Brackets are Busted by the Three
Date: 4/8/2019
Tools: Python & Microsoft Power BI
As the 2019 NCAA basketball tournament is coming to a close, I decided to dive into the analytics of the first round bracket busters. Using the BeautifulSoup python app, I screen scraped data and used Microsoft Power BI to help visualize my findings. The criteria for a "bracket buster" in the first round was limited to 11-16 seeds winning over their corresponding 1-6 seeds since 2000. This criteria returned 97 games that fit the parameters.
​
Originally, I expected the underdog teams to have better shooting stats across the board. To my surprise, in only 40% of the games did the underdog team have a higher 2-point shooting percentage. On the other hand, having a higher 3-point shooting percentage is much more indicative of overcoming a team of much higher skill. In over 60% of the games, the lower seeded team shot better from behind the arc.
The other factor at play is shot volume. An underdog team could have lower three point and two point shooting percentage and still win if they had a higher shot volume. While the underdog team having a higher shot volume definitely played a role in some of these games, it was not as prevalent of an indicator as 3-point shooting percentage (53% of games had the underdog team with a higher number of shot attempts).
