displayed below:
Clearly, there is no significant difference (p = 0.79) between teams with and without high-scorers in terms of performance against expected. In both groups, the PASE was almost exactly zero. Similarly, if the outcome variable is changed to overall tournament wins, the teams are almost indistinguishable. Teams with a star scorer average 0.950 tournament wins, while those without one rack up 0.947 wins on average. For every Luka Garza (36 points in a 2nd round loss to Oregon), there is a Marcus Zegarowski (a game-high 20 points to lead Creighton to the Sweet 16).
However, one could argue that points scored alone is not a good enough metric to identify whether a team is reliant on one player. There are many other methods that could be used to do so, including share of team’s points scored and minutes played. In addition, some players may not rack up as many points, but they still may play a lot and affect the game in other ways, such as drawing fouls. To acknowledge these types of players, usage rate was included as a dependent variable as well. This is a more holistic measure which indicates the share of a team’s possessions which end with the ball in a player’s hands (either by way of a shot, foul, or turnover). On this metric, players with usage rates of 28% or higher were deemed “go-to players” in order to match the designations used by KenPom. As in the points scored case, a t-test on PASE between teams with and without a “go-to player” was run. This time, results were restricted to the past decade, because that is as far back as Sports-Reference’s usage rate data goes. The results of this test are graphed below:
There is a very slight lean towards teams with a “go-to player” performing better in the tournament, as the PASE difference is +0.06 wins for teams led by a star. But while this difference is a little larger than the one derived from defining stars by only points scored, it still is not statistically significant (p = 0.48), and neither type of team performs significantly better than expected. It appears that no matter how a star player is defined, having a star does not affect whether a team succeeds in the tournament.
So while star players might get all of the press, don’t count on them to carry their teams to the title. And despite the struggles of teams led by superstars Cade Cunningham and Luka Garza, don’t listen to the cynics who say that depth is essential for tournament success. In the end, other factors, such as turnovers, team balance, tournament experience, and offensive rebounding, are all probably more important.
Editor’s Note: If you have any questions about this article, please feel free to reach out to HSAC at harvardsportsanalysis@gmail.com. Thank you for reading!