Digital Experience Data

AWS EC2 Benchmarking

AWS EC2 Benchmarking

o Applied statistical methods (Neural Network, NLP) to build quantitative algorithms in Python (with TensorFlow, Pytorch, Keras, NLTK) and in C# (with ML.NET).o Built and improved the Algorithm Framework in LEAN (QuantConnect's open-source trading engine) to speed up the backtest process and enable it more pluggable for users to create new algorithms.o Developed data pipelines for different kinds of financial data (Equities, Options, Futures, Forex, and CFD) to perform data-preprocessing and keep them automatically updated in the online platform.

o Constructed three Python-based frameworks for trading cryptocurrency on HuobiPro based on its RESTful and WebSocket APIs, where users can perform live trading, backtest quantitative strategies and attain historical data respectively.o Developed a live data feed pipeline through the WebSocket API and built a trading API for users to perform live trading.o Built a backtest framework where users can evaluate strategies with holdings, orders, statistical criterions, and visualized plots.o Extended the RESTful API so that users can automatically obtain csv historical data from Json without limitation on amount.

o Initiated and developed two quantitative strategies based on python and R, including a stock strategy for ZZ500 and HS300 stock pools and an options strategy for 50ETF in Chinese stock and options market.o In the stock strategy, I adapted the multi-factors method with three fundamental factors like the return of equity and one technical factor and computed the weighted average of the ranks of the factor values in different industries to evaluate the stocks. Moreover, I set different capital for selected stocks based on the market capital percentage of their industries in the whole stock market in order to diversify risk. This strategy performed a 20%+ annual rate of return with 5%- max drawdown on 1-year historical data.o In the option strategy, I applied the arbitrage method based on the convexity of the option price. On 1-year historical data, there were 100+ possible opportunities for this kind of arbitrage, especially when the price of underlying assets fluctuated violently.