Decoding Sector Valuation Dynamics
Technologies:
Data Manipulation and Visualization Libraries:
Pandas, Matplotlib, Seaborn, Plotly
Machine Learning Libraries:
cuML, LightGBM
Other Libraries:
SHAP, SciPy, Statsmodels, NumPy, CuPy
My honors thesis, “Decoding Sector Valuation Dynamics,” explores the drivers of equity valuations within Global Industry Classification Standard (GICS) sectors. It takes the form of a predictive analytics project, where I utilize company fundamentals, economic indicators, and various commodities to explain the variance in stock returns. I conducted the project using Python, employing multivariate regression and gradient-boosted decision tree models within Jupyter notebooks. The final report was composed in LaTeX.
Précis
This study examines the key drivers of equity valuation across various sectors and industries. Equity valuation is a critical metric that captures a company’s perceived worth and influences investment decisions, mergers and acquisitions, and risk management strategies. Understanding the primary factors impacting valuation is essential for market participants to make informed choices.
The research seeks to unveil determinants of equity valuation that may have been obscured by traditional linear regression techniques. With modern ensemble machine learning methods, particularly gradient-boosted decision trees, the study aims to leverage non-linear relationships among a diverse set of variables and fluctuations in stock prices, which serves as a proxy for shifts in company valuations.
The analysis employed both linear regression and gradient-boosted decision tree models to explain the variance in monthly stock returns across various sectors within the Global Industry Classification Standard (GICS). The results uncover substantial variations in the factors influencing equity valuation between sectors. Various commodities, economic factors, and company fundamentals emerged as the primary determinants, with their relative importance contingent upon the sector context. Particularly noteworthy is the consistent superiority of gradient-boosted decision tree models over linear regression methods, underscoring the significance of capturing intricate non-linear patterns in financial data analysis.
The identified drivers provide crucial insights that can enhance investors’ decision-making, improve valuation modeling practices, and strengthen risk management strategies. By uncovering these pivotal factors, the study aims to guide market participants in navigating the complexities of financial markets with greater foresight.
Future research could explore the incorporation of additional data sources, such as forward guidance from companies, analyst forecasts, news events, and social media sentiment, to further enhance the predictive power of the models. Additionally, investigating industry-specific, sub-industry-specific, and global market dynamics could furnish valuable insights into the determinants of equity valuation transcending the context of the United States and broad sector approach.
Overall, this study contributes to a more comprehensive understanding of the dynamic forces shaping equity valuations, empowering investors, analysts, and industry stakeholders to make more informed decisions and refine their valuation methodologies.