This article was updated on March 14, 2011.
Meet Gene, a 36-year-old basketball fan. Officially, he's an editor at a sports magazine in New York City. Unofficially, he's the office pool master. Mid-March every year, Gene takes the picks, rounds up the money ($10 per entry) and the NCAA brackets, and sends out regular e-mail updates — to the 400-odd people who enter the contest.
You’d think running a sizeable pool year after year might give Gene some insights into how best to make his own picks, but the week before March Madness starts, he’s like every other sports junkie: watching as many games as he can, listening to analysts on sports talk radio and ESPN, and immersing himself in articles and statistics. Kansas or Kentucky? Duke or Syracuse? What’s better, an experienced coach or a strong point guard? Two years ago, he says, “My girlfriend was laughing at me, because I fell asleep on the couch making my picks the night before they were due.” So how did he do? He finished 410th out of 416. Ouch.
There’s the rub: The people who can debate the case for Jimmer Fredette or Jared Sullinger to be National Player of the Year don’t necessarily do any better at these contests than the admin who can’t tell a full-court press from a French press. In fact, informed fans often do much worse.
- On CBSSports.com: Get the NCAA Brackets
So the mathematicians and game theorists who study NCAA tournament results — and yes, there are several — advise you to tune out the talking heads. There’s a better way to fill out the brackets to give you better odds of winning. Here’s how to do it.
Work the First Round
Get a pencil and a paper copy of the bracket. Most pools are run on sites like CBSSports.com or ESPN.com, saving pool masters hours of tabulating results, but you can enter your picks online after you’ve worked it out by hand.
Fill in the No. 1 seeds to win in the first round, because in 25 years no No. 1 seed has ever lost to a 16. Then, you’ll use Vegas odds to pick the remaining favorites in the first round. This saves you from having to research which starting center has a bum knee, or which team is playing just a few miles from campus. According to efficient market theory, “all the available information about teams is available in the Vegas odds, and so things like location and injuries are already taken into account without having to look into the details,” says Jarad Niemi, an assistant professor in the department of statistics and applied probability at UC Santa Barbara.
Are there any 50-50 games? Games between 8 and 9 seeds usually are pretty close to even odds, as are 7 and 10 games. Leave them blank for now.
Zoom to the Top Spot
Now it’s time to pick your champion. Yep, skip all the middle rounds, and go to your title game. Here’s where you want to pick a mild upset.
Mathematicians say to pick either the third- or fourth-ranked No. 1 seed, or the first- or second-ranked No. 2 seed to be your champion. In other words, you want your sheet’s winner to be a team that is ranked either third, fourth, fifth or sixth overall in the country before the tournament starts.
Steer clear of the overall No. 1 and No. 2 teams, which everyone else is picking. No team is so dominant that it’s a sure thing — especially this year. And going slightly afield gives you a mathematical advantage. “If everyone in your pool thought the same way you did, then you would tie them all,” says Tom Adams, who designed the site Poologic.com. “If you diverted a bit, and picked a different champion, you would have a higher probability of winning.”
That’s because in a standard scoring pool, getting the champion right gives you a motherlode of points. Nail that, and you won’t have to be perfect in the rest of your sheet. (Also, you’ll maximize your payout, because you won’t have to split your winnings with a bunch of other folks who tied your score.) But if you pick the same champion as everyone else, the only way to win is to pick the right winners early on. “So many people pick the heavy favorites that you need to be amazingly lucky in early rounds in order to be the best,” says Bryan Clair, a professor in the department of mathematics at St. Louis University. “But picking a less favored winner, or even Final Four, you only need to get lucky with one thing.”
ESPN, CBS Sports and Yahoo are helpful. Wait until Tuesday to file your picks, then go to those sites, which run huge, millions-of-entrants pools, and see which teams have been the most popular choices. ESPN publishes a “Who Picked Whom” chart; generally, these patterns reproduce in smaller pools. Steer clear of the top two.
“This year it looks like Ohio State and Kansas will be the two favorites and then another three to five teams will have about the same chance of winning,” says David Letscher, another St. Louis mathematics professor, who, with Clair, wrote a paper on maximizing the probability of winning these mega pools. He publishes his picks — and the perceived probability of a team reaching the finals — every year. “I expect both Ohio State and Kansas to be overpicked, so I wouldn’t pick them to win it all. Look for one in the next tier of teams that is being picked by the fewest people.
“That doesn't mean you should have Ohio State and Kansas losing in the first or second round,” Letscher adds. “They’re very good teams. I will probably have them losing in either the elite eight or the national semifinals.”
Know Your Rivals
While you’re picking your champion, use a little bit of psychology. Work within 100 miles of a No. 1 seed, for example? Fuhgeddaboudit: You know Arthur from HR and Sue from sales are going to pick it, too. You want to be different — not highly improbable, but different enough so that you don’t have to share your winnings.
Also, people tend to like teams that they’ve seen play. “I live in a Big Ten town, so as a rule, the Big Ten is overbet,” says Bradley Carlin, a professor at the University of Minnesota who has co-authored two papers on NCAA tournament probabilities. “Because everyone has seen a lot of Gopher games, they’ve seen Purdue, Indiana, Illinois. They’re thinking, ‘I saw them play the Gophers, and they were pretty good that day.’”
Use Online Tools
More than a decade ago, Adams, a computer scientist, spent some spare time pondering how to maximize points in a pool. He came up with a calculation for it and designed a Web site that spits out the ideal brackets. On Poologic.com’s calculator, you can enter the scoring system of your pool, pick from one of five probability models, which Adams describes on his site, and constrain yourself to the winner you like. Click “Calculate Now,” and the site produces the bracket designed to give you the highest score. Adams encourages people who win using the site’s picks to make a donation to The V Foundation, a cancer charity. He estimates his users have donated $25,000 to that organization so far.
Those probability models are where the really mathy folks have made their mark on this process. These guys rate teams according to scoreboard data: wins and losses, home court advantage, opponents’ strength, and in some cases, margin of victory. Two of the most popular, which Poologic incorporates, are Jeff Sagarin’s and Joel Sokol’s. Sagarin, an M.I.T. grad, has been publishing ratings in various sports in USA Today since 1985, and you can see his college basketball picks here. Sokol, a Georgia Tech professor, makes his calculations based on a different formula, LRMC, which stands for logistic regression and Markov Chain. Sokol says he’s analyzed data from 10 years of NCAA tournaments and “it’s statistically significant that our system predicts more winners.” Play around with that function, marvel at the brainpower behind them, and choose what you like.
Change a Few Games
Now, you could go back to your sheet and enter Poologic’s picks verbatim. But you want to put your own mark on it. Here’s where you go back to that first round. When No. 8 and No. 9 seeds play, as well as No. 7 vs. No. 10 seeds, those games tend to be statistical dead heats. Go to CBS’s Matchup Analysis, or ESPN’s Tournament Challenge and check that table again to find cases where one team is vastly overbet. Say an 8 is favored over a 9 by 60 percent of the entrants: In that case, pick the 9 seed.
“It’s almost like arbitrage,” says Letscher. “We’re trying to find those little inconsistencies in the pool.”
But don’t go nuts picking upsets. Typically that team is going to lose the next game anyway. Says Clair, Letscher’s colleague: “People tend to pick upsets early and go conservative in the Final Four. If you’re in a big pool, you need to flip-flop that: Go boring in early rounds, and slightly strange in the Final Four.”
Every statistician we talked to enters more than one sheet per pool, if the rules allow. They simply change the champion in each sheet and perhaps flip a few more early-round games, like 8 and 9 seeds. (You could also flip a couple of winners of second round or Sweet Sixteen matchups.)
If you’re willing to shell out a few more bucks, consider following suit. Just know that this could annoy the pool master, especially if he or she is tabulating results by hand. Gene from the sports magazine had 12 sheets from one person a couple of years ago. “It’s kind of against the spirit of the pool and the tournament,” he says. But you’ll increase your chances of finishing in the money.
Let the Games Begin
Once your brackets are turned in, you get to kick back and watch the games themselves. Here’s where you may notice the downside to the contrarian approach: Favorites often win. In 2009, North Carolina was the popular pick, and the team came through.
But picking against favorites isn’t just a good betting strategy — it’s also very much in the spirit of March Madness. “It’s fun to watch these games you could otherwise care less about when you’re rooting for the underdog,” Niemi says. “At worst, it does no damage to your pool sheet; at best, it can help you.”
Will you feel like an idiot if you’re not in the running going into the Final Four? Possibly. You can handle it. Says Niemi: “If Kansas goes the whole way, everyone else will say, ‘Couldn’t you see this coming?’ And you tell them, ‘Wait till next year.’”
More on MoneyWatch:
for more features.