Monthly Closing Price Forecasting of Soybean Grain in Paraná Using Sarima Modeling With Intervention

Abstract

This article presents a procedure to find a suitable model for forecasting financial time series of agribusiness commodities considering the negotiation and circumstantial factors that influence them, especially seasonality and abrupt economic changes. The used data were the series of monthly closing prices of soybean in the state of Paraná between the months of February 1998 and January 2004. This series and its transformations were tested empirically in terms of ACF analysis (Auto-Correlation Function) until stationarized changes were found. For this, the application of seasonal adjustment was necessary due to the characteristic of the series and interventions (dummy variables) due to structural breaks occurred during the analyzed period. The seasonal adjustment was done through the X-12-ARIMA method, which implements the univariate ARIMA model (AutoRegressive Integrated Moving Average) and the multiplicative seasonal (SARIMA). The combined use of SARIMA model and of interventions got, in terms of the statistical analysis adopted, modeling the negotiation factors known to influence their oscillations and realize the forecast with 1, 2 and 3 steps ahead.