The statistical lessons of Harvard's tremendous NCAA tournament upset victory
All across the country, thousands upon thousands of people are staring in disbelief at their NCAA tournament brackets, cursing the day they underestimated the Harvard Crimson.
Almost no one expected Harvard — the Ivy League champion and a 14-seed in this year's tournament — to do much of anything against third-seeded New Mexico. The Lobos finished the regular season with a hefty tally of 26 wins and a Mountain West conference title, and entered the tournament as a dark-horse pick to go to the Sweet Sixteen and beyond. But there was one flaw in New Mexico's game that Harvard was perfectly built to exploit: The Lobos had been awful at defending three-pointers all year, and the Crimson had spent all season lighting teams up from beyond the arc. That's exactly how the Crimson pulled off the upset, nailing eight of 18 three-pointers in a 68-62 shocker.
As Harvard moves on and New Mexico goes home, analysts will surely debate whether the Crimson's win was a total fluke or an upset pick that we could have seen coming based on the beyond-the-arc mismatch. But really, just how possible is it to see a result like this coming? And is there a common thread linking a bracket buster like Harvard to past double-digit seeds who have marched deep into the tournament?
First of all, it's important to identify exactly what it is that helps a team win. So let's turn to Ken Pomeroy, a statistician who's become the Nate Silver of the college basketball world. He compiles reams of advanced metrics on all 347 Division I teams, and ranks each team by offensive and defensive efficiency — that is, just how good that team is at making and denying shots. Within those calculations are four key offensive figures: Effective field-goal percentage (a measure of shooting that gives extra value to three-pointers), turnover rate, offensive rebounding rate, and free-throw rate. In essence, Pomeroy charts how good a team is at shooting from all over the floor, protecting the ball, recovering its own misses, and getting to the foul line.
The high seeds of the last 10 years differ in terms of how they produce points or prevent them. But on the whole, the bracket busters share a few things in common: They have high effective field-goal percentages, they shoot well from all spots on the floor, they're strong enough defensively to keep the top teams from pulling away, and they're excellent at preventing turnovers.
Take Virginia Commonwealth. In 2011, the 11th-seeded Rams made it all the way to the Final Four, leaving a trail of big-name programs in their wake. But their success wasn't built with smoke and mirrors. By Pomeroy's metrics, VCU was one of the nation's top teams when it came to offensive efficiency, as they shot 37 percent from three-point range and only turned the ball over on 17 percent of their possessions.
That combination of three-point prowess and protecting the basketball worked to the Rams' advantage in a third-round upset of sixth-seeded Georgetown, a team that struggled to create turnovers and was merely average at defending against the three. In that game, the Rams attempted a whopping 25 threes — over half their shots came from beyond the arc — and hit 12 of them, while only turning the ball over six times.
In 2006, 11th-seeded George Mason stunned the sports world by knocking off powerhouses Michigan State, North Carolina, and Connecticut en route to a Final Four appearance. Just like the Rams, the Patriots were a highly efficient team on offense — 24th best in the country, according to Pomeroy. But unlike VCU, George Mason didn't rely as much on the three-pointer. Instead, they did their damage inside the arc, hitting 54 percent of their two-point shots and holding teams to 43 percent on two-point attempts, as well as 32 percent from three. In their third-round 65-60 win over defending champion North Carolina, the Patriots worked their gameplan to perfection, hitting 45 percent of their two-point shots and allowing the Tar Heels to shoot just 36 percent, and only 33 percent from three-point range. In the Elite Eight against Connecticut, George Mason shot 50 percent from inside the arc while shutting the top-seeded Huskies down on the perimeter.
Those same advanced metrics made a good case for Harvard against New Mexico. Harvard's one major weakness — its defense against two-point attempts — wasn't really threatened by New Mexico. The Lobos shot a woeful 46 percent on their attempts inside the arc this season, one of the worst rates in the nation. That let Harvard concentrate its defense on the perimeter, where it kept New Mexico to 2-14 shooting from three. Meanwhile, the Crimson's hyper-efficient offense lit up the Lobos from inside and out, hitting 58 percent of their two-point shots and 44 percent of their threes, a performance better than but not too far out of line with the team's season averages.
Not all double-digit seeds are built the same. For every Harvard or VCU, there's a Belmont or Montana that gets drilled to the surprise of almost no one. But as the Crimson helpfully reminded the sports world Thursday, sometimes it helps to look past the seed and into the numbers. You might just find a hidden gem.