Game Theory & Monte Carlo Simulations for Predicting Tennis Match Outcomes (Ongoing)

-  Conducting analysis of different match formats using game theory and Monte Carlo simulations.

​

Detecting AI-Generated Text with DeBERTa_V3

-  Enhanced BERT model to identify AI-created text, achieving 99.93% accuracy, significantly outperforming non-tuned models like logistic regression and XGBoost(~95.5%).

​

Time Series Analysis of the COVID-19 Impact on the US Airline Companies Based on ARMA model (ICCIR '21)

- Completed in-depth analysis on the stock time series’ dynamics and made predictions based on the ARMA model.

​

LSTM Stock Prediction and Trading Strategies

- Utilized Long Short Time Memory(LSTM) recurrent neural network to develop a trading strategy: the first trading day of each month we want our model to tell us if we are going to stay in the market for the current month or not. Considering trading cost and tax, the model performed roughly a 45.28% net annual yield.

​

DMAC Analysis for Tesla (TSLA)

- The Dual Moving Average Crossover (DMAC) is a popular trend-following strategy used in technical analysis. It involves tracking two moving averages: a short-term average and a long-term average. When the short-term average crosses above the long-term average, it signals a potential bullish trend, known as the "Golden Cross". Conversely, when the short-term average crosses below the long-term average, it indicates a potential bearish trend, termed as the "Death Cross". In this notebook, we'll explore the DMAC strategy using Tesla's stock price data to identify potential buying and selling opportunities.

​

Decade-Long Dive: Equally Weighted vs. Dynamic Investment Strategies

- We meticulously examined the stock market, anchoring our study on an equally weighted portfolio of 12 stocks, steadfastly maintained for a decade against titans like the S&P500 and Nasdaq. Venturing further, we delved into prevalent investment tactics, from the renowned max Sharpe ratio to max return and the defensive min drawback strategies. The debate on integrating shorting strategies added another layer of intrigue. By determining the optimal weights for our 12-stock ensemble under each strategy, we vividly showcased their returns and portfolio values over time. Through compelling visuals, we juxtaposed these dynamic strategies against the industry's benchmarks, offering a holistic view of their long-term viability.

 

A Study on the Downforce Variation of F1 Racing Wings at Different Speeds, Air Pressures, and Angles of Attack

- Using ANSYS to analyze the downforce generated by F1 racing wings of the 2022 new regulations car under airflow, and compare the difficulty of following the car under the old rules.