Dynamic Volatility Arbitrage Strategy Design
This is the capstone project for the course Time Series Analysis & Statistical Arbitrage. We implemented a volatility arbitrage strategy by predicting S&P 500 future volatility using GARCH-type models, contrasting it with VIX-implied volatility. Then, we employed dynamic trading rule taking long or short positions guided by a threshold of forecasted errors. The best model achieved a 14.56% annualized return with a 0.4747 Sharpe Ratio in Year 2018-2023 backtesting.
Research Lab Inventory Management System
Scenario: A student recently took up an on-campus job, managing inventory for a research lab. To streamline the process and improve efficiency, the student proposes a Research Lab Inventory Management System.
Project Details: We designed an application which includes GUIs to add new products, update product details, manage stock levels, and generate sales reports. Inventory data will be stored in files, simplifying record-keeping. The system significantly reduces manual workload and the likelihood of human error, ensuring more accurate inventory tracking.
Spotify Music Classification
In this project, I built machine learning models based upon 45,000 songs from Spotify API. After careful preprocessing, train/test split, and normalization, I adopted dimensionality reduction and clustering methods including PCA, T-SNE, and MDS. Then, I used the silhouette score to determine the optimal number of clusters and added the result of k-means as an extra label to boost classification efficiency. Eventually, I classified 5,000 songs into 10 genres via 4 models (SVM/Neural Network/Random Forest/Adaboost) under the convention for supervised learning. Model performances were evaluated using the ROC curve and the highest AUC score achieved was 0.92/1.00.
SQL Database - New York Stock Exchange
In this project, I created an entity relationship (ER) Diagram for traders, companies, and stock trades. Following the ER diagram, I built six different tables to assemble a simple stock exchange database which is capapble of performing complex queries including secure and structured transactions.