Working Paper Series

Multivariate lasso-based forecast combinations for stock market volatility

Volatility forecasts indicate future risk and are key inputs in financial analysis. We fore-cast the realized variance, an observable measure of volatility, of several major international stock ma ...

Author(s) :

Jeroen Rombouts

Ines Wilms

Abstract :

Volatility forecasts indicate future risk and are key inputs in financial analysis. We fore-cast the realized variance, an observable measure of volatility, of several major international stock market indices and account for the differing predictive information present in jump, continuous, and option-implied variance components. We allow for volatility spillovers of different stock markets by using a multivariate modeling approach. To obtain the forecasts, we use an ordered lasso-based forecast approach. A simple forecast combination of the uni-variate and multivariate model outperforms the benchmark model for all considered forecast horizons and stock markets.

Keywords: Forecast combination, Ordered Lasso, Realized variance, Volatility forecasting.

Working Paper - Multivariate lasso-based forecast combinations for stock market ...
(-1.00 B)
Date : 21/09/2018
Working Paper Number: WP-18-003

See Also

The EDHEC professor Gianpaolo Parise invited to join the CEPR
News
- 07-01-2020
Gianpaolo Parise, Associate Professor in Finance at EDHEC Business School have been...
Can financial theory make sense of the factor zoo?
News
- 20-11-2019
EDHEC PhD in Finance Newsletter - November 2019  Editorial signed by Laurent Calvet,...
Appointment and Grant for EDHEC Professor Raman Uppal
News
- 19-11-2019
At the end of September 2019, EQDerivatives launched The Journal of Systematic...
50th thesis defence for the EDHEC PhD in Finance Programme
News
- 11-11-2019
On 8 November 2019, on the EDHEC London campus, the doctoral thesis “Two Essays on...