Documentation is currently being written
Abstract
Deep reinforcement learning (DRL) has proven to be a viable way for automatic portfolio management in many financial investment products, including the top cryptocurrencies in market capitalisation. However, implementing a practical and beginner-friendly DRL agent can be difficult because of the different goals, strategies, and libraries available. In this paper, we are implementing the FinRL has proven to be a viable way for automatic portfolio management in many financial investment products, including the top cryptocurrencies in market capitalisation. However, implementing a practical and beginner-friendly DRL agent can be difficult because of the different goals, strategies, and libraries available. In this paper, we are implementing the FinRL library for automated cryptocurrency trading and comparing the different agents across several librairies on different trading strategies. Furthermore, we are exploring the viability of the models by validating during periods or turmoil and bear trend, allowing an assessment of whether they are viable even in bad market conditions. Our models were trained and tested on the top 10 cryptocurrencies by market capitalization at the start of training period.) library for automated cryptocurrency trading and comparing the different agents across several librairies on different trading strategies. Furthermore, we are exploring the viability of the models by validating during periods or turmoil and bear trend, allowing an assessment of whether they are viable even in bad market conditions. Our models were trained and tested on the top 10 cryptocurrencies by market capitalization at the start of training period.
├── experiments
└── Deep_Reinforcement_Learning_Models_Performance_Cryptocurrency_Trading.pdf ...