Testing the informativeness of non-price variables with MIDAS touch

Abstract

We examine whether open interest and volume-based price predictors are informative in predicting the underlying assets’ price on options’ expiry. We apply MIDAS estimation procedure on the data extracted from NSE and CMIE-Prowess database for the period between 2011 and 2016. The results of MIDAS estimation uphold the informativeness of non-price variables for 10 stocks and Nifty50 index. In-sample fit verification highlights that MIDAS estimate fits better than time aggregation and distributed lag estimates to actual price. The out-of-sample forecast evaluation coupled with the Wilcoxon test further confirms the relatively better performance of our model over the other two models.

Keywords

• Derivatives
• Forecasting
• Informational content
• MIDAS
• Non-price
• Open interest