College basketball enthusiasts should reconsider relying solely on artificial intelligence to create a flawless March Madness bracket.
While the advancement of artificial intelligence into everyday life has made “AI” one of the buzziest phrases of the past year, its application in bracketology circles is not so new. Even so, the annual bracket contests still provide plenty of surprises for computer science aficionados who’ve spent years honing their models with past NCAA Tournament results.
The technologically inclined are chasing goals even more complicated than selecting the winners of all 67 matchups in both the men’s and women’s NCAA tournaments.
According to data analyst Chris Ford, March Madness predictions require a blend of art, science, and an understanding of human psychology. Even casual fans often base their bracket decisions on factors like team momentum or standout player performances.
Video link : https://youtu.be/CzgDCuyrBTM?si=5MzLbVVL53Y6i9Sz
Meanwhile, technologically savvy individuals are striving to develop sophisticated mathematical models to predict tournament success objectively. Some are leveraging AI to enhance their algorithms or determine which team attributes should carry the most weight in predictions.
However, the odds of crafting a perfect bracket remain slim for any competitor, regardless of their tools. Even with informed assumptions based on historical data, the chances of achieving perfection are extremely low, as highlighted by mathematics professor Ezra Miller from Duke University. According to Miller, predicting the outcome of an evenly matched game in the NCAA Tournament is akin to selecting a random person from the Western Hemisphere.
While artificial intelligence excels at assessing the likelihood of a team’s victory, Miller notes that even with sophisticated models, choosing the winner of a closely contested game remains essentially a random decision.
For a decade now, Kaggle has been hosting “Machine Learning Madness,” a unique competition where participants must submit their confidence levels as percentages for each team’s advancement, rather than filling out traditional all-or-nothing brackets.
Utilizing a comprehensive dataset of past tournament results provided by Kaggle, competitors develop algorithms based on various team statistics such as free-throw percentage, turnovers, and assists. These algorithms aim to identify the most predictive factors of tournament success.
Jeff Sonas, a co-founder of the competition and a statistical chess analyst, emphasizes that it’s a level playing field, allowing both basketball enthusiasts and data-savvy individuals to compete based on their respective strengths.
Meanwhile, Ford, a Purdue fan, has taken a different approach since 2020 by attempting to forecast the 68-team field for the NCAA Tournament. In his most successful year in 2021, Ford’s model accurately predicted 66 of the teams. He employs a “fake committee” comprising eight distinct machine learning models, each considering various factors such as strength of schedule and quality wins against formidable opponents.