A scientometrics review on combining forecasts in financial markets

Combining forecasts has been widely applied as one of the approaches to increase the predictive power over the past few decades. With the promising results, it has seen an explosion in the number of articles on combining forecasts, of which review of the literature become resources taxing and causes researcher's cognitive burden. This study applies a scientometrics analysis to demonstrate the intellectual maps about the networking of research work through co-citation analysis and collaboration analysis. A total of 184 publications in 1965 to 2020 related to combining forecast in financial markets were obtained from WoS database and were analysed. The keyword analysis identified five clusters of research: 1) model performance; 2) time series model error and volatility; 3) time series and tests; 4) forecast evaluation; 5) model efficiency in the combining research area and the evolution of combining forecasts over the study periods were also discussed. The visual mapping results provide an overview of combining forecast development which helps other researchers in formulating their research directions.