Warren Buffet offers $1 billion to any AMERICAN w/ perfect March Madness bracket

[Robbob]

Originally Posted by Daradon

^^^ I assume you only get to enter once.

pffff..rules

[Rathji]

Still, get a pool of 20 or so people to submit the 20 most likely options. Wouldn't be a bad idea

[Bring_Back_Shantz]

Originally Posted by Rathji

Still, get a pool of 20 or so people to submit the 20 most likely options. Wouldn't be a bad idea

Ohhhh, 20 people. Now you're on to something.

If only we can find 20 people to submit the most likely brackets, we're gold.

1:9.2 Quintillion......not a chance
1:460 Quadrillion.....now you're talking baby. It's in the bag.

[You Need a Thneed]

Originally Posted by Russic

According to this article the odds are actually 1 in 9.2 quintillion, but adjusted for there being a reasonable guess at some of the match ups, it's probably closer to 1 in 128 billion.

He doesn't explain how he gets to that 1 in 128 billion number. Frankly, I think that's absurd. 2^37 is about 137 billion. So saying that it's only 1 in 128 billion is saying that you know 26 places on the bracket guaranteed - which is nearly the entire first round.

Giving the 1 seeds a 99% chance of winning and the 2 seeds a 95% chance of winning in the first round doesn't get you even remotely close to knowing 26 spots on the bracket.

I made a speadsheet, giving each game a percentage likelihood of getting the right answer. In round 1, I gave the 1 seeds a 99% chance of winning, 95% to the no 2s, 85% to the no 3s, 80% to the no 4s, 75% to the no 5s, 65% to the no 6s, 53% to the no 7s, and 50% to the no 8s. In the second round, i gave 75% to the no 1s, 65% to the no 2s, 55% to the no 3s, and 50% to the no 4s. In the third round, I gave the no 1s a 60% chance, and the no 2s a 50% chance. In the last three rounds, I made each game 50/50.

That gives about a 1 in 1 trillion chance of writing down a perfect bracket. (1 in 1,049,887,751,485 to be exact) If everyone in the US filled out one bracket every year, there would be one perfect bracket filled out on average about every 3500 years.

I suppose the 1 in 128 billion is only off about 1 order of magnitude, closer than I thought probably.

[worth]

[You Need a Thneed]

It's actually quite interesting using my spreadsheet, how much things can change by changing the percentages a little bit. I found a list of historical percentages between different seed numbers. If I use the historical seed percentages (and assuming that the higher seed wins every round), i come out with only a 1 in 53 billion chance. Changing some percentages up by 5-10 percent can make huge differences. That drop is even with me dropping the chance of winning for the 2, 3, 4, and 5 seeds. A bunch of the rest of the numbers that I used were off 5-10% from the historic averages.

I left the 1 vs 16 at 99%, since it will happen someday. If I change that 99% to the 100% historical average, it jumps down to 1 in 51 billion.

I used historical records from here.

[FireFly]

Based on the improbability, he should probably open the contest up to everyone everywhere.