Statistical Methods in Finance 2025

Beyond Returns: Count Time Series Methods in Finance

By:TV Ramanathan
Plaksha University

Traditional financial statistics is mostly centered on continuous variables such as prices, returns, and volatilities. However, many financial phenomena are inherently discrete and take non-negative integer values. Count time series models offer a rigorous framework for analyzing such data, capturing serial dependence, overdispersion, zero-inflation, and other stylized features common in financial markets. In recent years, these models have become essential tools for studying discrete-valued financial processes that are not adequately represented by continuous-valued approaches. This talk explores the theoretical foundations and key applications of count time series methods in finance, with particular emphasis on modelling transaction counts, credit defaults, extreme events, and other integer-valued financial indicators. We also discuss extensions of count-based modelling to settings where the underlying financial processes evolve in continuous time.