Comprehensive Product Insights

(1) Use data to understand competitors’ performance and to identify improvement opportunities.(2) Build statistical, optimization, and machine learning models for applications including pricing, targeting, and experimentation.(3) Present findings to senior management to inform business decisions.(4) Collaborate with cross-functional teams across disciplines such as product, engineering, and operations to drive competitor intelligence development end-to-end from ideation to productionization.

(1) Create automation pipeline for Company's core product(2) Work on Amazon Redshift, analyzing COVID impact, customer behaviors on different products(3) Use Lasso regression, Deep Learning Algo to improve Actuary assumptions(4) Help write machine learning Python customize library (5) Simulate non-exsit data distribution to study product behavior(6) Study customers' abnormal behavior and create windfall for company(7) Detact error of implementing exsiting model which worth $10M(8) Helped on building normalized database on Redshift

NLP analysis on Chinese

• Developed efficient algorithm on 100M+ global equity weekly data queried from SQL database, winsorized and industry neutralized variables value before regression; carried out unit tests to validate code• Used Fama-Macbeth weight-least-square regression studying the relationship between ESG pillars and future stock crash risk, adjusted by fixed effects and residues are double clustered on firm and time• Customized stable skewness measure combining Fisher-Pearson and Bowley methods; created firm crash risk factor by exponentially weighting the skewness, developed betting-against-crash strategy

• Analyzed data distribution, filled NA with neutral filling method; designed algorithm for data processing on 20M+ data with high code efficiency; implemented Three-Pass Regression filter obtaining PLS index• Carried out Fama-Macbeth regression with Newey-West adjustment, verified feasibility of trading model• Built two-quarter lag regression model predicting future return; created PLS long-short trading strategy in Python, achieved annual return of 13.8%, Sharpe Ratio of 1.86 on 30-year back-test• Improved strategy by quantile transformation in Sklearn package on PLS index, with Sharpe Ratio raising to 2.1

• Applied k-nearest neighbor algorithm to impute missing value for fund’s characters• Ran Classification and Regression Tree (CART) Model and Random Forest Model to identify superior investment managers, achieved F1-score of 0.92 on test data, with area under receiver operating characteristic curve (ROC) of 0.8

· Developed machine learning trading strategies using Random Forest algorithm; monitored the working of main strategy managing futures worth over 45 million dollars, modified critical parameters when necessary· Studied 80+ strategy papers, designed a trading indicator based on momentum, bringing over 10% extra annual return on exsiting algorithm· Formed self-adjusting system encapsulated as a class in Python to automatically update parameters, increased main strategy's Sharpe ratio by 50%· Created IC index future middle-frequency strategy; unlevered annual return rose up to 25% on 30-day back-test

· Analyzed 12,000+ news items on Wind for clothing industry, selected about 400 that were important enough to potentially influence investor decisions· Researched 40+ papers on HUAFU stock (002042.SZ), led the intern team to analyze the company using fundamental tools like DCF and ratio analysis

· Filtered and standardized 10,000+ pieces of customer data with tools implemented in Python; selected approximately 14 effective indicators out of 50+ through linear regression and lasso analysis· Established risk expectation model in C++ with intern team; model is used by manager to predict customers' credibility

· Built four versions of data managing systems in Python for sorting data by time indexes, converted raw data into timetables; ultimately sorted data for 30+ stocks and helped to form investment decisions· Collaborated with strategy managers to back test on 80+ indicators; worked closely with team of interns to implement strategies using MACD applied on portfolio designed by senior staff