Johnny Y's Senior Project Blog

Project Title: The Meme Economy: Utilizing Natural Language Processing to Predict “Meme” Cryptocurrency Price Movements BASIS Advisor: Rob Lee Internship Location: University of Arizona Computer Science Department Onsite Mentor: Dr. Mihai Surdeanu

Project Abstract

This project aims to determine whether natural language processing (NLP) can effectively predict price movements of "meme cryptocurrencies"---digital currencies driven primarily by internet culture and social media phenomena---by analyzing social media sentiment. This question is particularly relevant because meme coins, unlike traditional cryptocurrencies, are primarily driven by social factors rather than fundamental value. Success would demonstrate NLP's potential as a valuation tool for assets driven by "mindshare" and "social support," while contributing to the growing body of research at the intersection of artificial intelligence and financial markets. The research builds on existing work applying NLP to stock market and major cryptocurrency analysis, but focuses specifically on the understudied meme coin market, which has a significant market cap of $123 billion as of November 2024. The project was conducted at the University of Arizona Computer Science Department under Dr. Mihai Surdeanu, an expert in NLP and financial text analysis. The project includes development of a sentiment analysis classifier to analyze social media posts from Reddit about selected meme coins. The model was optimized through trial and error to generate trading signals, with performance measured against that of a strategy utilizing ARIMA, a statistical model. The research addresses challenges like filtering out automated "bot comments" that attempt to artificially inflate prices. The goal is to be able to use sentiment analysis to achieve significantly higher returns than ARIMA.