Code Bracket. Because Science.
9I don't know a lot about the teams in the NCAA tournament this year. But I do know the statistical winning percentage over the past 29 years of each seed broken down by round:
So I took this data and wrote a March Madness simulation using Coders Brackets which allows you to quickly make a smart college basketball bracket in about two minutes using JavaScript.
You can find my code and the bracket it generated here: https://www.codersbracket.com/code_bracket/532857336f24af220016783f
It has Florida winning the tournament, but my favorite part is St. Louis going to the final four. Science.
Oh, and I also made a donation to STEM education because we need smart computer scientists to help fill all these job openings we have at Mediocre Laboratories.
- 11 comments, 3 replies
- Comment
If you're able to scientifically produce great results you may be inadvertently starting a new side business for yourself every march......or even a more full time gig in Vegas
Good thing people play the games and not computers. It's much more interesting that way. Most of mine are random picks, but I enjoy picking Kansas and other Big 12 teams, even if the odds are against them.
I look at fivethirtyeight's % chances and then pick just about every underdog that has at least a 30% chance of winning that game. (It'd obviously be better to always pick the team likely to win, but how boring is that?)
I'm also posting this to shame anyone who auto-picks 538's bracket without making any personal customization.
@dave, as a Nate Silver fan, 538 was my natural place to look, especially as I don't know squat about college basketball. But I enjoyed personalizing based on nothing in particular. Louisville beats Kansas in the final. Sorry @jkearney & @mediacre
Is there a mediocre bracket group? I'd pitch in for a prize for the "winner" who finishes in the middle of the pack.
@joshaw Great idea, I will work on getting a bracket set up for our community
@jkearney Better hurry! Deadline is soon
@jkearney Did you get'er done?
Earlier this week, I was wondering how I should fill out my bracket. Stat comparison? Just go by the seed number? Coin flip? Numerological divination?
That's when I realized: why fill out my own bracket when I can get other people to do it for me?
Using a module I hacked together to count Twitter search results, I searched "bracket win (teamname)" for every team, with additional qualifiers for some teams so that, for instance, "bracket win Oklahoma" wouldn't match "bracket win Oklahoma State." Then, for each matchup, I compared the number of resutls from the past four days. The team with more results was the winner, resulting in this bracket:
It's going better than I expected, honestly. My bracket of crowds isn't winning my pool, but I'm not losing, either. At the very least, it's doing better than Shawn's science bracket, which just goes to show that, if we work together, we can beat science. Thanks, James Surowiecki!
Picking from the heart has been proven to fail... at least this year. Next year, though, I know they can do it. Sad 'Hawks.
Only six teams left. Hopefully @harrison is doing better with Twitter crowdsourced data.
So this is interesting... I get 4 days to gloat about being in 1st place. But I have almost the lowest possible final point total, so my chances of even taking 3rd are not good. I guess I need to keep cheering for those annoying double-digit seeds still in the tournament.
I have eight left. My final 2 is wrong, though (curse you, Kentucky), so things aren't looking great here.
@harrison, Looking at your picks I was thinking it ended up just based on seed. It mostly did, but there are some exceptions. I'm mostly surprised Wichita had more results than Louisville.