Data Scientist
Resume:
THOMAS LI
CONTACT
INFORMATION
Courant Institute of Mathematical Sciences +1-347-***-****
New York University ******@****.***.***
New York City, New York State 10012, U.S.A.
EDUCATION New York University, New York City, United States of America Doctor of Philosophy in Mathematics, August 2019
Dissertation Topic: Dynamic Pairs Trading Strategies Utilising Stochastic Optimal Control Guidance Committee: Peter Paul Carr, Andrew Papanicolaou, Deane Yang, G a oyoˇng Zh a ng Master of Science in Financial Engineering
Research Topic: Credit Risk of Secured Loans under Jump-Di usion Process Award: Received Merit-Based Postgraduate Scholarship Beijing Information Science and Technology University, Peking, P. R. China Bachelor of Science in Computer Science and Technology Dissertation: Interactive Network Educational System Award: Received Merit-Based Undergraduate Scholarship RESEARCH
INTERESTS
Financial mathematics, machine learning, and stochastic optimal control. PUBLICATIONS Machine Learning Methods For Trading VIX Futures, joint with A. Papanicolaou, submitted 2020.
• We study an application of machine learning methods for trading of VIX futures and ETNs. We implement tabular Q-Learning algorithm and deep Q-Learning method, which allow us to show the ability of the models to learn basic risk-on/risk-o trading rules for various VIX ETN trades using the contango feature of the VIX future curves. We build a simulation engine via a vector autoregression model to train the algorithm, and preform out-of-sample tests using Sharpe ratios, P&L distributions and the ROC curves. Dynamic Optimal Portfolios for Multiple Co-Integrated Assets, joint with A. Papanicolaou, under review 2019.
• Developed a multi-asset model that is used for long-term statistical arbitrage strategies based on the techniques from stochastic optimal control. A key feature of the model is that all assets have co-integration with a benchmark, which if sustained, allows for long-term positive pro ts with low probability of losses. The optimal controls are solutions to Hamilton-Jacobi-Bellman equations, to which we can introduce portfolio constraints, such as market neutral or dollar neutral, and if there is long-term stability we can project the portfolio’s growth rate. Kernel Method for Testing Co-integration and Causality of Time Series, Working Paper, 2017.
• Developed a data-driven machine learning method to study the co-integration and causality properties of time series. The kernel function of this model is formed through Euclidean distance formula with a penalty term that lters out the time series that are either not correlated or spurious correlated. The model is simpler than Engle-Granger test, and test results of SP 500 stock prices data in Python showed that it is as good as Engle-Granger model.
Optimal Pairs Trading with Time-Varying Volatility, joint with A. Tourin, International Journal of Financial Engineering, 2016.
• Developed a dynamic optimal pairs trading strategy that incorporated a time-varying volatility of the constant elasticity of variance type such that the trading pro t is maximised. The trading strategy is obtained by solving a Hamilton-Jacobi-Bellman equation through the nite di erence method. The parameters were estimated by the generalized method of moments. The model was tested in MATLAB with the prices of dual listed stocks, the average return and Sharpe ratio of out-of-sample tests are 65 percent and 1.6 respectively. WORK IN
PROGRESS
Multivariate Pairs Trading with Common ETF Benchmark Under Drawdown Constraint, joint with A. Papanicolaou.
• Developed a multi-asset pairs trading model based on the approaches of stochastic optimal control. This optimal pairs trading model incorporates the drawdown constraint of Grossman- Zhou Problem.
1 of 2
Momentum Trading by Stochastic Delay Di erential Equations.
• Developed a dynamic momentum trading strategy in Python based on short term momentum and long term mean reversion empirical properties of equity return. The model is developed with the stochastic delay di erential equations, and stochastic volatility. The trading strategies are calculated numerically by solving the associated Hamilton-Jacobi-Bellman equation through the radial basis function-generated nite di erences method. INDUSTRY
EXPERIENCE
XU Consulting Engineering, 2019-Present, New York City, U.S.A.
• Data Scientist. Develop and implement machine learning models and deep learning models. Hua-Ke Financial Holding, 2017-2018, Peking and New York City
• Actuary. Developed and implemented two actuarial models for pricing critical illness insurance products. The model implementation involved (i) identifying and sourcing date, (ii) performing data cleaning, (iii) developing the models in Python, (iv) back testing the models and comparing them to the current market prices, and (v) supporting models go live in production environment. Jia-Heng Investment Co., Ltd., 2006-2010, 2010-2014, Peking and New York City
• Financial Engineer. Implemented portfolio management strategies that were leveraged by the front o ce asset managers to conduct their portfolio. The portfolio management strategies were developed using stochastic control approach and mean variance portfolio optimization with Value-at-Risk constraint.
• Market Risk Analyst. The analyses involved (i) evaluating daily volatilities of commodity prices using ARCH and GARCH models to predict the future volatilise, (ii) analysing the trend and seasonality of commodities prices through moving average models, (iii) providing investment suggestions to the front o ce asset managers.
China Information Institute of Machine Industry, 2005-2006, Peking, P. R. China.
• Statistical and Data Analyst. Used collected sales data to develop marketing strategies by implementing linear regression, lasso regression, and ridge regression to support decision making of industry research report sales teams.
• Translated the book Technical Analysis: Power Tools for Active Investors by Gerald Appel from English to Chinese.
Digital China Co., Ltd., 2004, Peking, P. R. China.
• So ware Engineer. Implemented so ware testing tools to establish whether the developed so wares met target requirements, responded correctly to di erent kinds of inputs and threshold conditions, executed within the allowed run time, and delivered the expected results. TEACHING
EXPERIENCE
Instructor
Computational nance, Quantitative methods for nance, Probability and statistics, Recitation courses for stochastic calculus and option pricing. Teaching Assistant
Stochastic Calculus, Dynamic Assets and Option Pricing, Quantitative Methods for Finance, Statistical Arbitrage, Computational Methods for Finance, and Machine Learning for Finance. PROFESSIONAL
ACTIVITIES
Presentations
Finance Symposium at East China Normal University 2018, The Eastern Conference on Mathematical Finance 2017, Bloomberg Quantitative Finance Seminar 2017, International Conference on Computational Science 2016, Bachelier Finance Society World Congress 2016, Morgan Stanley Quantitative Finance Seminar 2015, SIAM Mathematical Finance and Partial Di erential Equations Conference 2015.
MISCELLANEOUS Computer Skills: C++, MATLAB, Python, R, Bloomberg, and LaTex Languages: Chinese, English, French (Beginning Level) Candidate for the Fellow of the Society of Actuaries Professional swimmer, national athlete level 2 of China 2 of 2
Contact this candidate