Are there gains from using information over the surface of implied volatilities?

We investigate the out‐of‐sample predictability of implied volatility using the information over the implied volatility surface. We show that implied volatility surface is useful for the out‐of‐sample forecast of implied volatility up to 1 week ahead. Trading strategies based on the predictability of implied volatility could generate significant risk‐adjusted gains after controlling for transaction costs. Significant results also depend on the way of modeling implied volatility surface. We then calibrate a two‐factor stochastic volatility option pricing model to implied volatility data. Results show that implied volatility is better explained by both long‐ and short‐term variance factors.