If it was truly that low then they'd actually be at a significant risk. As it stands... the odds are much much worse. As stated earlier... if the games are all 50/50 then it's like 2^67. It's like a billion times a billion.
By my calculations... if every game was 60/40 and you had a billion people enter every year it would take more than 700,000 years to get a perfect bracket on average.
My math is probably wrong but... (1/(1-(1-.6^67)^1*(10^9))) in the right ballpark I think.
Presuming all four play-in games are included, there are 67 games in the tournament. So there are 2^67 possible brackets. The most obvious way to see this is in each initial game, call one team "Heads" and the other team "Tails", and keep those names constant throughout the tournament. Then you just have 67 coin flips. 2^67 is 1.4E20, or 147.5 billion billion.
Say an extraordinarily good picker can predict a game with 70% accuracy. It is reasonable to assume the average contestant can get over half the games right, mainly because a good chunk of games are in the first few rounds where there is a wide difference in seeds, and where the higher seed wins a great deal of the time. But overall, a 70% picker can can do very well at Vegas by just betting money lines until he gets cut off. I've never heard of anyone getting cut off from sports betting without trying to fix a game or otherwise cheat the casino, which should tell you how likely this is.
Anyway, the chance a 70% right contestant gets the bracket perfect is 1 in 24 billion (23.902 and change billion, actually). So the expected value of participating in this pool would be slightly over four cents.
We can calculate the level of accuracy required for the expected value of an entry to be one cent. It's just the 67th root of 100 billion (1E11). This comes out to .68521, or 68.5%, just over two-thirds. You'd have to get 46 out of 64 games right on average. One way to do this is to get the ENTIRE play-in round, the ENTIRE Round of 64, and 10 out of the Sweet 16, and everything wrong the rest of way. That's not just getting lucky, that would have to be your AVERAGE bracket.
What Buffet really should do is allow anyone to enter an unlimited number of brackets, but charge one cent for each entry.
Think of it this way. There are 68 teams at the start of the tournament and 1 team at the end of the tournament. Each game eliminates exactly 1 team. So, 67 games.
I think it's possible that someone picks a perfect bracket in our lifetime, but it'd probably have to be during a year like 2007 where the tournament is super boring with few upsets.
Oh whatever, considering I'd have to dance around all those taxes, leverage the remainder, pay estimateds, blarrgh, what a pain.
Except that that's not really the case. I mean, it is in your coin-flip scenario, but in reality, Syracuse has a much better than 50/50 chance to win their first few games. If you were truly "guessing" rather than "picking," you would have that blind-squirrel chance. However, I like to think that as a bit of a fan, I can guess whether Syracuse beats Directional State University more than half the time.
It's a damned complex equation.
Darn it, where's Paul when you need him?