statsmodels.tsa.ardl.ARDLResults.fpe¶ ARDLResults.fpe¶ Final prediction error using Lütkepohl’s definition. \(((nobs+df_{model})/(nobs-df_{model})) \sigma^2\)