Update: Ultimately, data science prevailed. Kansas won the tournament by 3 points. Here’s the final bracket results:

Data lead to a 99 percentile showing ranking in the top 14 thousand brackets placed on CBSSports.com. In final analysis St. Pete was a force to be reckoned with. They started the tournament with a fantastic upset win and was on a tear for three rounds. Them beating Purdue effectively killed the left side of my bracket. I was playing for points on that side and that was it. UNC stunned Baylor in the second round effectively cutting off the bottom left side of my bracket. I rode Gonzaga as far as I could, but Arkansas halted them before they could play against #2 ranked Duke.

I had some stumbling blocks along the right side of the bracket, but my Final Four matchup of Kansas playing Villanova proved out. Furthermore, Kansas’ 16 point win insured I still had a chance to win the big game.

It was fun watching data play out game-by-game. In the end, humans; not machines, show up to play.

Data Science has served my bracket well, but human behavior always exceeds the margin for error.

It’s been a wild crazy weekend for college basketball fans. There’s a reason why they call it March Madness and we’ve seen it play out in full effect. I’ve made a lot of noise about “The Bracket Picked by Data Science” this year, and for the most part it’s working out, but not every scientific pick has worked out.

My bracket is not broken.

10 of my Sweet 16 picks are still sweet. 6 of my Elite 8 picks are still in. All of my Final Four picks are still in it. Hopefully, to win it.

Now, I did deviate from the math for a couple of picks. Upsets are staple of the Big Dance. My choices against the algorithm have been a mixed bag.

Now that the smoke has settled I’m going to stop and analyze what happened.

The Shocker

The Peacocks from St. Peter’s stunned everyone, including those who picked them, in their upset of Kentucky. 15 seeded teams have only beaten second seeds ten times in NCAA tournament history with a 9-135 record. I felt fine after witnessing this loss. The math has Kentucky losing to Purdue in the Sweet 16, but my points and rankings were compounded later by upsets.

Showing Up to Play

Picking down rank gets more involved and you have look at offensive power to make better picks. Sometimes things look good on paper, but you need the humans to show up and play. UConn lost to New Mexico State in a seven point game that was well played. Juwan Howard led his Michigan team to beat 6th ranked CSU in their corner of the bracket. Iowa State bested LSU by 5 in a close game. Miami beat favored USC by 2. By two points! All were well played games.

Upset Specials

My deviations were…mixed. I chose San Francisco to beat Murray State. It was a delightful game, but they couldn’t mete out a win. Not necessarily an upset, but I did pick #9 Memphis over Boise. I chose Creighton over San Diego, and Richmond to beat Iowa. My 8/9 picks were on, but uggh down rank not so much. In the end, I wound up scoring 52 upset points. UNC bested Baylor in the second round, something I wasn’t expected at all. I had Baylor getting beat by Purdue in the Elite 8.

There you have it.

Right now, I’m still ranked in the top 94% of the millions of brackets filed. We’ll see what happens this week.

You can cheer and cry along with me by checking out my bracket at NCAA.com.