The STEM field can yield advantages in calculating March Madness brackets and engage with math and tech in a fun and unique way.
This year’s NCAA Men’s Basketball Tournament may be over, but the concepts of how to use statistical and mathematical analysis to predict better brackets remain.
For those who are already thinking about how to win next year’s March Madness office pool, Kevin Brennan, AP and graduate statistics educational expert on the Varsity Tutors platform, offers his insights.
STEM and students
Scott Matteson: How does STEM (Science, Technology, Engineering, and Math) tie in with March Madness?
Kevin Brennan: The key to creating a successful March Madness bracket is half basketball knowledge and half game theory. March Madness incorporates both technology and math to generate predictions, outcomes, and team rankings. The STEM approach correlates to this.
Scott Matteson: What can students learn from the correlation?
Kevin Brennan: Students can learn about percentages and probability by formulating a March Madness bracket. They can use mathematical equations to determine each team’s odds of winning and develop a more statistical approach to decision-making.
It’s best to pick teams that are statistically more likely to win, regardless of what other people think, and it’s important to pay attention to specific match-ups likely to produce an upset. This gives students a real-life example of conditional probability, where the likelihood of an outcome is affected by the conditions surrounding it.
Scott Matteson: Can you provide some concrete/subjective examples of how March Madness incorporates both technology and math to generate predictions, outcomes, and team rankings?
Kevin Brennan: Personally, I use a series of math equations to formulate the most informed decisions when making bracket predictions. For example, a common match-up in the bracket difficult to predict is the #10 seed vs. the #7 seed. Typically, this match-up occurs in the first round of every regional, so it’s important to use a strategy. Historically, the #7 seed wins this match-up 61.3% of the time.
So, you should pick the higher seed, right? Not so fast. This is where we should “metagame,” or consider how likely our peers are to pick the #7 seed. Let’s assume that each player gets one point for the correct guess and zero points for a wrong one, and you’re playing against one person who will pick the #7 seed 80% of the time. Figure A shows the decision tree related to that choice.
The math reveals picking the #7 seed in this situation gains us an average of 0.0452 points (not great) while picking the #10 costs an average of 0.1808 points (noticeably worse). For this situation, it’s best to pick #7.
The Google March Madness connection
Scott Matteson: How does technology factor into the concept?
Kevin Brennan: Over the past decade, technology has significantly modernized basketball and helped with sports predictions. Many aspects of our lives are becoming more data-driven, and sports analytics is a huge part of this. Basketball has been progressive with technology, including individual player and ball-tracking technologies, incorporating historical metrics of shooting efficiency and passing accuracy for viewers to have the most information to make logical predictions, especially with compiling March Madness brackets. New metrics like player efficiency ratings help players, coaches, and fans better understand the game.
Most recently, Google recruited college students to deliver March Madness analysis (SEE: Students create NCAA March Madness predictive analysis via Google Cloud) in order to show how to use analytics and machine learning tools for bracket selections.
In terms of technology, there are many data analytics and tools available to help inform decisions. Google Cloud is a great example, which uses data analytics to inform viewers of the game and build predictions for what might happen. Adobe is another great choice; its predictions finished in the 98th percentile last year.
Scott Matteson: You mentioned that Google recruited college students to deliver March Madness analysis. Can you share any details about how it worked out?
Kevin Brennan: While predicting the perfect bracket is nearly impossible, people are eager to use data to improve their chances of making the correct picks. Recruiting undergraduates for this initiative is a great way to promote STEM for students.
Last year, the Google Cloud had accurate analytics and real-time predictions, and this year it’s looking to be the same. It will only improve year after year, with more data to analyze and inform decisions. By combining storage of the raw files with BigQuery to host data tables and model outputs, the platform represents only the beginning of what the NCAA can do with its data.
Preparing for next year’s March Madness
Scott Matteson: What do you recommend for next year’s March Madness for those interested in STEM?
Kevin Brennan: I think March Madness allows viewers to engage with math and technology in a fun and unique way. STEM subjects can seem boring to some, but when you pair it with another personal interest like basketball, it makes technical tools or mathematical equations interesting and relatable, while also beneficial in helping you win your bracket.
For those who want a leg up on the competition, there is a wide range of advanced statistics available that go beyond traditional metrics like the number of rebounds or points per game. These stats can help you determine, which team is more likely to score on a given possession or how important a single player is to the outcome of a game.
March Madness is a large, single-elimination tournament. Therefore, there will always be a lot of upsets and busted brackets. The most important thing is to have fun. That, and to improve the odds of beating your colleagues in the office betting pool.