W. Hoops Seeks Good Seed

When junior Tricia Tubridy of the Harvard women’s basketball team was asked about the Crimson’s NCAA seeding chances shortly after being presented with the Ivy League championship trophy, she took a moment to answer. Her response was lacking in optimism.

Tubridy first thought back to the first half of the season, when Harvard suffered its only four losses of its season—all to Top 25 teams—by an average of 35.5 points.

“I think the blowouts early in the year are going to hurt us a little in our seeds,” Tubridy said. “They’ll be in the back of people’s minds, but we’re obviously on the right track now.”

If history is any indication, however, Tubridy need not worry about the blowouts affecting the Crimson when the selections are announced live on ESPN on March 16. Her intuition relied on the assumption that the NCAA selection committee will actually put thought into Harvard’s seeding, but there’s little evidence in support of that belief.

The seeding practices of the past three years have shown that the committees don’t do much thinking about the bottom third of the bracket. Instead, they let the computers do all the work.

And the good news for Harvard is that the computers the selection committee uses don’t care how badly the Crimson lost to Rutgers, Vanderbilt, Minnesota and BC. They use the ratings percentage index (RPI)—a weighted average of a team’s record, its opponents’ records and its opponents’ opponents’ records. Wins and losses are all that count.

The committee’s selection of 33 at-large teams for the NCAA tournament and seeding them into four 16-team brackets along with 31 automatic qualifiers is often mired in controversy. When the committee has to make the tough decisions—picking the No. 1-4 seeds and selecting the final at-large teams—it uses more than just than the RPI. In fact, the RPI has been a poor predictor of these decisions.

But when seeding teams from the 18-or-so conferences like the Ivy League which typically send just one team to the tournament, the RPI has been an almost perfect forecaster for the bottom third of the bracket.

Last year, the RPI correctly predicted the seedings of the 18 automatic qualifiers in the bottom third of the bracket. For example, Harvard’s RPI projected to a No. 13 seed, which was the seed the Crimson received.

The RPI’s predictive power wasn’t unique to last year either. In the last three years alone, only two of 50 automatic qualifiers in the bottom third of the bracket received a seed lower than what it would have predicted. That’s a success rate of 96 percent—over four points better than the best free-throw percentage in the Ivy League.

The question now is, where does the RPI predict Harvard will be seeded this year? That depends largely on how the Crimson performs in its two remaining games at Yale and Brown this weekend.

Jerry Palm’s web site www.collegerpi.com posts RPI calculations on a daily basis, and right now, his calculations list Harvard’s RPI at No. 42 in the nation, ahead of 19 other projected automatic tournaments qualifiers.

This means that if the Crimson can beat Brown and Yale this weekend, Harvard will be at least a No. 12 seed, which would be the highest ever in the history of Ivy women’s basketball.

And Harvard could do even better for a couple reasons. If there are upsets in conference tournaments, teams with lower RPIs than Harvard could make the tournament, which would push up Harvard’s seeding. Also, Palm’s site suggests that the NCAA selection committee actually uses a weighted RPI which rewards good wins and penalizes bad losses. Few of the Crimson’s competitors have bad wins, and most of them have bad losses. Harvard, however, is one of just seven teams nationwide that has yet to lose to a non-Top 25 team. Such considerations suggest that Tubridy’s worries about the early-season blowouts were overly pessimistic.

A higher seed will provide several benefits to Harvard. It will give the Crimson an easier opponent in the first round and lower the likelihood that this game will be on its opponent’s home court. Palm’s latest bracket projection is among the best site arrangements Crimson could hope for. He projects Harvard at No. 12, playing South Carolina at Old Dominion, who would earn a No. 13 seed. Other nearby sites where Harvard could end up include Penn State, Cincinnati, NC State, Georgia and Purdue.

Harvard coach Kathy Delaney-Smith gave her take on Harvard’s seeding on Saturday. She thinks her team should be no worse than a No. 13 seed if it wins its last two games, though she chooses not to spend too much time dwelling on possibilities.

“[If] you predict those things…then you’re angry because the committee didn’t see it the same way you saw it,” Delaney-Smith said.

Not to mention, the Crimson needs to focus on winning two games next weekend. The incentive to go undefeated in the Ivy League and earn the best seed in league history gives the team plenty to play for, and the players know it.

“We want to win these games,” Tubridy said “We want to win them handily. We want to send the message that we’re a different team than we were in December.”

—Staff writer David R. De Remer can be reached at remer@fas.harvard.edu.