How to Use AI to (Possibly) Beat the March Madness
For most people, predicting the results of the NCAA March Madness with any degree of accuracy is foolish. The elimination tournament, which features 64 teams, one of which must win six games in a row to become champions, is notoriously unpredictable: unknown Davids upset the old Goliaths almost every year.
Enter the cars. Given the popularity of predictive algorithms and data models used in elections in recent years, it’s no surprise that the March frenzy has swept into frenzy, with a wave of AI trying to predict the big dance. Knowing this, perhaps looking at the daunting parenthesis, begs the question: Should you consult an AI before filling in the parenthesis?
The answer, which depends on who you are and your understanding of the intricacies of student basketball, is quite definite.
Are AI better at predicting a tournament?
According to Matt Osborne, a mathematician and PhD researcher at Ohio State University, an AI-powered model can help better inform your group in three ways.
As he explains to Lifehacker, a prediction model can give you a good idea of which direction a tournament will take, but it definitely cannot predict a tournament directly. He explains that there are certain criteria that can improve your chances of accuracy if you use AI: overcoming your personal biases (which are widespread in sports), assessing historical results, and an overall likelihood score (which he compares to traditional Las Vegas). … rates).
He delves into the details in an email, writing:
You may have personal biases that will not be in the data. For example, you think Team A is good because you’ve heard of it before, but it turns out that they are not that good this year.
Using historical results can help you guess better than guessing. Team randomization is essentially like a quarter flip, but Team A’s chances of beating Team B are usually not 50-50. For example, 16 seed points knocked over only 1 seed in tournament history (UMBC vs Virginia in 2018), so you can it is safe to select the best dogs in each region.
[Algorithms] can give you an estimate of the likelihood that something will happen (similar to the odds in Vegas), which will help you judge for yourself how likely this frustrated choice you are pondering is.
Osborne says AI tools tend to be useful for laypersons and well-researched experts alike. He adds, however, that hardy fans “are probably better equipped (due to their current knowledge) to understand the input and output of the instrument and to check if what the machine has to offer really sounds realistic.”
How accurate are usually AI brackets?
Asking the AI to predict a college basketball tournament is a little more difficult than asking it to predict the presidential election. While the latter can be prone to violation – as it was in 2016 – models receive a constant stream of polls that are based on the more reliable act of voting by people. Basketball, especially in an environment like the March Madness, is more volatile.
Osborne says first hand, “Only the best braces are ever covered, and since the tournament is pretty random, the best braces are usually not created with a data analysis tool.” As an example, he cites the case of 12-year-old Sam Holtz, who filled the perfect parenthesis in ESPN’s March Madness 2015 challenge by beating more than 11 million other participants. Surprisingly, Holtz did not even watch basketball with any degree of regularity and was guided by simple guesses when choosing his winners. And then, somehow, he made history by filling in the perfect parenthesis that challenged the giant odds that one business professor estimated was “somewhere in one of a number of quintillions.”
No predictive model has ever done this, and it is unlikely that it will. But that doesn’t mean that an AI brace is useless if you choose to take part in a competition. Osborne recommends the Five Thirty Eight (which is free) and the Sportsline (which is optional) as reference points.
Even with a model at your disposal, you should feel great guided by instinct and prejudice. After all, we are talking about sports.