Hayes: Are we overanalyzing March Madness?


Bob Hayes, Columnist

For many American sports fans, including myself, March represents the most exciting time of the year. Scintillating basketball stars on the cusp of greatness face off against guys who appear no more athletic than you or I. For a couple weeks in March, schools hardly anyone can locate on a map dethrone perennial powers while millions of Americans scream at their television sets.

As fun as these three weekends in March and April are, it is important to be wary of drawing conclusions from the always-shocking results of March Madness.

Over the last few days, statistics have been thrown around Twitter and sports blogs saying only 1.3 percent of brackets had Connecticut in the Final Four and 3.3 percent had Kentucky — both teams, of course, made it. Of the more than 5.9 million people who filled out ESPN.com’s Tournament Challenge brackets, only two people picked all four teams correctly. The precise numbers will differ depending on where you look because the percentages have been derived from different pools, but the conclusion remains the same: Hardly anyone picked a correct Final Four.

What is important is that just because somebody picked a correct Final Four does not mean they are actually better at predicting basketball games than 99.9 percent of the world. Instead, they got pretty lucky.

This misconception stems from the common trap of sports fans’ drawing conclusions from high-variance outcomes of a small sample. In the case of March Madness, this sample size is a series of one-game results, each subject to an element of randomness.

A good way to express what I mean is Connecticut’s tournament run. The Huskies are a well-coached, talented team that has had a number of outstanding performances in the past few weeks. However, though everyone lauds their victory over Michigan State, does anyone remember what happened in the first round? This same team took a number of improbable plays to eke out an overtime victory over 10th-seeded Saint Joseph’s. Connecticut was a foul call or missed shot away from crashing out of the first round. The Huskies went on to upset No. 2 Villanova, then edged third-seeded Iowa State, who had lost key player Georges Niang in the first round – something nobody could have predicted in his or her bracket.

It is unfair for me to single out or discount Connecticut’s tournament run. Similar things could be said about each of the four remaining teams. It undoubtedly takes a strong team to win four straight games against top competition, but just because a particular event happens once does not mean that it is a likely outcome.

Mercer’s win over Duke does not at all mean they are the better team, though we can reasonably conclude that Duke relies too much on 3-pointers and lacks size, while Mercer was a well-coached team that could compete with great teams. As much as we love to talk about Mercer’s prevailing over the Blue Devils, in reality, if these teams played 100 games, Duke would probably win at least 80 of them.

Dayton, the darlings of this year’s tournament, were a potentially botched foul call and missed shot away from relatively forgettable losses in each of their first two games. Instead, head coach Archie Miller is one of the hottest coaching commodities, even though he had no control over Aaron Craft’s missed buzzer-beater in the opening round.

Finally, the same people bragging about correctly picking the Final Four are generally the same people talking about the failures of bracket prediction models, most notably FiveThirtyEight’s Nate Silver’s algorithm. However, these critics only prove their own ignorance of the system. Silver said his title favorite, Louisville, had just a 15 percent chance of winning the tournament. That means that 85 percent of the time, someone else is expected to win. Silver never said Louisville would win the whole tournament — far from it — he simply calculated that if the tournament were played an infinite number of times, Louisville would win the most times.

We all love March Madness because of the players’ passion and the games’ unpredictability, but we must understand that with this unpredictability comes the urge to overanalyze. When we look at bracket predictions, one bounce of the ball or one swallowed whistle throws off the entire tournament. If played 1,000 times, the tournament could easily have 1,000 different results. Thus, correct bracket picks really only mean somebody got it right just once out of the nearly infinite possible outcomes. Just because a result happens does not mean it is a likely and, consequently, an intelligent prediction.

Fortunately, though, the insanity of these three spring weeks is the beauty of the tournament — and what keeps fans coming back year after year.

Bob Hayes is a Weinberg freshman. He can be reached at [email protected]. If you would like to respond publicly to this column, send a Letter to the Editor to [email protected].