Picking and Grinning: Secrets to Successful Bracketology

By Dan Steinberg
Sunday, March 16, 2008

Discussing Dan's Theories -- Q & A Transcript | Research Your Picks -- NCAA Tournament Database
Kentucky 1996
One golden rule of 21st Century sports blather goes like this: Don't write about your fantasy football team. Your kicker's tragic missed extra-point or your miracle Monday night win were surely dramatic, but no one cares. Sorry.
A similar rule applies to the NCAA tournament. Convoluted stories of past failures and successes quickly lead to glassy eyes. And yet I'm going to tell you such a story anyhow: during one glorious month in the spring of 1996, I clinched victory in two NCAA tournament pools before Kentucky (right) even played the national championship game. I still love Erick Dampier.
Three years later, I moved to the District and made a pledge: if I nailed the Final Four, I was going to write letters to Michael Wilbon and Tony Kornheiser demanding that they hire me to work for this paper. I had Duke, Ohio State, U-Conn. and Kentucky. The first three made it. Kentucky lost in the regional final. I never wrote the letters.
My primary guiding principle in filling out a bracket goes like this: I don't care if you know a lot about college basketball. I don't care if you can name the sixth man on the eighth best team in the Pac 10, or if you saw 17 games every weekend of February. Bracket-changing games go down to the final second every year, and no amount of expertise allows you to predict whether, for example, UCLA's Tyus Edney will hit a last-second shot to eliminate Missouri in the 1995 second round, allowing the Bruins to survive long enough to win the national championship.
That being the case, I worry less about picking every head-to-head result correctly and worry more about following broad principles that have worked for me in the past. I'm no math major; heck, it's possible that all of my principles for success are statistical goop. But having solicited and graded hundreds if not thousands of brackets from friends and family members over the past 15 years, I feel like I have at least a few guiding principles worth sharing.

I know what you've heard: almost every year, an underdog 12 seed beats a favored 12 seed. For the next four days you'll hear experts debate which 12 is most likely to pull such an upset. Push mute. Ignore them. In tournament history, 5 seeds are 63-29 in the first round. Yes, that's less than you might expect for such a high seed, but still, a 68 percent winning percentage is all the argument you need to stick with all four favorites. Recent years have been even better; the five seeds are 9-3 in the past three tournaments.
This advice, I should note, is based on standard scoring systems. If you're in a pool that offers big bonuses for upsets, this rule would no longer apply. But if, on average, almost 75 percent of five seeds survive the first round, you're simply better off taking all four favorites and accepting one miss than blindly chasing after that one 12th-seeded winner.
(In fact, as shameful as it might feel, don't be afraid to play it safe by embracing virtually every higher seed in the first round, guaranteeing yourself a spot in the soft middle of the standings and living to fight another round. Excluding the 50-50 games between 8 and 9 seeds, higher seeds have gone 26-2, 20-8, 23-5 and 25-3 in the past four seasons. That's a winning percentage of 84 percent. That's good enough. For two exceptions to this advice, see rules 3 and 7 below.)

On a related note, it's terrible but true; the big boys have an incredible rate of success against the upstarts. As fans, we all love the George Masons of the world; as bracket entrants, we should shun them.
Take the Missouri Valley, for example, possibly the most celebrated of the so-called "mid-major" leagues. The conference has produced some notable tournament teams in recent years (think Southern Illinois and Wichita State in the past, and possibly Drake, right, this year), and yet its composite first round record since 2003 is an abysmal 4-9. Likewise with the Mountain West, an atrocious 3-9 in first-round games since 2003. And likewise with perennial West Coast Conference upstart Gonzaga, whose total tournament record in those years is a blah 5-5.
On the other side of the coin is the ACC, which is 23-3 in first-round games over that span. Now, might a 10th seeded ACC team lose a first-round game? Sure, look no further than Georgia Tech last year. But 23-3 sort of speaks for itself.
Further, the six major conferences, plus Memphis, have accounted for 35 of the past 40 Elite Eight teams; two of the five gate-crashers have since joined a major conference. So for both the first round and the end game, swallow your underdog dreams, toss out your fairy tale metaphors and give the power conferences a big sweaty hug.

I know, broken record, but it's true; give another sweaty hug to the big boys. The lesser lights in the best conferences often provide the choicest upsets in the tournament's second and third rounds, which is also where the bracket standings begin to separate.
Last year, among the tournament's biggest surprises was Vanderbilt, which defeated Washington State behind Derrick Byars, right, and made the Sweet 16 as a 6 seed. Two years ago, seventh-seeded Georgetown and sixth-seeded West Virginia advanced that far. In 2005, 10th-seeded N.C. State was one of two double-digit seeds in the Sweet 16 and seventh-seeded West Virginia was the biggest surprise in the Elite Eight, and in 2004 it was eighth-seeded Alabama. The 2002 event also had a major-conference surprise, 12-seeded Missouri, which became the lowest seed ever to make the Elite Eight. All of those schools come from power conferences.
Do those big-time programs have any similarities? It's not perfect, but they often share a proclivity for the long-distance shot. Five of the seven teams referenced above finished in the top 25 in three-pointers made per game. Both West Virginia teams and Missouri were in the top 10. Last year's Vanderbilt team was 13th in the country in three-pointers made per game. West Virginia ranked second and sixth in that category during its two surprising tournament runs. The 2005 N.C. State team was 21st in the country in three-pointers, while the 2002 Missouri team ranked seventh nationally in that statistic. Bear in mind that there are more than 300 Division I teams, and yet each of these four teams was in the top 25 in three-pointers made.
So, you're now looking for a big-conference, lower-seeded, three-point-shooting team to pull an upset or two this March? Try ninth-seeded Oregon (26th in the country in threes), fourth-seeded Vanderbilt (17th in the country in threes), 11th-seeded Baylor (27th) or fifth-seeded Notre Dame (31st).

I see it happen every year: Bracket pushers believe they're showing their courage by chasing, say, an 11-6 first-round upset, which, even if correct, would do virtually nothing for their prospects in most scoring systems. But then this courage goes out the window in the later rounds, when courage is the only thing that matters, and they pick four one seeds to advance to the Final Four.
Remember the last time even as many as three one seeds made the Final Four? It was 1999. Since then, 11 top seeds have made the Final Four, barely more than the eight second seeds that made it in that time. But still, the masses embrace the ones, and so if you hit the right two seeds, and even the occasional oddball three or four, suddenly you're speeding past the rote bracketeers whose Final Four looked like a prison calendar.
Now, will this principle occasionally make you look foolish? Absolutely. If the top seeds actually do come through, you're entirely sunk. But if you're going to take a few risks, they might as well be risks that would mean something if they actually come to pass.

I debated whether even to include this rule, because I struggle to follow it every March, but the logic seems pure. If the point is to give yourself the best chance to win a prize, lining up behind the top pre-tournament favorites is sheer folly. Concentrate your thoughts on the secondary group of strong teams: the weakest one seeds, the two seeds, and the strongest three seeds.
Let me explain: because most scoring systems provide disproportionate weight to the final two rounds, the bracket winner almost always has chosen the champion correctly. But virtually every pool I've helped organize has seen an overwhelming majority of entrants cluster around the top three or four favorites. In a 100-person pool this year, a third of the folks might choose North Carolina as their champion; that means that even if the Tar Heels win it all, their backers still have to be better than 32 other UNC supporters. Those brackets now have a much smaller margin of error in the early rounds, and can be mathematically eliminated fairly quickly.
A school like Texas or Kansas or Georgetown, on the other hand, is still one of the dozen or so pre-tournament heavyweights, but will gather scant national championship support, often less than their Las Vegas odds would suggest. Which means if you select such a secondary squad as your champion, you can make several major missteps in earlier rounds and still have a chance to finish first if your school becomes a surprise national champion.
Now, do I think Kevin Love, right, and UCLA has the best shot of winning it all next month? Yup, I sure do. But would I rather have a 50-50 share of Louisville's chances than a four percent share of UCLA's chances? No question.

All that being said, don't follow Xavier or Tennessee off a cliff and blame it on me. For eight years in a row, and for 11 of the last 12 seasons, the eventual national champion has made at least one recent Final Four appearance. ("Recent" meaning "within the past seven years.") There have been plenty of surprising Final Four teams in that span--think Marquette and Utah and U-Mass. and yes, George Mason--but those teams haven't broken all the way through for a national title.
What does that mean for this year? In addition to Xavier and Tennessee, be wary of selecting top-seeded Memphis as your national champs.

Once you've got your national champion, go back and make sure you've given that team's conference the proper respect in earlier rounds.
There's obviously no real link between, say, Duke's performance in the last round and Clemson's performance in the first, but I don't care: I always choose one major conference to ride throughout the tournament. The strategy makes theoretical sense-that there would be some kind of interior consistency in the final results-and it seems to offer major rewards.
Last year the SEC went 4-1 in the first round, produced three Sweet 16 teams (tied for the most) and the national champion. In 2006, the SEC went 5-1 in the first round, produced two Sweet 16 teams (tied for second) and the national champion. In 2005, the ACC went 5-0 in the first round, produced three Sweet 16 teams (tied for the most) and the national champion. The previous year, the Big East went 5-1 in the first round, produced three Sweet 16 teams (tied for the most) and the national champion.
In fact, over the past six seasons, the eventual national champion's conference has gone 27-3 in first-round games and produced an average of a shade under three Sweet 16 teams each year. So if you like, say, Louisville to win it all this year, attempting to discriminate between the Big East's fourth and fifth teams is futile; just take 'em all in the early going.

Steinberg's head
To recap, think favorites early, upsets in the middle rounds, and a mildly surprising national champion. Be biased in favor of power conferences, and against one-hit wonders. Do I claim that this system will help you win your pool this month? Absolutely not. But the point is, it will give you a CHANCE to win a pool; it will allow you to submit picks that could theoretically win. As long as the champion you pick doesn't lose in the first round.
Discussing Dan's Theories -- Q & A Transcript | Research Your Picks -- NCAA Tournament Database
© The Washington Post Company