The New York Times
Printer Friendly Format Sponsored By


March 13, 2006

Don't Always Follow the Crowds

By BRYAN CLAIR and
DAVID LETSCHER

Connecticut and Duke, here we go again. These teams received top seedings in the N.C.A.A. men's basketball tournament, so pundits across America will fall all over themselves predicting a Duke-UConn battle for the national title.

They should, too. Duke and UConn have led the polls all season. They have great coaches, great players and a history of tournament victories. But if you want to win your office pool this year, you might be better off picking someone else.

To win a pool you don't need to pick the winner of every game, you just have to beat your co-workers. They know what everybody knows: Duke and UConn are mighty fine teams. But a basketball pool, like a horse race, is parimutuel: the more people who back a team, the less that team helps you win.

The question is: do you back the favorite or go for the long shot?

In a small pool, with you and a handful of friends, the right strategy is to be really boring. Pick good teams to win, leave upsets alone.

In a 50-player office pool, other players' picks make a difference. Avoiding the crowd can be worth the risk.

In a huge pool, like the million-player online pools run by Yahoo and ESPN, choosing the favorites gives you almost no chance of winning.

Thousands of players end up picking Final Fours involving the big favorites. The only way those players can win the pool is to have near-perfect picks in early rounds.

To avoid the crowd, you need to know what the crowd is doing. The large online pools often tell you, posting the percentage of players picking each game, even before the games begin.

What you'll find is this: People like to pick favorites. Over the past five years, 41 percent of the fans picked the highest-ranked team in the country to win the tournament, and about 17 percent picked the second-ranked team. The other top-seeded teams will have much smaller followings.

Last night on Yahoo Sports' Tournament Pick'em, 32 percent of the entries had selected UConn and 28 percent had chosen Duke. Villanova, which also received a No. 1 seeding, was next at 9 percent, followed by North Carolina, seeded No. 3, at 5 percent.

Are the top teams really that good? Maybe. Since 1985, the No. 1- and No. 2-ranked teams in the Associated Press poll combined to win the championship eight times, about 38 percent of the time. On the other hand, in 16 of 21 years, one of the A.P. top two has missed the Final Four.

The tension between quality picks and crowd avoidance makes for an interesting math problem. After the 2003 tournament, when more than half of America's pool entries had their Kentucky picks go down in Round 4 (the Wildcats were a No. 1 seed at the time), we set out to answer it.

The big online pools let you look at other people's picks. We took advantage of this, and looked at half a million. While each individual has a strategy, groups of people pick as if they were tossing weighted coins. Whether you're a hoops aficionado or you pick your teams for their mascots, your pool score will land on a bell curve. We take data from the online pools, feed it into a computer, and estimate where that curve will be.

Using this data, we formulated a strategy to find the picks that maximize the chances of winning a pool and wrote software to make our picks. With about nine quintillion possibilities, the search takes a little time. But it runs fast enough to get our pool brackets in on time.

In the last two years, we've noticed some patterns in our optimized picks. These tips may give you an edge when filling out your bracket.

First, think differently in your Final Four. One-quarter of the country concentrates on the five most likely Final Fours. Decide which teams are being overpicked and which are being underpicked. Put a good but unpopular team in your Final Four. If it wins, you'll score points that few others will get.

One of the big favorites may not make the Final Four in Indianapolis. It has happened before, so try picking it. These Final Fours are nearly as likely and keep you away from the most common picks.

In the early rounds, pick the better teams to win. So far, our methods have never suggested picking a team seeded lower than 10th to win in the first round.

There's a popular belief that you should always pick a team seeded 12th to win in Round 1. Although the 12's have a history of success, there's not much benefit to scoring an extra point. Take your risks in games that really count: the Final Four. If your unpopular Final Four comes true, the rest of your bracket won't matter at all.

Unfortunately, picking underdogs in the Final Four means that in most years, you'll take it on the chin. You're going against the favorites, so be prepared for some miserable scores when these teams do win. Last year, with heavily favored Illinois and North Carolina in the championship, crowd-avoiding picks were especially embarrassing. But don't be deterred; it was the first time that had happened since the field expanded to 64 teams.

This year, if Duke or UConn cuts down the nets, expect to be sitting at the bottom of the pool. But get one or two breaks in the Final Four, and you'll go home a winner.

Bryan Clair and David Letscher are professors of mathematics and computer science at St. Louis University. Follow their N.C.A.A. picks, posted by Wednesday, at http://dehn.slu.edu/sports.