Home

This website is created for a Washington University ESE 499 Capstone Design Project by Sam Jones, advised by Professors Zachary Feinstein and Jason Trobaugh. The abstract and a link to the full report are below:

View Report

Cryptocurrency is a new asset class that struggles attracting investors with low risk tolerance due to their high volatility. The goal of this project is to use historical cryptocurrency trade data and machine learning models to generate prices that are better than market prices. This project implements an autoregressive linear model, several baseline machine learning models, and systems of machine learning models. It tests the accuracy of each in predicting the future 1 hour percent change in price of 10 different cryptocurrencies. Several models show significant predictive power and potential to make markets more efficient. However, this project is not able to conclude that margin trading cryptocurrencies provides a higher return on investment than a buy and hold strategy.