Finoptix Portfolio Project

Welcome to my Finoptix portfolio repository! This project highlights a series of assignments and a final capstone implementation showcasing my skills in quantitative trading, machine learning, and advanced portfolio optimization.

Table of Contents

1. Overview
2. Project Structure
   • 1. Quantitative Trading
   • 2. Machine Learning
   • 3. Portfolio Optimization
   • 4. Final Implementation
3. Setup and Installation
4. Technologies Used

Overview

Throughout this project, I developed robust financial models ranging from basic technical trading strategies to complex, machine-learning-driven portfolio allocation frameworks. The highlight of the project is the Final Implementation, which blends predictive modeling (XGBoost) with financial heuristics (PE/DE ratios) and the Black-Litterman optimization model to construct a dynamic, robust stock portfolio.

Project Structure

The repository is divided into four main sections:

1. Quantitative Trading

File: 01_Quantitative_Trading/Bitcoin_Trading_Strategy.ipynb

• Developed a backtested trading strategy for Bitcoin (BTC-USD).
• Utilized technical indicators including MACD (Moving Average Convergence Divergence) and RSI (Relative Strength Index) to generate buy/sell signals.
• Integrated risk management techniques using ATR (Average True Range) for stop-loss and trailing take-profit mechanisms.

2. Machine Learning

File: 02_Machine_Learning/ML_Feature_Selection_and_Classification.ipynb

• Performed extensive Exploratory Data Analysis (EDA) on a given dataset (correlation heatmaps, distributions).
• Executed feature selection using Mutual Information, PCA (Principal Component Analysis), and Random Forest feature importances.
• Evaluated multiple classifiers (Logistic Regression, Random Forest, Decision Tree) with cross-validation to achieve the best ROC-AUC score.

3. Portfolio Optimization

File: 03_Portfolio_Optimization/Fama_French_Black_Litterman_Optimization.ipynb

• Implemented the Fama-French 3-Factor Model to estimate the expected returns of a diverse stock portfolio.
• Transitioned into the Black-Litterman Model, integrating market equilibrium returns with specific investor views.
• Calculated optimal weights for the portfolio, utilizing PyPortfolioOpt to maximize the Sharpe Ratio and plot the efficient frontier.

4. Final Implementation

File: 04_Final_Implementation/Stock_Ranking_and_Portfolio_Allocation.ipynb This notebook brings everything together into a robust stock picking and allocation pipeline:

• Feature Engineering: Calculated rolling volatilities, moving averages, momentum, and lag returns for a large universe of Indian stocks.
• Predictive Modeling: Trained an XGBoost Regressor to predict expected future returns for each stock.
• Fundamental Ranking: Ranked the top stocks by combining the ML-predicted returns with fundamental metrics like PE Ratio, Debt-to-Equity, and Market Cap.
• Portfolio Allocation: Applied the Black-Litterman Model on the top selected stocks, formulating views based on the model's predictions, to generate the final, optimal portfolio weights. Simulated the performance against an equal-weight market portfolio.

Setup and Installation

1. Clone the repository:

   git clone https://github.com/your-username/finoptix-portfolio.git
   cd finoptix-portfolio

2. Install dependencies: It is recommended to use a virtual environment.

   pip install -r requirements.txt

3. Run the Notebooks: Launch Jupyter Notebook or Jupyter Lab to explore the files:

   jupyter notebook

Technologies Used

• Data Analysis & Processing: Pandas, NumPy
• Machine Learning: Scikit-Learn, XGBoost
• Financial Modeling: PyPortfolioOpt, Statsmodels
• Data Gathering: yfinance
• Visualization: Matplotlib, Seaborn, Plotly