Dynamic functional time-series forecasts of foreign exchange implied volatility surfaces

TL;DR

This paper develops dynamic functional time-series models to forecast foreign exchange implied volatility surfaces, demonstrating improved accuracy and potential economic benefits over traditional methods.

Contribution

It introduces dynamic functional principal component analysis for volatility surface forecasting, showing significant accuracy improvements over existing approaches.

Findings

Dynamic models outperform static ones in forecast accuracy.

Significant improvements for EUR-USD, EUR-GBP, and EUR-JPY surfaces.

Potential economic gains demonstrated through a stylized trading strategy.

Abstract

This paper presents static and dynamic versions of univariate, multivariate, and multilevel functional time-series methods to forecast implied volatility surfaces in foreign exchange markets. We find that dynamic functional principal component analysis generally improves out-of-sample forecast accuracy. More specifically, the dynamic univariate functional time-series method shows the greatest improvement. Our models lead to multiple instances of statistically significant improvements in forecast accuracy for daily EUR-USD, EUR-GBP, and EUR-JPY implied volatility surfaces across various maturities, when benchmarked against established methods. A stylised trading strategy is also employed to demonstrate the potential economic benefits of our proposed approach.