The value-at-risk measurement (VaR) is an important indicator when a financial institution wishes to evaluate its exposure to market risks. Referring to I. Khindanova, S. Rachev and E. Schwartz (2001), the traditional Gaussian estimation of VaR generally overlooks the heavy-tailed properties exhibited financial data series, thereby incurring unexpected losses under extreme circumstances. This paper follows I. Khindanova, S. Rachev and E. Schwartz (2001) to show that heavy-tailed properties exist for Taiwan’s financial market as well and aims to prove that this issue can be effectively solved by fitting the financial data to an α-stable distribution.