With the rapid development of the global economy in the last two decades have come intensified price fluctuations in the metals markets due to uneven product distribution and inadequate productivity. Risk management techniques such as Value at Risk (VaR) have been demanded and have increasingly become the basis of planning for the nonferrous metal industry. As the traditional ex-post approaches to VaR estimates, such as the ARMA-GARCH model, leave little room for further performance improvement, this paper proposes the ex-ante based approach to VaR estimates and introduces the wavelet theory to strike the balance between the needs of data characteristics categorization and model calibrations. Empirical studies in four nonferrous metals markets show that WDVaR can significantly improve performance, compared to the existing ARMA-GARCH models, besides facilitating greater flexibility in tuning models for a specific market under investigation.