Skip to content

willhammondhimself/poker-tracker

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AI Poker Coach

Session tracker with AI analysis I built for my Ignition grind. Wanted something that does bankroll math properly and finds leaks automatically instead of a spreadsheet.

What it does

  • Session logging - Track live sessions or log completed ones
  • Hand logger - Record hands with a card picker (keyboard shortcuts work)
  • LeakFinder - Finds your -EV spots and ranks them
  • My Edge - Shows your top exploits and overall BB/100

The quant stuff

Monte Carlo bankroll sim

Runs 1000 trajectories to estimate risk of ruin. Has a Kelly calculator too. The fan chart looks cool but honestly you probably already know if you're rolled for the stakes.

Tilt detection

Looks for downswings (>10bb in 50 hands) and whether you start spewinig after. Gives you a 0-10 score. Mostly useful for post-session review.

Opponent tagging

Auto-tags villains based on VPIP/PFR/AF. Categories are the usual - whale, nit, LAG, etc. Suggests exploits but they're pretty obvious (don't bluff the calling station, etc).

Hand replayer

Step through hands street by street. Green felt table, looks nice.

GTO Radar

Scatterpolar comparing your stats to a baseline. Shows where you're deviating.

PDF tearsheet

Generates a report you could theoretically show a backer. Has your hourly, session winrate, trailing 15 sessions.

Luck bucket

EV calculator for all-in spots. Tracks how much you've run above/below EV.

Quant Lab (the fun stuff)

Volatility (GARCH)

GARCH(1,1) on session PnL. Flags if variance is abnormally high/low compared to your history. Useful for knowing when to take shots or move down.

Opponent clustering

PCA + K-means on villain stats. Groups regs into archetypes (nit, lag, etc). Needs ~50 opponents with decent sample to be useful.

Winrate confidence intervals

Bootstrap resampling (10k iterations) to get 95% CI on your actual winrate. Shows P(you're a winner) which is humbling with small samples.

Caveats

  • Need 50+ opponents for clustering to mean anything
  • GARCH needs 10+ sessions minimum
  • Winrate CI is wide until you have 10k+ hands
  • Only parses Ignition Zone format right now

Setup

git clone https://github.com/yourusername/poker-tracker.git
cd poker-tracker
pip install -r requirements.txt
streamlit run app.py

AI Coach (optional)

Add your Perplexity API key to .streamlit/secrets.toml:

[perplexity]
api_key = "pplx-your-key"

Structure

poker-tracker/
├── app.py              # main streamlit app
├── analytics/          # GARCH, clustering, bayesian stuff
├── components/         # UI components
├── utils/              # parsers, calculators, data loading
└── data/               # JSON storage

Built to track sessions and practice scipy/sklearn. The quant stuff is mostly for fun - you don't need GARCH to know you're running bad.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages