A team of mathematicians from the University of the Philippines Diliman is making waves in the realm of stock market analysis with their innovative use of topological data analysis (TDA).
Led by Ela Mae Riñon and Dr. Rachelle Sambayan from the College of Science Institute of Mathematics, this groundbreaking research focuses on predicting critical market events, such as crashes. By analyzing stock price data from January 2019 to January 2021 for three prominent Philippine companies—Century Pacific Food (CNPF), PAL Holdings (PAL), and Cebu Air (CEB)—the researchers have unveiled patterns that could change the way investors approach market volatility.
Decoding Data with Topological Insight
TDA offers a unique lens through which to view complex datasets, revealing underlying structures that traditional analysis might overlook. The researchers compare this process to identifying constellations in the night sky. At first glance, the stars appear random, but with careful observation, distinct patterns emerge. Similarly, TDA classifies data into three homology groups: connected components (open shapes), loops (closed shapes), and cavities (three-dimensional forms).
Notably, the study found that as stock markets near a downturn, data points begin to cluster together, indicating a shift in their topological structure. This clustering leads to a decrease in homology group persistence, serving as an early warning sign for potential crashes.
Implications for Future Research
The findings of the study, published in the Philippine Journal of Science in June 2024, reveal that CNPF and PAL demonstrated remarkable stability during the early COVID-19 pandemic, while CEB faced greater volatility. Despite minor fluctuations, the resilience of PAL and CNPF highlights the model's predictive power.
The researchers suggest further exploring TDA’s applicability to additional markets and time periods, as well as its use in analyzing various time series data, such as exchange rates. This could provide deeper insights into stock behavior across different economic climates and enhance forecasting methods.
The findings of this research were published in the Philippine Journal of Science in June 2024, marking a significant advancement in financial analysis techniques.
