The purpose of the manuscript is to outline and demonstrate problems with the use of the HP filter, and to propose an alternative strategy for inferring cyclical behavior from a time series that features seasonal, trend, cyclical and noise components. The main innovation of the alternative strategy involves augmenting the series in question with forecasts and backcasts obtained from an ARIMA model, and then applying the HP filter to the augmented series. Comparisons presented in the paper using artificial and actual data demonstrate the superiority of the alternative strategy.