Scientific Strategies for Choosing March Madness Winners
Probability matching is tempting, but strategists are better off going by the numbers when choosing winners in the NCAA men’s basketball tournament
BLOOMINGTON, Ind. — Soon Americans nationwide will begin poring over NCAA men’s basketball tournament brackets in their annual attempt at glory — and maybe even a little cash — in winning the ubiquitous, albeit illegal, office pool.
Some will go by the numbers, picking the team in each matchup with the best ranking or seed. Others use intuition, sports knowledge, favorite colors, mascot preferences — it’s not called March Madness for nothing — or other somewhat unscientific methods for picking winners and more importantly, picking the upsets.
New research from Indiana University and the University of Wyoming has found that strategists, regardless of their sports expertise, would be better off sticking with the numbers — but what’s the fun in that? Bettors often think picking the upsets will give them an edge, and that they know how to pick them.
“Picking the lower seed is a good strategy, but people think, ‘I can’t win by doing that because everyone else is doing this,’” said Ed Hirt, professor in IU Bloomington’s Department of Psychological and Brain Sciences. “The upsets people pick are no better than chance. People have this idea that they know how many upsets will occur, but can they predict the ones that will occur? They pick upsets but not the right ones and end up sabotaging their efforts.”
Hirt’s study, co-authored by Sean M. McCrea, University of Wyoming, was published in the “Journal of Applied Social Psychology.” McCrea said they were surprised by how little expertise or favoring an underdog really explained people’s tournament predictions.
“Instead, it seems that people who follow basketball are aware of the possibility of upsets and fool themselves into believing that they can figure out which upsets will happen,” he said. “The problem is that the tournament seedings summarize most of the useful information one could use (win-loss record, strength of schedule, etc. ) and so the upsets are much less predictable than one might think.”
Other studies have shown that making NCAA bracket predictions based on rankings from other experts, such as sportswriter polls or gambling bookies, are no more successful than choosing the lower seeds. Hirt and McCrea sought to examine whether bettors used probability matching to pick upsets, if this approach was more successful than picking winning teams based on seeding, and whether people use probability matching because they viewed basketball as a skilled, non-random activity that could be predicted — essentially, thinking they just know better.
Probability matching describes a scenario where individuals predict a specific outcome based on an existing rate of occurrence. For example, in the first round of the NCAA tournament, prognosticators often expect an upset in the contests between No. 5 and No. 12 seeds, so bettors often attempt to pick which of the four games involving a 5-12 matchup will see the upset.
Hirt and McCrea examined bracket strategies as a way to study this common decision-making behavior, which frequently is seen when individuals make predictions or judgments in areas involving skill, such as hiring decisions, outcomes of races or predicting stock prices. Hirt says this behavior relies on a confidence that an individual’s insight can trump variability or discern patterns in randomness.
For the study, they examined NCAA tournament results from 1985-2005 and the first-round predictions of more than 3 million entries in an ESPN Tournament Challenge. They also designed a series of studies involving varying degrees of perceived randomness.
Their study provides one of the first demonstrations that probability matching is used more frequently for predictions of social behavior than for predictions of random events.
“We want to deny the fact that there’s variability, that there are bad days,” Hirt said. “We want to think we can predict these things. It’s human nature to think that things aren’t random, serendipitous, that we should be able to predict what someone will do or outcomes of situations that we care about.”
On the Net: