Due to the rapid grow up of transaction volume of derivatives in the financial market, the Black-Scholes options pricing model (BSM) is played an important role recently and widely applied in various options contract. However, this theoretical model limited by the influences of many unexpected real world phenomena caused due to its six unreasonable assumptions, which often make the miss-pricing result because of the difference of market convention in practical. If we were to soundly take these phenomena into account, the pricing error could be reduced. In this paper, we provide a signal-decomposition oriented framework via wavelet analysis to improve the precision of BSM using integrated wavelet-based feature extraction with support vector machines (WSVMs). We investigate the techniques for transforming the noticeable signal from the mark to market price into estimating the option fair value and hence gain better precision estimation then pure support vector machine, in which has recently been introduced as a new technique for solving a variety of time series forecasting. Compare with the original GARCH method, adaptive neural-based fuzzy inference system (ANFIS) and pure SVMs, the performance of the presented method show the best. Using evidence from the warrants market in Taiwan, it supports our claims. This paper helps to provide an alternative way to refine the options valuation.