The present investigation was carried out to determine which are the best models to forecast the main indicators of the financial system of the Loreto region, which are direct loans and deposits. For this, the most appropriate time series econometric techniques reflected in the Box-Jenkins methodology were used, from which it was possible to corroborate that the model that performs a better fit to the data for both loans and deposits is the ARIMA(0,1,1), since this is the one that generates the smallest value of the Schwarz information criterion; however, this model is not the one that performs the best job in terms of forecast execution, since the values it yields when calculating the Root Mean Square Error (RMSE) for all models are not the lowest. For this reason, the models that have the best level of efficiency in terms of projections are the ARIMA(2,1,1) for direct loans and the ARI...