Inspiration
We are driven by two primary passions: the potential for market profitability and the intellectual challenge of implementing complex algorithms. We wanted to bridge the gap between theoretical computer science and real-world financial markets.
What it does
This project is an automated trading bot that executes strategies within the Optiver simulated market. It autonomously manages buy and sell orders to capitalize on market movements.
How we built it
We engineered the system using Python, iterating through a wide variety of strategies ranging from modern Machine Learning approaches to classic financial algorithms. Through rigorous testing, we eventually pivoted to a robust Market Making strategy.
Challenges we ran into
Our team faced a steep learning curve regarding the intricacies of High-Frequency Trading (HFT). Additionally, balancing rapid, intuitive prototyping ("vibe-coding") with the need for systemic stability proved difficult.
Accomplishments that we're proud of
We are proud that we ventured into the complex domain of HFT and successfully deployed a functional strategy in a competitive environment, moving beyond theory into execution.
What we learned
We learned the critical importance of API documentation analysis. Specifically, we gained a deeper understanding of rate limiting, realizing too late that the 25 RPS (Requests Per Second) limit applied globally across all queries, not per endpoint.
Implementation Details
We used a market-maker strategy. Our algorithm looked at stocks whose price was stable at the moment (there was no news about the stock and the price had not changed significantly in the last 3 seconds) and offered to buy/sell stocks at slightly less favourable terms than the best offer. Trivial filters based on regular expressions were used to classify news.
