With the rise of artificial intelligence and machine learning, neural network models have shown great promise for making investing and trading decisions. For beginners looking to get into neural network investing, having the right resources and frameworks is crucial. In this article, we will provide an overview of the best open source neural network investing libraries like FinRL, resources like research papers and tutorials, as well as tips on how to get started.
FinRL – An open source deep reinforcement learning library for quantitative finance
FinRL is one of the best open source libraries for applying deep reinforcement learning to quantitative finance problems like stock trading, portfolio management, etc. It provides implementations of state-of-the-art DRL algorithms like DQN, DDPG, PPO tuned specifically for financial applications. FinRL also contains simulated trading environments using actual market data and comes with extensive backtesting capabilities out of the box. For beginners, the detailed tutorials make it easy to build and evaluate neural network trading strategies.
ElegantRL – High performance deep reinforcement learning library
ElegantRL is another excellent open source DRL library aimed at high performance and flexible customization. It implements many of the latest deep reinforcement learning algorithms for continuous control problems. For neural network investing, ElegantRL enables efficient strategy optimization and backtesting. The library is well documented with Jupyter notebook examples spanning use cases like trading cryptocurrencies, stocks, futures etc.
Awesome Deep RL in Finance – Curated list of research papers and other resources
For those interested in researching the intersection of deep reinforcement learning and finance, the Awesome Deep RL in Finance list aggregates all the latest breakthrough papers in this field. It covers interesting topics like applications of meta learning, multitask learning in finance and much more. This is a great starting point to survey the landscape of deep RL in quantitative finance research.
Start simple and iterate – Tips for beginners
For complete beginners, it is best to start simple and not get overwhelmed by the advanced techniques. Begin with basic DRL algorithms like DQN, use simulated environments for validation before applying strategies to live markets. Leverage prebuilt libraries like FinRL to avoid heavy coding. And most importantly, iterate extensively via backtesting and strategy optimization. This will build intuition on what works and what doesn’t.
Resources like FinRL, ElegantRL combined with relevant research papers and tutorials are invaluable for beginners aiming to use neural networks for investing. Focus on quick experimentation and iteration rather than advanced algorithms at first. Over time, significant alpha can be generated using these techniques.