Yeah, but couldn’t there be some fluke postseason where an unusual number of higher seeds win? Everybody’s brackets are balls this year because who would have predicted 8 10+ seeds winning the first round, but that’s not always going to be the case.
Maybe we have a season where there’s definitely Haves and Have-Nots. Very few first round upsets, and the Final Four ends up with a 1, 1, 1, & a 2. Out of millions of brackets submitted nobody randomly got there?
1st round stats, since tournament expanded to 64 in '85:
1 seeds win 100%
2 seeds win 96%
3 seeds win 84%
4 seeds win 79%
5 seeds win 68%
6 seeds win 69%
7 seeds win 63%
8 seeds win 42%
If you pick the favorites, in every game, (which statistically give you the best chance to be perfect), based on these numbers, if my math is right, you have around a 1 in 25,000 chance for a perfect first round.
So, if you get that far, there’s still 32 games to go, likely half of them coin flips. calling 16 straight coinflips is about a 1 in 50,000 shot. combined with your 1 in 25,000 perfect first round, you’re already at over 1 in a billion, and you still got 16 more game you need to call. Easier games to guess, but even if you had an 80% chance of calling each of the other 16 games, your overall chance of a perfect bracket would still grow to around 1 in 50 billion.
My dad was bizarrely good at picking these things - he won the office pool every year except one.
But even then he realized he had a fluke bracket. Had the thing framed (it was done on a newspaper sheet), but it was lost after his death a few years ago. 
Harvard (a #16 seed) beat Stanford (a #1 Seed) in the 1998 NCAA Basketball tournament. Check that, Make it the 1998 NCAA Womens Basketball Tournament.
Come, now, silly. Wimmin don’t play basketball.
How much do you think “correct” ranking would factor in. According to the stats provided, #9 seeds beat #8s most of the time. If prognostication on the part of the ranking committee were better, how much would that matter?
Article about autistic teenager from Chicago who has a perfect bracket through round 2
Crap! I did that same analysis years ago, and came to the same conclusion: from the regional semifinals on, seedings are irrelevant. It kind of made sense, in retrospect–a team that gets that far in the tourney can’t do it on luck, they’ve really proved something. Eg. UNI maybe got lucky & caught Kansas on an off night and played the game of their lives, in which case you can expect MSU to crush them like a grape. Or, they’re really that good, in which case the Spartans are in trouble*. And if that happens, and the Panthers advance to the Elite Eight, you can’t really think of them as an underdog in any of the subsequent games.
Who knew I could have gotten interviewed by a magazine for that (fairly simple) bit of math?
*Not happening, of course. Go Spartans! 
Nice story, and as much as I’d like to buy it…I just can’t.
I’ve been filling out brackets for a long, long time, and I’ve never seen anyone do this. Ever. And with the upsets this year?
Rainman angle or not, I just can’t buy it.
I can buy that he got all the picks right so far. What I’m not buying is him having Purdue to win it all!
I heard his father on the radio this morning. He said it was a combination of being a fan of basketball and luck.
Couldn’t someone write a computer program to figure out every possible bracket and save them to disks, and then after the tournament claim that one of them must be perfect?
I have not done the math on how much processing time or disk space this would require. Probably a lot.
Am I missing the point of that article? It doesn’t make sense to me. The NCAA tournament has 4 1-seeds, 4 2-seeds. At elite eight all teams either are a 1-seed, 2-seed or beat a high seed to get there (or beat someone who beat someone).
Essentially, isn’t he saying that all the 1-seeds and 2-seeds are about equal?
How is this shocking?
According to an ESPN story, there are 9 quintillion possibilities, and filling out one bracket a second would take 3 billion years.
A computer could do it faster than one per second, though. Once you decide on a format, the whole bracket can be represented by 63 bits. And it’s independent of the teams, so we could start work on our 3,000,002,011 bracket today.
Knock yourself out. 
I have no idea how fast a computer could do it, or if in fact the information I gave is even correct. It was just something I saw on one of those ESPN March Madness specials.
No, they wouldn’t. Scores in general would be higher because the penalty for missing in the earlier rounds would be less, but the rate of perfect scores would remain the same. It doesn’t matter whether I chose Kansas to win it all in a bracket format or merely to win that individual game, either way in terms of being “perfect”, I have lost.
It won’t hold. He has Duke losing to Purdue! 
Doesn’t matter, you’d still need about a billion terabyte drives to store them all (I did that in my head, so I’m not 100% sure on the order of magnitudes, but my basic point is correct).
But, see, the trick is that you store them in a compressed format. Then, after the tournament is over, you just uncompress the appropriate part of the file, and show it off.