Hello

FinTech Project

• Utilized scikit-learn to compare multiple machine learning models for predicting weekly price trends.
• Applied Value at Risk (VaR) as a grid search method for risk management in a moving average crossover trading strategy.
• Employed covariance shrinkage and the efficient frontier approach to optimize portfolio performance.
• Leveraged Perplexity API to monitor local news and utilized local AI tools for sentiment and event analysis.
• Develop a trading interface website based on AWS cloud architecture to realize price monitoring, notification push and strategy backtesting functions.
• Used Selenium to web scrape a decade’s worth of 10-K financial reports, building a financial ratio database. Applied Random Forest to analyze the relationship between financial ratios and stock price movements post-filing.
• Web-scraped and compiled 10 years of HKJC horse racing data, including historical odds, into a detailed database. Applied EDA, Optuna, and LightGBM to develop a predictive model for race outcomes, achieving an AUC of 0.798.