Duo (Ted) Sun
Cornell University, College of Engineering, New York, NY Master of Engineering in Financial Engineering, Financial Data Science Certificate, GPA: 3.6/4.0 December 2018 CME Futures Trading Competition Finalist Spring 2018 Imperial College London, Department of Natural Science, London, UK Bachelor of Science in Mathematics and Statistics for Finance (G1GH), Upper Second Class Honor July 2017 Imperial College Undergraduate Research Excellence Award of 1,500 GBP Summer 2016 Selected Coursework: Statistical Data Mining, Time Series Analysis, Stochastic Calculus, Monte Carlo Methods, Quantitative Trading Strategy, Quantitative Portfolio Management, Optimization in Finance, Market Microstructure & Real-time Risk, Financial Statement Analysis, Fixed Income & Derivatives, Spreadsheet Modeling SKILLS & CERTIFICATES
Technical & Language: Python, R, SQL, MATLAB, C++, Linux, Excel VBA, Bloomberg, German (Intermediate) Certificates: Algorithm Tool Box and Data Structure in C++, UC San Diego, Coursera Summer 2017 EXPERIENCE & PROJECTS
Alpha Generation from Fund Analysis, Project Sponsor: Rebellion Research, Cornell University, New York, NY Fall 2018
• Led a group of six to analyze 13-F filings data from 8000 funds over 20 years in order to capture alpha by predicting stock returns and replicating intelligent funds’ portfolio ideas, heavily using Python and MySQL.
• Obliged SEC regulations to warehouse 13-F filings and cleaned price data ready to build features and construct models.
• Used correlation analysis to classify funds into categories, such as industry sectors, value or growth, idiosyncratic risk, survival time, money inflow and number of funds change holdings, then used those classifications to construct features.
• Applied XGBoosting and Logistic Regression on above features to predict stock price movement and built equally weighted portfolio, which had a 15% annualized return and 0.98 Sharpe ratio from March 2013 to December 2017.
• Chose high return funds with small turnover to replicate their holdings, with 19.8% return and 0.78 sharpe ratio. Quantitative Analyst, Hite Capital Management (Recurve Asset), New York, NY Summer 2018
• Constructed momentum strategies on equity & vertical spread options and built backtests, using Python, SQL and R.
• Programmed over 30 GB NASDAQ100 Composite Stocks’ historical options chain from 2000 to 2018 from third party I-Volatility and cleaned adjusted price with dividend and split.
• Generated long and short trading signals from Exponential Moving Average, On-balance Volume, Moving Average Convergence/Divergence and Relative Strength Index, and also analyzed the parameters of signals ranking constraints.
• Coded the strategy and conducted performance & holding period analysis and generated report in LaTeX. Natural Language Processing in Stock Market Prediction on Daily News, Cornell University, Ithaca, NY Spring 2018
• Used daily “top 25” news headlines over 8 years to predict Dow Jones Industrial Average (DJIA) movement, using Python and Classification Machine Learning techniques heavily.
• Constructed datasets in two ways: first with the overall daily sentiment obtained from VADER analysis, and then with the words’ dictionary of news from standardized frequency of appearance based on TF-IDF analysis.
• Reduced dimension of the dictionary by Truncated SVD and TSNE to better visualize data and prevent overfitting.
• Built classification models, such as KNN, LASSO regularization, rocchio classification, random forest and support vector machine in order to predict next day’s Dow Jones movement direction. Macroeconomic Factors Hedging Model in Portfolio Management, Cornell University, Ithaca, NY Spring 2018
• Chose a stock pool from Quantopian based on the scoring system on their fundamental performance and run dynamic linear regression analysis of those stocks’ prices versus leading macro-economic factors to build sensitivity matrix.
• Applied Markowitz theory to minimize the portfolio CVaR with suitable constraints in MATLAB CVX solver.
• Improvised an alarming system based on the change in Consumer Sentiments in order to better gauge rebalancing.
• Built a portfolio with 0.71 Sharpe from 2004 to 2017 and only 19% yearly maximum drawdown in 2008 Crisis. Markov Regime Switching in Credit Portfolio Construction, Cornell University, Ithaca, NY Spring 2018
• Conducted Markov Regime Switching model on S&P 500 to find the normal and crisis state.
• Applied PCA on collections of Fama Bliss Yield from 2000 to 2017 in different states and used top two PCs to run regression on ICE BofAML US Corporate Effective Yields.
• Constructed credit portfolio by longing corporate bonds and shorting a Treasury Bond portfolio with weights associated to regression coefficients and PCs in order to capture the credit premium. The return of BB corporate bonds were 3.15% with 1.6 Sharpe; the return of CCC corporate bonds were 2.93% with 3.36 Sharpe. Consumer Credit Risk Analyst, Industrial and Commercial Bank of China (ICBC), Beijing, China Summer 2015
• Built logistic regression model for over 20,000 credit card applications to predict creditworthiness and set credit line.
• Applied advanced machine learning techniques, such as discretization, WOE transformation, stepwise selection and decision trees on R, during logistic regression model construction.
• Performed due diligence in Know Your Customer, including Name Screening and Customer Credit Report refresh.