Two weeks ago, I explained how most of the numbers floating around regarding the “perfect bracket” were missing several key aspects to prediction methods.
Most notably, humans don’t pick games by tossing a coin (edge, humans), but they are also much more inclined to pick favorites than underdogs, underestimating the chances of upsets (edge, computers, at least as far as trying to pick a perfect bracket).
The question I attempt to answer is the one I which I believe most people are truly interested in:
What are the odds that a random individual fills out a bracket with every game picked correctly?
To answer this, I extracted three sets of game probabilities:
(1) ESPN consensus numbers
(2) Ken Pomeroy predictions
(3) Nate Silver predictions on 538.
For 538’s numbers, I used a spreadsheet sent by a loyal reader (Dan Z), who extracted pre-tournament predictions (538 updated its numbers as each round went on).
Next, I took the probabilities that each prediction method method gave the winners who advanced to each round. For example, ESPN consensus gave UConn about a 1 in 300 chance of winning the title, while Pomeroy and Silver’s numbers gave the Huskies a roughly 1 in 130 chance of winning it all.
Here’s a screenshot of the probabilities for each team, round, and prediction site.
Here are the odds of a perfect bracket given the numbers above:
(1) ESPN consensus numbers: 1 in 107 quadrillion
(2) Ken Pomeroy predictions: 1 in 3.3 quadrillion
(3) Nate Silver predictions on 538: 1 in 1.1 quadrillion.
For reference, here’s what a quadrillion is: 1,000,000,000,000,000 (yes, I had to look this up).
Given the unlikely set of finalists in 2014, these odds are likely more extreme than usual. Still, I liked how using statistics to inform picks would have greatly increased the odds of a perfect bracket (by a factor of 30 or 100, depending on whether or not you used KenPom or 538’s numbers).
As it turns out, these numbers ended up being quite similar to an article in the WSJ a few days ago.
What else has a 1 in 107 quadrillion chance of happening?
Well, given that one has about a 1 in 175 million chance of winning Powerball, your odds of winning the Powerball on two consecutive tries, while buying 1 ticket at each attempt (1 in 31 quadrillion) are better than your chances of picking a perfect 2014 March Madness bracket using the ESPN consensus. So, well, good luck with that.
Thanks for reading, and I’m looking forward to 2015 already.