Pickin'

I am an amazing college basketball picker.  I can pick the outcome of games with remarkable consistency.  Sadly, not in a good way.

Through the first round of the NCAA tournament (not the fake first round they invented a few years ago) I did . . . poorly.  I ended that round in the 3.8th percentile.  Yup, 96.2% of people are better than me.  I'm not sure if ESPN counts brackets that are left blank.  I hope not.  One thing they do count is all those silly brackets where someone picks Manhattan and Coastal Carolina in the championship game.  Those utterly ridiculous brackets are, well, ridiculous.  And I'm barely beating them.  So out of people who actually tried, I'm doing even worse than 3.8%.

How you ask? Well, the 1 through 4 seeds almost never lose in the first round.  This year, those teams went 15/16 (only Duke lost).  My bracket had all 16 of the 1-4 seeds winning, which means I got 15 right.  The other 16 games in the first round are a bit harder to pick for everyone, and near impossible for me.

For all the attention that 5/12 upsets get, 5 seeds still win 66% of the time.  6 seeds win 68%, 7 seeds 60% and 8 seeds 46%.  Pick wise, these games represent something reasonably close to a coin flip, which is why those perfect brackets are so hard.  If all 16 games were coin flips, the odds of getting a given number of games correct would be:

0 or 16 games: 0.0015%
1 or 15 games: 0.024%
2 or 14 games: 0.183%
3 or 13 games: 0.854%
4 or 12 games: 2.777%
5 or 11 games: 6.665%
6 or 10 games: 12.219%
7 or 9 games: 17.456%
8 games: 19.638%

Unfortunately for all of you, I don't really feel like taking the time to make a graph of that, or doing the math for the specific tournament games this year using number from kenpom or something like that.  (This will be left as an exercise for the reader.)  But, it's clear that in guessing coin flips, you're generally going to come out close to 50%.  In our 16 flip example, you're going to get 6, 7, 8, 9, or 10 right 79% of the time.  How did I do?  I got 3 of the 16 games correct.  The odds of getting 3 or fewer games right? 1%  And that's how you end up in the 3.8th percentile.

While my picks were obviously terrible, the majority of the games were quite close:
I picked VCU over SF Austin.  They lost by 2.
I picked OSU over Dayton.  They lost by 1.
I picked UNM over Stanford.  They lost by 5.
I picked Cincinnati over Harvard.  They lost by 4.
I picked Providence over UNC.  They lost by 2.
I picked St. Joe's over UConn.  They lost in OT.
I picked Oklahoma over North Dakota.  They lost in OT.
I picked Kansas State over Kentucky.  They lost by 7.
I picked St. Louis over NC State.  They lost by 3.
I picked ASU over Texas.  They lost by 2.

All told, there were 16 first round games decided by less than 10 points, and I got 12 of them wrong.  Maybe I'm saving my luck up for one of these years when I can win big.