Files
bracket/readme.md
rconnorjohnstone 0ee953ae69 Added a readme
2021-03-17 13:48:28 -06:00

1.6 KiB

Bracket Generation Script

Generating the perfect bracket using science

(actually, just standing on other peoples' shoulders)

Before I get started, I want to specify that this is not a production-ready application. This is a script. That I wrote in about an hour of work. If you want to change anything, just change it in main.py. Maybe one day I'll clean it up.

Anyway, this tool is based on the Massey Composite Basketball Rankings site. It compiles a massive list of basketball rankings by various methods. rankings.csv is sourced from there. The principle is, just choose the ranking systems that you'd like included inside of main.py and it will filter out all the rest. I recommend using the Moog Analysis to find the most predictive systems in recent months.

The script will then average these rankings and use them to compare teams and build up a bracket. The only thing it does other than directly compare the rankings is that it attempts to account for upsets. Basically, I figure an upset is more likely to occur when a team's "actual ranking" (by the composite of the ranking systems) is higher than their "expected ranking" (generally the seed of the team that should be expected to be at that point in the tournament times four). In that case, they realize at some point in the game that they might actually have a shot and play harder. Idk, upsets are hard to predict, but I had to do something or these brackets would be very boring.

There aren't any requirements (other than python3) so running is as simple as:

python main.py