Autoregressive moving average approaches for estimating continuous non-negative time series

This study performs a comparative analysis of autoregressive moving average models for non-negative time series within a financial context, aiming to identify the model that offers the best fit and forecasting accuracy. The analysis is applied to two financial datasets: the stock trading volume of Banco Bradesco and the insurance volume of Porto Seguro. Four models are fitted in the process: Autoregressive Moving Average (ARMA), Rayleigh Autoregressive Moving Average (RARMA), Generalized Autoregressive Moving Average (GARMA), and Generalized Linear Autoregressive Moving Average (GLARMA). Model performance is evaluated through fit comparison metrics and forecasting accuracy measures to determine the most effective model.