Uso del Modelo Autorregresivo de Duración Condicional para predecir la caída del dólar en el mercado cambiario colombiano

Abstract

The fundamental objective of the Autoregressive Conditional Duration (acd) model is the modeling of time series with non-equidistant periods. Given the leptokurtic nature of trm returns and the durations behavior associated with them, a Rayleigh Distribution with transmutation is used, which allows approaching in an adjusted form to a heavy tail distribution, coherent with this stylized fact in returns. It is concluded that the time elapsed between dollar falls, on average, is between 3 and 6 days.