Researchers
Chanakiya Sivakumar
Sriram Bharadwaj
Faculty Advisors
Dr. Khaldoun Khashanah
Summary
The paper focuses on cryptocurrency trading strategies, primarily on Bitcoin due to its accessibility and liquidity. The volatile nature of Bitcoin trading allowed us to experience a wide range of market conditions within a relatively short time frame. By analyzing different combinations of trading intervals and market environments, we evaluated the performance of three distinct trading strategies. These strategies were designed to maximize returns under specific market conditions: “Bull” and “Bear” served as foundational approaches, while “Flipper” utilized machine learning techniques, such as a Random Forest model, to dynamically switch between Bull and Bear based on market indicators.
The study found that both Bull and Bear strategies outperformed the market in their respective conditions, with the Bear strategy demonstrating strong resilience across all market types. In volatile periods, the Bear strategy excelled over the other two due to its trend-following approach, which aligned well with Bitcoin’s inherent price fluctuations. On the other hand, while the machine-learning-based Flipper strategy was able to switch between Bull and Bear conditions, it did not outperform either in a notable way.
Future work could focus on enhancing the machine learning model to improve its accuracy in identifying market conditions. Additionally, live testing using real-time data, as opposed to historical datasets, would provide a more robust evaluation of these strategies. If live results align with the historical data, incorporating technical analysis into these trading strategies could further optimize returns.
