Big Data Probabilities Of March Madness by Attain Capital
For the next 4 days, more than 70 million brackets (40 millions people) will be filled out by Americans, entranced with the fallacy of picking a perfect bracket for the Men’s 2015 NCAA Basketball Tournament. Between now and Thursday, millions will evaluate win-loss records, RPI rank, strength of schedule, points per game, and free throw percentages. But that’s all surface level.
You also have to consider who has the easiest road to the final four, which seeds historically have a higher probably of defeating the other. March Madness isn’t just for college basketball fans anymore, it’s turned into who has the best bracket amongst their friends (regardless of they’ve watched any games before the games begin).
Last year, the Oracle of Omaha, Warren Buffett, offered up $1 Billion to anyone who successfully picked every match up correctly. For another year, no one was able to predict each game correctly. While Buffett isn’t offering up $1 Billion this year, The American Gambling Association estimates that $9 Billion will be wagered when it’s all said and done (That’s the same GDP size as The Bahamas). Enough for us to ask again, the probability of picking a perfect bracket.
We wrote a post last year about the actual odds of accurately choosing every game, and the numbers varied. If you flipped a coin, for each probability, the odds were 9.2 quintillion. Here’s the brief description again via, Business Insider:
“If all of these brackets are equally likely — if each game in the entire tournament is a 50-50 tossup, and picking the winner is basically a coin flip — we then get the odds of a correct bracket at one in 9.2 quintillion.
Of course, flipping a coin 63 times is probably not a very good strategy for deciding how to fill out your bracket. Most of the games are not 50-50 matchups.
Consider the first round (the round of 64) of the NCAA Tournament. Of the 32 games in the first round, there are four games in which four of the best 64 teams (1st seeds) play four of the worst 64 teams (16th seeds).
Since 1985, when the tournament first expanded to 64 teams, no 16th seed has ever beaten a 1st seed in the round of 64.
If we’re comfortable assuming that this trend continues, we can safely fill in the four 1st seed vs. 16th seed games on our brackets.
Now we have 59 games to pick, and if we flip coins for all those, we have a one in 259, or about one in 576 quadrillion, chance of winning the tournament. Still pretty terrible odds, but by making this one assumption, we have boosted our chances by a factor of 16.”
Will there be someone out there this year to defy the odds? Is big data getting closer to cracking the code of predicting winners? We wish everyone the best of luck. In case you want any assistance, here is FiveThirtyEight’s interactive bracket predictions, and Bing’s bracket predictions.
Photo Courtesy: NCAA.com