Análisis del comportamiento del precio de una acción en el mercado de valores argentino utilizando un modelo ARIMA

Abstract

In the present investigation, the statistical estimation was presented through ARIMA models – autoregressive integrated moving average (p, d, q) – of the behavior of the price of a financial asset in the Argentine capital market. The development involved relying on the Box and Jenkins modeling methodology, in which the stages of identification of the intervening variables, their estimation and, finally, the application of the control and diagnosis of the model used were followed. The selection of this type of model was found in its potential for forecasting in the financial field. At the same time, it allowed to carry out an order in the determination of the price lags that influence the present price of financial assets. Following this methodology allowed us to estimate two possible ARIMA (2,1,2) and ARIMA (4,1,2) configurations. At a 95% level, it was observed that the model that meets the characteristics of constant variance - validated by means of the Engle statistic - and well-behaved partial autocorrelation functions, is the ARIMA model (2,1,2) including the constant. This being the model with the highest prediction power because it has the lowest mean square error. The estimation of this model allows us to affirm that it is the one that best represents observed data and allows us to affirm that the price of the shares of the GGAL group, in the period from 02/01/2016 to 01/01/2018, can be explained with robustness to through an ARIMA model (p, d, q).