Bole Lin Email & Phone Number

Shanghai, China

Developed operators in C++ based on the company’s source code to process analyst forecast data and constructed various factors based on analyst predictions for Net Debt, ROE, and Free Cash Flow, ultimately obtaining several robust factors through different time-series and cross-sectional methodologiesBuilt a module to analyze analyst forecast divergence, derived factors related to divergence and market sentiment, and backtested factors across different time periods and stock universe to validate their robustnessProgrammed a data analysis module for the backtesting system’s web application using JavaScript, adding interactive features and visualization for the IBES dataset (multi-dimensional analyst forecast data)

Shanghai, China

Developed a daily-frequency CTA and minute-frequency cryptocurrency backtesting framework, optimizing the backtesting process through vectorized programming to enhance computing speedRefined factors such as warehouse receipt factor, inventory cycle, and operating rate by integrating fundamental logic, adjusted parameters to verify the robustness of factors, achieving a 30%+ improvement in out-of-sample performanceAutomated API-driven processes to retrieve and store real-time cryptocurrency price and volume data, resolving various errors and implementing automatic reconnection from the original point after disconnectionUtilized minute-level cryptocurrency data along with technical indicators, such as CMO, to develop momentum and reversal factors, re-balancing based on volume information and factor signals, achieving a Sharpe ratio of 1.6+

Shanghai, China

Utilized MySQLdb in a Linux environment to access and cleaned commodity futures fundamental dataConstructed factors based on production willingness and inventory cycle indicators, with the inventory cycle factor distinguishing between active and passive restocking, achieving a Sharpe ratio of 2Built an MLP model on residuals of ARIMA model for futures inventory and combined the predicted residuals with the ARIMA forecast results, with a R2 of over 90% and an improvement of over 20% compared to the base model

Shanghai, China

Cleaned data using Pandas and reconstructed order book by aggregating millisecond-level tick data, and extracted features from transaction data, with the addition of convolutional layer to reduce data dimensionsTrained Transformer deep learning model and tuned hyperparameters through grid search and OptunaDeveloped order splitting strategies using TWAP and MVWAP algorithms to minimize market impact and to achieve lowerthan-market execution price through volatility prediction

Shanghai, China

Bachelor Of Economics, Finance

Activities and Societies: Finance Club, Quantitative Investment Club

Education record