We utilise novel functional time series (FTS) techniques to characterise and forecast implied
volatility in foreign exchange markets. In particular, we examine the daily implied volatility
curves of FX options, namely; EUR-USD, EUR-GBP, and EUR-JPY. The FTS model is shown
to produce both realistic and plausible implied volatility shapes that closely match empirical
data during the volatile 2006-2013 period. Furthermore, the FTS model significantly outperforms implied volatility forecasts produced by traditionally employed parametric models. The
evaluation is performed under both in-sample and out-of-sample testing frameworks with our
findings shown to be robust across various currencies, moneyness segments, contract maturities, forecasting horizons, and out-of-sample window lengths. The economic significance of the
results is highlighted through the implementation of a simple trading strategy.
Item Type:
Article (Published)
Refereed:
Yes
Uncontrolled Keywords:
Exchange rates; implied volatility; forecasting; functional data analysis; functional time series