Cryptocurrency Portfolio Strategies with Machine Learning

Level: MA
Responsible person: Yu Zhang
Keywords:  Cryptocurrency, Portfolio, Machine Learning

Hundreds of cryptocurrencies, such as Bitcoin and Ether, are listed and actively traded on centralized exchanges, offering investors a new avenue for investment. Unlike traditional assets like stocks, cryptocurrencies lack underlying businesses to support their valuations, which makes the investment in cryptocurrencies very risky. A key characteristic of cryptocurrency prices is extreme volatility, with most cryptocurrencies exhibiting high positive correlation, which provides distinct opportunities for us to research assets portfolio trading for arbitrage profit. This thesis explores the use of machine learning methods to construct investment portfolio strategies that can help traders develop safer investment tools with still comparable profit rates by selecting which cryptocurrencies to long and short and recognizing proper trading signal. The key workload of the thesis lies in implementing different trading strategies from literature on cryptocurrencies and choose some strategies that can provide stable profit flow. 

References: 

Huck, Nicolas. "Large data sets and machine learning: Applications to statistical arbitrage." European Journal of Operational Research 278.1 (2019): 330-342.