Is It Worth Applying Intraday Value-at-Risk Models in the Financial Industry? Evidence from the EURTRY FX Market

This paper examines the volatile EURTRY FX market for the period 2005–2022. We follow the guidelines on Value-at-Risk (VaR) set by the law and we test popular and widely applied models, namely Delta Normal VaR, Exponential Weighted Moving Average VaR (EWMA VaR), and Historical VaR (HVaR), which use close to close prices, versus more challenging models that require intraday data (Garman-Klass and Parkinson). The results show that HVaR is the best of the conventional close-to-close models, but the Garman-Klass (GK) model is considered the best of all the examined models when we adopt the legal standards in the evaluation process. Moreover, we present qualitative information, which the legislation does not examine, such as that GK estimations are more accurate than HVaR estimations without being overly conservative, and that when GK presents overshootings, it falls closer to reality than the HVaR. Thus, this paper may be useful to risk managers who want to compare the GK and HVaR models in terms of VaR accuracy in an effort to improve estimations and decide whether the increased complexity that the GK model entails, as well as the increased volume of data it requires, is worth it in the end.