Carry Trade Using Bond Interest Rate Signals

- 1.245 Sharpe, 83.3% Win Rate, 15,132 trades executed in a 10-year time period, 175.762% PnL (10.68% YoY) matching S&P 500 returns from 2010-2019, little to no correlation with S&P 500 and USD is not used as a long or short currency

- Created a carry trade algorithm across 4 ForEx currencies (Swiss Franc, Japanese Yen, Australian Dollar, Canadian Dollar) using 16 separate bond interest rates from federal reserve data (1962 - 2024) and 10 year out-of-sample backtesting verification

- Programmed 5 separate algorithms to optimize performance: Linear Regression, Histogram-Based Gradient Boosting Regression, Convoluted Neural Network (4 hidden layers), Long Short-Term Memory Network, Weighted Decision Tree + Random Forest

- Utilized various data manipulation techniques like sliding windows, lookback periods, absolute and relative data margin, and imputers/filters for NaN data points as well as continuous training even during out-of-sample testing

Categorization of Tennis Swing Using a Recurrent Neural Network in Human Activity Recognition

- Utilized an LSTM to classify the following tennis strokes: forehand, backhand, backhand slice

- Novel approach implementing sliding windows in HAR, detailing model performance in attached research paper (unpublished)

- Proved effectiveness of LSTM, outperforming RNN, CNN, GRU; Achieved 99%+ prediction accuracy when categorizing

- Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), Convoluted Neural Network (CNN), Gated Recurrent Unit (GRU)

Not in publication

Pairs Trading Optimized Using Genetic Algorithm

- Implemented a pairs trading algorithm using PNC (long) and BAC (short), backtesting the strategy from 2013-2023, including and excluding 2020

- Cross-correlated a set of banks as assets to determine a possibly profitable pair with a maximum bound correlation coefficient

- Chosen banks for testing included: Bank of America (NYSE: BAC), JP Morgan Chase (NYSE: JPM), US Bancorp (NYSE:USB), Citigroup (NYSE: C), Wells Fargo (NYSE: WFC), PNC (NYSE: PNC), Charles Schwab (NYSE:SCHW), Morgan Stanley (NYSE: MS), Goldman Sachs (NYSE: GS), Truist (NYSE: TFC), ICICI Bank (NYSE: IBN)

- Tested various executions and optimized trading timing using a Genetic Algorithm (PyGAD), an algorithm typically used to identify certain genes responsible for behaviors or attributes

- Achieved Sharpe Ratio of over 3, indicating highly successful returns in the 2013-2023 testing window; however, this Sharpe Ratio doesn't include transaction costs, projecting higher-than-actual returns

H&M Supply Chain Optimization Machine Learning Backtesting Algorithm

Note: This project was completed during the final round during the interview process for the BI, Data & Analytics Track under the H&M Group Trainee Program

- Implemented a backtesting algorithm for two machine learning models aiming to optimize the clothing brand's resources to maximize profit and reduce waste

- Given a large dataset containing the real sales info, and then comparing it to the model's predictive outputs, the algorithm created a new metric to identify which algorithm continued to predict an efficient, profit-mazimizing strategy comparatively

- Numerous metrics were tested and visualized before concluding which machine learning model was more accurate and could provide higher sales for the clothing brand while minimizing waste

Artificial Intelligence and Machine Learning in Financial Risk Modeling Presentation

- Created a high-level presentation covering the basics of Machine Learning from linear vs non-linear models to supervised versus unsupervised models

- Presentation time was around 30 minutes and the goal beyond the basic algorithm was to describe how these models could help provide a stronger method for identifying risk in markets, later shown in the case studies

- Models and methods presented included: Principal Components Analysis, Ridge regression, Partial Least Squares, LASSO, LARS, Elastic Nets, Neural Networks, Deep Learning, Support Vector Machines, Decision Trees, K-,X-means

IBM Stock Pitch

Note: if access to model is requested, please contact me via the email address or phone number provided in the Contact section

- Analyzed and modeled the company International Business Machines (NYSE: IBM); rated a BUY in December 2023 (+17% return)

- Learned about the company's business model, products, segments, leadership, and trends; Compared products to market size and competitors to understand if IBM had a reasonable competitive advantage

- Performed a full financial analysis using the company's 10-K via BamSEC, proposing a roughly 5% upside in its base case

Abercrobmie & Fitch (NYSE: ANF) Product Improvement Project

- Created a survey to gather information from over 100 participants; performed data analysis on the data combined with market information

- Suggested three separate implementations to help increase sales for the Hollister subsidiary of Abercrombie & Fitch