Read Your Stocks Via Screenshots

In this talk, we’ll build a Python app that extracts stock transaction data from screenshots or documents. We’ll refine screenshot extraction accuracy using OCR and regex in an interactive lab environment, store structured data in DuckDB, and visualize insights with Streamlit—transforming raw data into actionable trading insights. This approach is highly adaptable and can be applied to various industries.

Basic Python skills

In this talk, we’ll build a Python app that extracts stock transaction data from either screenshots or documents. To ensure high OCR accuracy, we‘ll develop an interactive lab environment to visualize every step and dynamically refine regex patterns.

Once extracted, the data will be stored in DuckDB, and with just a few lines of code, we’ll build a beautiful, interactive, and insightful dashboard using Streamlit. This will allow us to track performance, analyze profitable stocks, and gain insights into trading behavior—all from screenshots and documents!

What’s more, this approach is easily adaptable beyond finance. The same techniques can be applied to invoice processing, contract analysis, medical records, and more. By the end of this talk, you’ll learn how to transform unstructured screenshots and documents into structured, actionable data, unlocking new automation possibilities across industries.