CMStatistics 2021: Start Registration
View Submission - CFE
A1715
Title: Forecasting VVIX using forecast combinations and LASSO Authors:  Yushuang Jiang - Peking University (United Kingdom) [presenting]
Emese Lazar - University of Reading (United Kingdom)
Abstract: Motivated by the success of forecast combinations and the LASSO-type shrinkage methods, we attempt to answer the following question: is there an optimal VVIX forecasting method? If yes, then is this based on forecast combinations or LASSO? We show that forecast combinations perform best. We compare the forecasting performance of three individual models, eight combining methods and two LASSO-type models out-of-sample. The results show that the simple median combining method delivers the lowest forecasting errors across the years. In addition, we discuss the model selection results of two shrinkage methods. Interestingly, instead of daily changes in the VVIX, the changes in monthly VVIX are key to predicting the VVIX.