Villanova over Georgetown in 1985, Bucknell over Kansas in 2005, Middle Tennessee State over Michigan State two years ago … NCAA Tournament history is littered with unexpected upsets. It’s how March Madness got its name. If you’re trying to win a bracket challenge, we suggest you embrace the madness and try to call several upsets this year.
A common strategy for filling out a March Madness bracket is to simply pick the top seed to win each game. That strategy will typically get you close, but won’t get you to the top of your pool. Upsets happen with enough frequency that you must predict some if you want to win.
While upsets are inevitable, how many should you pick, and where? The following article offers round-by-round historical data and other pointers to guide your upset selections. For more information to help your March Madness bracket, read our other article on the behavioral biases that can negatively affect your bracket selections.
When Calling Upsets, Seek a Goldilocks Scenario
Picking upsets is a necessity to win a bracket pool, but don’t go overboard. You want a Goldilocks scenario: not too many, but not too few. Since the NCAA tournament field expanded to 64 teams in 1985, an average of 28% of all games have ended in an upset.¹ You should actually predict fewer than that.
Last year, the Wall Street Journal ran a simulation of different pool sizes and predictions to determine the optimal number of upsets one should predict. In a small, 10-participant pool, the analysis found you should call an average of 8.7 upsets. In a pool of 25, the optimal number was 10.5 and in a pool of 50 it was 11.1.²
Round One: Don’t Discount Double Digit Seeds
In the first round (we’re excluding the play-in games and referring to the round of 64), upsets have occurred in a quarter of all games since 1985. A 16 seed has never toppled a one seed, but don’t discount the potential of other double digit seeds: Historically, 10-15 seeds have averaged a total of six upsets every tournament.³
Round Two: The Mighty Start Falling
By the second round, the real madness begins. One and two seeds typically survive, advancing to the Sweet 16 roughly 87% and 64% of the time, respectively. Three seeds, however, advance only 51% the time, while four seeds make the Sweet 16 only 47% of the time.⁴
Expecting a double-digit seed to advance to the Sweet 16 isn’t a safe bet, but it’s not outlandish either. Historically, 10 seeds advance past the second round 18% of the time, while 11 and 12 seeds advance 15% of the time.
Rounds Three and Four
In the third round, upsets have historically occurred in 28% of all games, but the rate of upsets in this round varies considerably each season. Since 1985, there have been two years in which no third-round upsets occurred, but there have been seven years in which upsets occurred in half the games. The fourth round appears more ripe for upsets, as 45% of lower seeded teams walk away with a surprising knockout.
As you make Final Four predictions, remember that you don’t have to be perfect. An analysis of all brackets filled out on NCAA.com in the last seven years found that less than 0.5% of all participants have correctly predicted all four teams.⁵ Last year, only 1.6% of all participants even went 3-4. Chances are, that person isn’t in your pool.
Correctly selecting two Final Four teams is a good goal and will likely provide a good chance of winning a pool. Once again though, you’ll need to consider some upsets: One seeds make it through their region only 41% of the time. That’s more than any other seed, but the NCAA.com analysis shows we tend to overpredict the one seed, selecting one seeds to win in this round 56% of the time.
Upsets occur frequently enough that you’ll have to go out on a limb and predict some to win your pool. Now the hard part: predicting them correctly. There’s no crystal ball foretelling this year’s Cinderella stories. The best you can do is to make sure your choices are rational. We often allow behavioral biases to influence our decisions, to the detriment of our brackets.
To learn about the behavioral biases that creep into our March Madness predictions, and how to prevent them, read our article on the subject before you make your picks. Good luck!