Skip to content

raymondhfeng/project_yuler

Repository files navigation

Project Yuler

An Django web app that remembers online poker data. Uses said data to helper players make better decisions on when to play.

Live site

http://e3726eaab85e.ngrok.io/ignition/chart (Main)
http://2955a2050a6d.ngrok.io/ignition/chart (Backup)

System Diagram

alt text

How It Works

  • Terminal script on Windows captures screenshot of the poker GUI information.
  • Image processing to "clean" the screenshot into an image that resembles a spreadsheet.
  • PyTesseract OCR to parse the image into integers and floats.
  • Django Celery Task that inputs the values into a database at a minute resolution.
  • Chart.js to display the most recent 2 hours of data at each stake.
  • ngrok forwards my locally running port to the outside world.

Data Analysis

  • Linear Regression: Regressing Number of players at NL25 on number of players at other stakes, time of day, and day of week.
  • CVXPY Regression: Regressing on the same variables, but using convex solvers to include an inequality constraint, which ensures that predictions for 50+ players will be greater than fifty.
  • Tobit Regression: In progress.

Naming

A pun that combines the famous Project Euler, and Udny Yule, the Yule in the Yule-Walker equations from time series analysis.

About

Infrastructure that screenscrapes dynamic data and stores in database for time series analysis.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published