Zhengyao Jiang

Chevron down

Cyptocurrency Portfolio Management with Deep Reinforcement Learning

I'm fascinated about IT and math since high school and started to do academic researching at the end of undergraduate Year 0. Thanks to Dr Jinjun Liang, who gave me a chance and became my supervisor, I embarked on a long-term research project about applying deep reinforcement learning to portfolio management problems. Our research paper was accepted for publication at Intellisys2017 in London \cite{Jiang2017}, and as the first author, I did an oral presentation at that conference. Thereafter, we improved this work and extended the conference paper to a journal article in Year 1. In this work, we designed a domain-specific neural network training algorithm that efficiently leverages the training data and proposed a novel network architecture, which eventually surpass the traditional algorithms. We also released a portfolio management research toolbox which also contains the source code of our published algorithms on GitHub \footnote{https://github.com/ZhengyaoJiang/PGPortfolio}. The repository, PGPortfolio, has achieved 700 stars and the algorithms have been applied to real-world trading by some third parties. The conference paper and arxiv preprint of the journal article has been cited 25 times in total according to the google scholar \footnote{https://scholar.google.co.jp/citations?user=J8pFrgwAAAAJ&hl=en}. Besides financial topics, I'm also interested in general AI topics and current doing related research projects, for example, about Markov Decision Process and Differentiable Inductive Logic Programming. I also once been in a team to do cryptocurrency live trading which tries to apply the algorithm in the real market.

Buttontwitter
This website uses cookies to ensure you get the best experience. Learn more