Project: JackpotMath

Hi, I'm Ivan S., a Data Scientist and recent MS graduate from the University of Colorado Boulder (Dec 2025).

JackpotMath began as an independent research project. My goal was to apply the advanced concepts I studied in graduate school to a real-world, high-entropy dataset. I wanted to determine if modern machine learning algorithms could detect signal in the noise of lottery drawings, or if they would rigorously confirm that the system is truly random.

This passion project turned into a bit of a rabbit hole in the best way. I used it as an opportunity to explore specialized techniques and apply methods in new contexts: implementing custom Bayesian models, diving into causal inference frameworks, experimenting with advanced time series analysis, and exploring network theory applications. Working through these implementations on my own deepened my understanding of when and how to apply different analytical approaches.

A Note on Methodology

Could this have been simpler? Absolutely. A Chi-Squared test would confirm randomness in minutes. But I approached this project as a rigorous exercise in signal processing: treating the lottery as a high-entropy dataset to determine if modern algorithms could distinguish meaningful signal from pure noise.

By testing for patterns using diverse methodologies (Bayesian inference, causal inference, information theory, deep learning, and network analysis), I established what statisticians call a "negative proof." When linear models, geometric approaches, and probabilistic frameworks all independently converge on the same conclusion, you can be confident in the finding: no hidden structure exists.

The lottery data proved to be exactly what it should be: high-entropy randomness that withstands scrutiny from even the most sophisticated analytical techniques. Sometimes the most interesting result is confirming that there's nothing interesting to find.

The Project Pipeline

Beyond the public tools on this site, I built a 12-module analysis pipeline to stress-test the data. This project allowed me to implement and evaluate a wide range of methodologies:

*Note: This project is for educational purposes. The comprehensive negative results across these models serve as statistical evidence that these lotteries behave as random systems.