March Madness 2019 Preview

With March Madness right around the corner, I thought it would be fun to visualize a bit of data about each region and team, with the end goal of being a bit better informed for filling out my bracket! The data below is mostly drawn from Ken Pomeroy’s college basketball rankings, and is intended to give a brief overview of the teams and regions in this year’s tournament.

Comparing Regions

NCAA Tournament Regions

Right off the bat, one of the things that jumps out is how tough the South region is from top to bottom. While Virginia stands alone in terms of combined offensive and defensive efficiency, the South possesses several other teams with particularly suffocating defenses (Kansas St. and Wisconsin) or electric offenses (Tennessee and Purdue).

By virtue of drawing Michigan State, Duke got the toughest #2 seed of any #1. While the rest of the East is relatively weak compared to the other regions, Michigan State has actually performed better on offense than Duke this season, and appears set to give Zion and Co. a run for their money.

Finally, the West boasts the two best defenses in the country in Texas Tech and Michigan, either of which would serve as a compelling foil to a Gonzaga team which led the country in offensive efficiency.

Region Avg. Team
Efficiency
Avg. Offensive
Efficiency
Avg. Defensive
Efficiency
Cumulative
Efficiency
East 13.77167 111.1889 97.41111 247.89
Midwest 15.65625 112.6688 96.99375 250.5
South 16.65312 112.6437 95.98125 266.45
West 14.59444 111.5056 96.90556 262.7

Top Seeds

Top Seeds

By comparing a team’s tournament seed with it’s season-long efficiency, we can look to see which teams were “under” or “over” seeded by the NCAA committee. While the tournament will be the ultimate arbiter, it’s clear that Auburn and Virginia Tech were particularly under-seeded by the committee. Based on their regular season performance, each deserve to have been slotted one or two seeds higher than currently placed. Louisville has an argument there as well.

On the flip side, LSU and Houston were both noticeably over-seeded relative to their performance during the year. They’re not the worst though - Marquette, the #5 seed in the West, actually rates out as worse than both #6 Buffalo and #7 Nevada, and #4 Kansas rates below #7 Wofford. Come on committee!

Clustering Teams

Team Clusters

In the chart above, I wanted to show how closely the different teams in the tournament are across a number of dimensions. To do so, I used k-means clustering to break the 68 tournament teams into 10 different clusters, with offensive and defensive efficiency, tempo, luck, and strength of schedule used as the clustering dimensions. A few fun observations:

  • Gonzaga and Wofford are clustered together, but with no other teams - I think this the result of both being small-conference teams with marginal strength of schedule but great offenses
  • Virginia sits in a cluster of its own due its unusual pairing of an extremely slow tempo (most deliberate in the field) paired with outstanding efficiency
  • North Carolina and Duke have the first and third-fastest tempos in the field, an interesting juxtaposition against Virginia for the three ACC #1 seeds
  • Kentucky and Michigan State are also in Carolina and Duke’s cluster, officially turning it into the “Blue Blood” cluster. Given the track records of these programs, it’s tough to bet against any of them to win it all!