Mingfeng Yang Email & Phone Number

United States

Boston, Massachusetts, United States

1. Strategic Asset Allocation2. Systematic Macro Strategy3. Portfolio Construction Automation and rebalance4. ETL and data pipeline

New York, New York, United States

IEOR 4150 Probability and Statistics

New York, United States

1. Worked in Quant Central Team to enhance the current low latency trading and pricing engine using graph computation in C++, used parallel computation and incremental computation to accelerate tick-level calculation for trading and pricing. 2. Converted the pricing engine into a computational graph, integrated the function ID and outcome ID within each node, employed edges to depict the pricing calculation flow, implemented a local cache by harnessing a hash table to store both functions and outcomes, and leveraged Merkle Hashes to connect nodes and outcomes. 3. Fine-tuned the engine to reuse previous calculation outcomes stored in the local cache in real-time tick-level trading, leading to an impressive 70% reduction in calculation latency. 4. Designed a score-based purging strategy with a combination of LRU and LFU to enhance the efficiency of graph computation engine's cache management, decreased the running time by 50% and memory usage by 30%. 5. Implemented C++ APIs in IDL files to import features of Graph Computation solution to Graph Driver solution. 6. Utilized CUDA C++ extension to extend and optimize the primary metrics functions for trade processing to leverage GPU Accelerated Computing, reduced the 60% running time of tick-level pricing of 100k+ trades vs. multi-threading on CPU. 7. Established a pipeline for automated testing and parameter optimization for the trading engine using Bash and Python, incorporated format and syntax testing into the pipeline with Perl.

Beijing, China

1. Worked in Quant Development team to build a backtesting platform for option trading strategies and pricing packages.2. Developed the backtesting microservice using C++ and deployed on Apache Server in UNIX with MariaDB, supported 10k+ backtesting requirements of 100+ accounts per day.3. Designed a UI using PHP and Node.js, allowed user to post JSON request and invoke the backtesting service, displayed the backtest performance result on a dashboard implemented with jQuery, Dynamic Charts.4. Constructed a C++ package for option pricing, enabling the calibration of Heston model through FFT and particle filter of VGSA model using Monte Carlo Simulation, increased pricing accuracy by 23%.5. Designed option volatility arbitrage strategies including historical-implied volatility strategy and volatility direction strategy tailored for China’s market dynamics, achieved an annualized return of 26.39% and Sharpe ratio of 2.6.

Chengdu, Sichuan, China

1. Worked as a Software Engineer to build a quantitative trading platform for 20k+ clients using microservices architecture. 2. Built an ETL data pipeline with C++ and Python to extract data from provider using Navicat, clean, load into MariaDB.3. Implemented the trading platform services using C++, built a technical indicator module in C++ with LRU Caching to improve the calculation efficiency, reduced the response time by 50% and memory usage by 30%. 4. Implemented Unit tests and Integration tests, reached the coverage rate of 95%, leveraged log file to build a performance supervision system, yielding low failure rate of only 0.01%.5. Developed a web UI like Jupiter notebook with LAMP stack, allowed user to load own trading strategy in python, created the trading logic by invoking trading platform using TCP/IP Socket to send trading order and receive order books. 6. Leveraged Docker to setup python programming environment for the quantitative strategy’s generation, enhanced the scalability to support 10k+ visits per day.

Beijing, China

1. Collected financial data of listed companies in bond market, deriving 32 liability and operational indicators.2. Exerted Logistic Regression and Xgboost to build 3 rating models for bonds; the AUCs of model are all above 0.87.3. Wrote a program to integrate data extraction, model prediction and rating reports generation in Python.

Beijing, China

1. Queried the car loan data of clients of Bank of China and constructed modeling database in SQL with 352 variables. 2. Established 6 marketing prediction models for 3 car brands with Logistic Regression and Xgboost in Python for potential loan customer selection. Performed Decision Tree Binning to do feature engineering. The AUC and KS of models are all above 0.87 and 0.5 and the PSI of models are below 0.05.3. Used models to predict probabilities of loans of 80 million clients, increasing marketing success rate by 200%.

Master'S Degree, Operations Research, 4.13/4.33(Departmental Fellowship Awardee || Top 1%)

Analysis of Algorithm, Algorithmic Trading, Natural Language Processing, Computing method in Finance, Simulation, Operating System, High.

Bachelor Of Economics, Public Finance, 93.94/100 || Rank 1/33

C++ for Programming, C Programming, Python Programming, Asset Pricing

Exchange Program, Exchange Program, 4.0/4.0

Machine Learning, Applications for Data Analysis, Numerical Analysis, Computational Structure of Data Science, Advanced Programming in R