Hudson River Trading

I interned as an Algorithm Development Intern on both the High-Frequency team and the ETF Arbitrage team in the New York City Office.

For my first rotation, I built a full end-to-end trading system for trading Brazilian equities. This involved reading in data from the exchange, creating a model for price prediction, and then implementing and deploying an order execution system that made decisions, tracked risk, and sent orders into the exchange. We were given capital to run our algorithms with, so my quotes were in the Brazilian exchange for a few days, which was super cool.

Between my first and second rotations, I had a training project to predict the impact of news headlines on price movements. I used embeddings and other categorical features to classify the magnitude and direction of certain classes of headlines. I was able to achieve some measure of statistical significance in categorizing price movements using non-parameterized ensemble methods.

For my second rotation, I had two projects. My first project was optimizing the holding cost of our completed arbitrages over many days. This involved deciding when to create/redeem our ETF positions with the fund to lock in an arbitrage. My second project was determining arbitrage opportunities outside market hours. For this one, I primarily focused on US markets and ETFs with US Equity underliers.