Warren Buffett And His NCAA Bet by David Merkel, CFA of Aleph Blog

I have no idea what premium Warren Buffett is receiving for insuring against one or more people having a perfect NCAA Tournament bracket, but it is unlikely that he will lose on this underwriting bet.  Those seeking insurance on unlikely events think the events are more likely than they actually are.  That said, for Quicken Loans, they don’t want to bet the company, and they do want publicity, so contracting with Warren Buffett is worth their while.

Warren Buffett Makes Big NCAA Bet: An Actuarial Look At Odds

Imagine for a moment that the average person submitting a bracket had a 78.6% chance of getting each game right, and the maximum 10 million people sent in their brackets.  What is the likely number of correct brackets?  One.

But does the average person get more than three out of four games right?  I don’t think so.  Are there some people that are better than others so that they get games right 90% of the time?  Well, if they are 1,163 out of the ten million, on average, one of them will have a perfect bracket.

Here’s a further problem.  Every tournament has significant upsets.  Someone who has a good understanding of how good the teams are will know how to pick the most likely team to win.  It is tough to pick the upsets, and tougher to pick all of the upsets.  There is no good model for upsets, or they wouldn’t be upsets.


As an aside, the prize is $500 million as a lump or $25 million for 40 years.  The breakeven yield rate on that is 4.21%.  Warren Buffett knows he can beat 4.21%.

This contest is like the lottery.  If I had one piece of advice for lottery winners it would be to take the payments over time.  The discount rates on most lotteries are far higher than 4.21%, and really, taking them over time gives you a chance to learn how to manage more money then you know what to do with.  Taking the payments over time gives you the freedom to learn from mistakes.  We all make mistakes, but when we get all the money at once, we make more.