在選擇權交易操作中，最重要是選擇權價格之評定。但目前有關選擇權定價模式的相關研究，其結果皆與市場價格誤差甚大，主要導因於研究中所提出之假設前提與實務差異甚大；故本研究利用資料探勘技術，避免不符實際狀況之研究假設前提，建構準確率較高之定價模式。本研究分別採用歷史波動性與隱含波動性，結合類神經網路以建構選擇權定價模式，並以MAE與MSE作為與Black-Scholes模式之比較基礎。研究結果發現，在使用歷史波動性方面，由本研究類神經網路所建構之選擇權權利金定價模式，無論在MAE與MSE之績效，皆大幅優於Black-Scholes選擇權定價模式。而在使用隱含波動性方面，本研究建構之最佳模式預測績效，在價平與價外資料中，其MAE與MSE皆優於Black-Scholes模式；而在價內資料中， MSE優於Black-Scholes模式，但MAE比Black-Scholes模式稍差。本研究建構類神經網路選擇權定價模式之重要因素，除了包含Black-Scholes模式的6個因素外，在歷史波動性方面，加入成交量、未平倉量、開盤價、交易稅及股利；在隱含波動性方面，加入交易稅及股利。

The amount of premium is the most important factor in trading options. There were some related researches about option pricing model, however, the option prices that got from those models were much different from market practices. The major differences between model prices and market prices were due to wrong research Pre hypothesis. Therefore this paper will employ data mining techniques to construct an option pricing model to provide more ac-curate results. This research associated history volatility, and implied volatility with Neural Network to create the option pricing model. The results we found that the neural network model using history volatility is better than Black-Scholes model, and the neural network model that using implied volatility in at-the-money and out-of-the-money options are better than Black-Scholes model. The neural network model that using implied volatility in in-the-money options is better than Black-Scholes model in mean square error, but it is a little worse than Black-Scholes model in mean absolute error. The critical factors we found in the neural network option pricing model including the 6 factors of Black-Scholes model in addition to the volume and open interest, open price, dividend, and transaction tax in history volatility; the dividend, and transaction tax in implied volatility.