Prédiction de la volatilité future dans le marché des devises à l'aide de la volatilité implicite

Abstract

This research examines the performance of implied volatility when used in the objective of forecasting future volatility in the foreign exchange market and more specifically for the two most traded pairs, EUR/USD and USD/JPY. An analysis of the performance of the EUVIX and JYVIX indexes is conducted for a period of 10 years, from 2007 to 2017. It is discovered that during this period, the last available implied volatility value is optimal and seems to contain the maximum amount of useful information about future volatility. The data concerning implied volatility is integrated in a GARCH-type autoregressive model. The conditional variances obtained from the resulting IVGARCH(1,1,1) model are compared with the conditional variances of a standard GARCH(1,1) model. The in-sample and out-of-sample results tell us that the addition of information concerning implied volatility significantly improves forecasting performance for the two pairs studied. The average forecasting error is reduced by approximately 6% when using the IVGARCH(1,1,1) model versus the GARCH(1,1). But most importantly, we discover that the IVGARCH(1,1,1) model allows us to anticipate periods of high volatility on several occasions. A method of application and the performance of implied volatility when using an original active investment system are also presented. The signals generated in part by the conditional variances of the model that is integrating implied volatility allows us to obtain profits that are significantly superiors compared to the profits obtained by a model that does not integrate it.

Date	10 Jan 2019
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Edmond T. Miresco (Supervisor)