As readers of this blog might have realized, my main research interest - applications/projects - is tightly related to forecasting `tendencies': I'm haunting `turning points' (of trends, cycles). Typically mean-square criteria are not well suited for this kind of application (let alone mean-square one-step ahead...). As a consequence, a vaste majority of entries on this blog are devoted to `customized' (turning-point) criteria. However, data providers are more picky about `small revisions' i.e. good mean-square performances. A SEF-Blog reader (and accessory I-MDFA user) asked about `mean-square stuff'. Here is some.
"A User Request: "I would like to see how (if any) DFA could help improve mean-square performance (not the trend/cycle stuff)"" vollständig lesen
