|Current Position||PhD Candidate|
|Employer||Stevens Institute of Technology|
|Degree||PhD Financial Engineering - May 2023|
I currently am a first-year Ph.D. student in financial engineering at Stevens. I got my bachelor degree of science in mathematics from Nankai University, Tianjin China. Then I got my master degree of science in financial engineering at Stevens. My research interests include machine learning algorithms and their applications in finance, reinforcement learning and optimal control.
Doctor of Philosophy in Financial Engineering, Stevens Institute of Technology, Hoboken, NJ
Expected May 2022
Master of Science in Financial Engineering, Stevens Institute of Technology, Hoboken, NJ
August 2017 - May 2019
Bachelor of Science in Mathematics and Applied Mathematics, Nankai University, Tianjin, China
August 2013 - June 2017
Hanlon Financial Systems Lab
Research Assistant, September 2017 - May 2019
- Research on deep reinforcement learning (RL), developed distributed simulation system for RL, implemented RL algorithms (PPO, DDPG, etc.) and controlled robotic agents across Internet by RL models.
Clearpool Group & Stevens Institute of Technology
Graduate Researcher, March 2018 - July 2019
- Research on dark pool liquidity detection based on ensemble machine learning and traditional statistical models; analysis on the liquidity signals and transition patterns. Improved models are deployed to production.
WorldQuant, LLC, Beijing, China
Intern and Virtual Quantitative Researcher, June 2016 - November 2017
- Intern: Research on technical alpha strategies for stocks with top liquidity in U.S. stock market.
- Virtual Quantitative Researcher: Developed alpha strategies based on fundamental analysis and analytical data. Applied machine learning algorithms to alpha strategy development on web based strategy backtesting system.
My research interests include Reinforcement learning algorithms and their applications in finance.
- Construct simulation system for Delayed Markov Decision Process (DMDP) model.
- Parametrized state transition model using variants of Recurrent Neural Networks.
- Combined main-stream actor-critic model with state transition model to improve its performance in delayed environment.
- Reformulated the optimal execution into dynamic control frameworks and solved by policy optimization.
- Applied continuous and discrete control reinforcement learning algorithms to nd optimal limit order price.
- Used realistic market replication system to justify the trained models and typical optimal execution benchmark.
- Constructed fundamental and price-volume database and data feeding interface using web scraping and Bloomberg API.
- Built event-driven back-testing system to seek for valid alpha signals, generate statistics and visualize portfolio performance.
- Built automate trading system aggregated with portfolio optimization in terms of target Beta and cash neutralization.
- Create a system which can control both simulated and physical robot through Robotic Operation System (ROS).
- Allow user to control multiple agents across internet and to stream camera input from agent to control side.
- Some main-stream reinforcement learning algorithms have been implemented and tested on this system.
- Provost Doctoral Fellowship Award, Stevens Institute of Technology, June 2019
- Outstanding Academic Achievement Award, Stevens Institute of Technology, May 2019
- Master's Fellowship Award, Stevens Institute of Technology, August 2017
- Meritorious Winner, Interdisciplinary Contest in Modeling, awarded by COMAP, April 2017
- Gold Medal (Global Rank: 62), WorldQuant Challenge, WorldQuant LLC, May 2016
- The 3rd-Class University Scholarship, Nankai University, November 2013