Forecasting option smile dynamics

Practitioners have long tried to exploit the predictability of the option implied volatility smile. Motivated by the recent developments in the literature focusing on market-based option pricing arguments, this paper proposes the introduction of trading volume into a vector autoregressive (VAR) model to improve forecasts of the smile dynamics. The augmented VAR-volume model produces quality forecasts of the smile surface and explains its dynamic changes over time relatively well. Our results suggest that the incorporation of trading volume leads to it outperforming other alternative forecast approaches, as well as being robust to a variety of perturbations of the data and offers scope for investors to more accurately predict option implied volatility in the future.