Volatility-adapted network model for hedging equity index options

Hedging index options reduce the risk from equity price changing using strategies as purchase or sale options, futures, or other derivatives. The complexity and visibility can hinder interpretation and transparency to making hard for stakeholders to fully comprehend or accept the decision-making process. This research uses non-linear neural networks to provide a novel hedging technique for stock index options called volatility-adapted network (VA Net). Numerous hedging strategies are applied and tested on the same datasets, and extensive experiments are carried out on the S&P 500 option market in the United States and CSI300 options market in China. In contrast with previous studies on neural network hedging, the VA Net presents the gate activation system, which could automatically choose the best input parameters to maximize their utilization for an improved hedging ratio. The results of the comparisons show that the VA Net performs well in determining the relative relevance of input characteristics and producing a good hedging ratio.