Juan De Dios Santos - Getting better at Pokémon using data, Python, and ChatGPT | PyData Global 2023

Used data analysis and Python to improve Pokemon Trading Card Game performance by tracking 100+ matches and analyzing win ratios, deck performance, and game patterns

Implemented hypergeometric distribution calculations to determine probability of drawing specific cards, helping optimize deck building and initial hand strategies

Created automated dashboard using Google Sheets and Apps Script to maintain running statistics including win ratios, remaining prizes, and match outcomes

Leveraged ChatGPT to analyze qualitative game notes and categorize insights into themes like mistakes, learning moments, and gameplay patterns

Key metrics tracked included:

• Win/loss ratio over time
• Prizes remaining at game end
• Opponent deck matchups
• First/second turn performance
• Game-ending conditions (concessions vs. completions)

Built data collection system using combination of:

• Python for statistical analysis
• Google Sheets for live dashboard
• Apps Script for automation
• ChatGPT for qualitative analysis

Analysis revealed specific insights like:

• 33.63% chance of drawing Battle Pass in initial hand
• 4.5 average prizes remaining in lost games
• Stronger performance against certain deck archetypes
• Common patterns in game concessions and mistakes

Documentation process helped identify areas for improvement through systematic tracking of mistakes, successful strategies, and learning moments from each match