Program Website: https://msfe.illinois.edu/ 
                            
                        
                            
                             
                            
                        
                            
                             
                            
                        
                            
                            * Implement production algorithms for S&P credit analytics statistical machine learning extensively served as Capital IQ back-end programs on both Windows and Linux platforms used by worldwide S&P clients 
                            
                        
                            
                            * Develop FX trading infrastructure using Python, C++, C# 
                            
                            * FIX (Pricing/Dropcopy), API (Rest, SOUP, RabbitMQ), trading platforms connection in C++, C# 
                            
                            * Develop tools for Traders, Supports, and Trading Operations 
                            
                            * Build online alert systems and web dashboards to display key data and configure trading settings using HTML,Javascript/JQuery, PHP, CSS/Bootstrap 
                            
                        
                            
                            * Lead the team to perform algorithm model training on limit order books and trade records of high frequency Crude Oil futures and E-Mini futures 
                            
                            * Merged data in Python to generate 76 attributes including mid price, volume imbalance, weighed book price, time lag trade volume, etc. to prepare model training 
                            
                            * Modeled high frequency market via various machine learning models (Logistic Regression, Neural Network,SVM, Decision tree) in C++, Python & R to understand market pattern and make predictions market trends 
                            
                            * Backtested all training models in the test set with real high frequency data and achieved 86% accuracy 
                            
                            * Reimplemented and improved iceberg order detection algorithms to identify iceberg orders 
                            
                            * Summarized the weekly progress of the team and wrote weekly technical reports to CME supervisors