Q: What is the Graves Bracket Challenge?
A: The #GravesBracketChallenge is a contest that my statistics students enter each year for the March Madness Tournament. Each student submits two entries to the contest.The first is a simulation that they run using an algorithm that I developed, the second bracket they are allowed to create however they see fit. I ask the students to post both of their brackets onto social media for a couple reasons. The first reason is that I will not return their submitted brackets until after the tournament, so this allows the students to check their progress without their actual submission. Secondly, it provides a verification of the time and date of the submission in the event that we do simulate a perfect bracket or a perfect region.
Q: How does the algorithm work?
A: The algorithm uses a random integer generator and a random binomial generator to predict upsets within the tournament. We use the empirical data from previous years and the seeding of teams to predict if an upset occurs. For example, if a 1 seed versus an 8 seed, the probability of an upset (in other words, the probability that the 8 seed beats the 1 seed) is 19.5 percent. We use that probability and then number of times a one faces an eight, to simulate outcomes. Then we compare the simulations to the actual outcomes of the tournament.
Q: What sparked your interest in making this algorithm?
A: I was attempting to create a lab that used probability distributions to simulate actual events and since a number of students find March Madness interested, it seemed like a good fit.
Q: How can students win your Bracket Challenge and what is the prize?
A: My statistics students win the #GravesBracketChallenge by getting the highest score on their bracket, either the simulation or their personal bracket. The student that has the highest score wins a $100 cash prize.
Q: How many students have won the challenge in the past?
A: This is my third year doing the #GravesBracketChallenge, so we have had two students in the past that have won. We have never simulated a perfect bracket, nor have we simulated a perfect region. Two goals that we wish to accomplish in the future by improving the algorithm over time.
