Abstract
In this article, we propose a wavelet-based approach to accommodate the stylized facts and complex structure of financial data, caused by frequent and abrupt changes of markets and noises. Specifically, we show how the combination of both continuous and discrete wavelet transforms with traditional financial models helps improve portfolio’s market risk assessment. In the empirical stage, three wavelet-based models (wavelet-EGARCH with dynamic conditional correlations, wavelet-copula, and wavelet-extreme value) are considered and applied to crude oil price and US dollar exchange rate data. Our findings show that the wavelet-based approach provides an effective and powerful tool for detecting extreme moments and improving the accuracy of VaR and Expected Shortfall estimates of oil–exchange rate portfolios after noise is removed from the original data.
•Financial models (EVT, copulas, DCC-EGARCH) are combined with wavelet transforms.•These wavelet-based models help accommodate the complex structure of financial data.•They are then applied to both denoised and original data of oil and exchange rates.•We find that wavelet transforms enable the detection of extreme events in return data.•The proposed models help improve the accuracy of portfolio’s market risk.