Why does forecast combination work so well?

Abstract Forecast combinations were big winners in the M4 competition. This note reflects on and analyzes the reasons for the success of forecast combination. We illustrate graphically how and in what cases forecast combinations produce good results. We also study the effects of forecast combination on the bias and the variance of the forecast.

[1]  F. Petropoulos,et al.  Improving forecasting by estimating time series structural components across multiple frequencies , 2014 .

[2]  Amir F. Atiya,et al.  A Bias and Variance Analysis for Multistep-Ahead Time Series Forecasting , 2016, IEEE Transactions on Neural Networks and Learning Systems.

[3]  Spyros Makridakis,et al.  The M3-Competition: results, conclusions and implications , 2000 .

[4]  R. Clemen.  Combining forecasts: A review and annotated bibliography , 1989 .

[5]  Fotios Petropoulos,et al.  Another look at forecast selection and combination: Evidence from forecast pooling , 2019, International Journal of Production Economics.

[6]  Michael P. Clements,et al.  Pooling of Forecasts , 2004 .

[7]  Derek W. Bunn,et al.  Review of guidelines for the use of combined forecasts , 2000, Eur. J. Oper. Res..

[8]  Robert L. Winkler,et al.  The accuracy of extrapolation (time series) methods: Results of a forecasting competition , 1982 .

[9]  Waleed A. Yousef,et al.  Learning algorithms may perform worse with increasing training set size: Algorithm-data incompatibility , 2014, Comput. Stat. Data Anal..

[10]  Amir F. Atiya,et al.  Combination of long term and short term forecasts, with application to tourism demand forecasting , 2011 .

[11]  Mary E. Thomson,et al.  Combining forecasts: Performance and coherence , 2019, International Journal of Forecasting.

[12]  Evangelos Spiliotis,et al.  The M4 Competition: Results, findings, conclusion and way forward , 2018, International Journal of Forecasting.

[13]  Fotios Petropoulos,et al.  Forecasting with multivariate temporal aggregation: the case of promotional modelling , 2016 .