Given the success of the previous three M-Competitions, the considerable number of participants attracted and the significant contributions made fundamentally changing the field of forecasting, we would expect similar achievements from the M4.
More specifically, we would expect to attract more than 100 participants from business schools and finance, computer science, and engineering departments of universities, financial analysts, departments of banks and financial institutions as well as individuals. The attraction to participate, beyond winning a prize, will be the experience in competing in an international challenge and the publicity to be gained from one’s accurate forecasts.
The Competition will provide us with information to be able to:
- Identify the similarities as well as the differences between the various forecasting methods, including ML ones, and their accuracy versus traditional statistical ones. A recent study discussing the accuracy of Statistical and Machine Learning methods, explaining why the performance of the latter is below that of the former and proposing some possible ways forward, can be found here .
- Compare the forecasts with those of random walks and similar benchmarks.
- Determine if the forecasts of some methods are consistently superior to others, in a way it is not by chance, and if this is the case find out why.
- Classify forecasters as optimists or pessimists and quantify the extent and consistency of their bias.
- Determine the accuracy of confidence intervals forecasts of the various methods.
- Evaluate the efficiency of the computed prediction intervals.
Preliminary results of the Competition will be announced at the 38th International Symposium on Forecasting – Boulder, Colorado, 17-20 June 2018. The final results of the Competition, including the winners and a leader-board of the competitors, will be announced latter.
The detailed results of the Competition will be published in a Special Issue of International Journal of Forecasting (IJF), coordinated by Fotios Petropoulos. A small number of papers describing the methods that performed best in the Competition will be invited, along with others discussing its results. All papers will follow the IJF’s refereeing process.