The Old New Thing

Well, Raymond, as of today you are 17 for 17.  You even got the opening game right.

You got skin in the game or just pontificating ? Surely your coworkers setup a $5 bracket or somesuch.

I'd say Raymond probably doesn't have any "skin" in the game.  If you check out this post < http://blogs.msdn.com/oldnewthing/archive/2006/03/16/552822.aspx > you'll see he doesn't know squat about basketball and just dos this for fun.  The first year the teams were ranked based on whose president served the longest, and the next year it was based on the pay of the head coach.  

I gave a copy of Raymond's brackets to an NCAA freak here at work and his jaw dropped when he say Raymond had predicted the UMBC upset of Georgetown.  I dunno, Raymond -- you might have found a good method here!

WHOO! Go Retrievers!

I graduated from UMBC last year, and I can't remember any sort of basketball team on campus. I have no idea how they managed to make it to the NCAA tournament, much less beat the #2 team.

Shouldn't they play the game before a winner is declared?  It's starting on 3/21 at 3:10...

OK, I'll bite: why are so few matches played between teams with an even seed difference?

You ought to read Isaac Asimov's 'The Machine That Won The War' for the most effective statistical algorithm ever created to make the right choices in situations like this.

>OK, I'll bite: why are so few matches played between teams with an even seed difference?

All the first round games are played with an odd seed difference (1 plays 16, etc).    More than half the games in any bracket are first round games, so assuming a 50-50 distribution of even and odd seed differences after the first round means that 75% of the games are odd.

Raymond, K St. over USC, impressive.

> OK, I'll bite: why are so few matches played between teams with an even seed difference?

Because the way tournament is structured, the best team plays the worst team. If all the favorites win (they often do), you will always have an odd number for the seed difference (only an upset produce an even seed difference next round). Only exception is when 8th seed plays the 9th seed in the first round, it's a virtual coin toss. Therefore, it's a 50-50 for a first seed to meet a 8th or 9th seed in the second round of regional action. (look at row 7 and 8, nearly identical).

So far you've only missed two... looks like you have a pretty good method.

Go Michigan St.!

I like the looks of your bracket, Michigan State out front like they should be!

Go Green!

I have nothing to add other than the fact that you have two South brackets and no West bracket.  I nitpick at thee!

PingBack from http://codesmithy.wordpress.com/2008/03/22/march-madness-craziness-part-one/

You did extremely well with your first round picks. Nice job.  I'm trying to understand your methodology. You wrote that you "went for the suboptimal algorithm of choosing q = p."  Can you explain that more? I don't understand but am fascinated by what you came up with.

@patrick -- it means that instead of always picking the favorite, he'll pick the favorite most of the time: for seed difference 1 (upset rate 46%), he's 46% likely to choose the upset (probably by RNG).

@ScottB & Raymond.  I don't understand... Take, for example, the choices when #4 vs. #13.  Twice the #4 seed is picked to win, twice the #13 seed is picked, but the upset rate for this differential is not 50%.  And the two times the #13 was picked (San Diego over UCONN, Siena over Vanderbilt), Raymond was correct. Why were these particular #13's chosen and not the other two #13s?  Just trying to figure out whether this was random chance.