本研究提出了一個利用時間序列動態行為的群集分析法，可用來估算期貨避險交易決策所需之最適避險比例，作為發展避險決策輔助系統雲端服務之基礎。市場波動的動態行為可以使用變異數、共變數、期貨和現貨間的價差和這些變數的一階、二階導數來衡量。具有相同行為模式的日資料，則可以利用擴展階層式自我組織映射圖(growing hierarchical self-organizing map，GHSOM)來進行分群，這些具有相似行為的日資料將依照相似程度被分配到階層式的子集合中，配合集群內重新取樣之演算法，來構成以傳統最小變異數法估算避險比例之樣本資料。我們使用了台灣加權指數之期貨與現貨來進行為實證，結果顯示本研究所提出的方法可以顯著的改善避險效能。
In this study, a novel procedure of time series dynamic behaviors clustering is proposed to improve the accuracy of minimum -variance optimal hedge ratio (OHR) estimation for future hedging. The dynamic behaviors of market fluctuation are extracted by measurement of variances, covariance, price spread, and their first and second differences. The behaviors with similar patterns are clustered using a growing hierarchical self-organizing map (GHSOM). The observations for OHR estimation are collected based on the hierarchical cluster structure and processed by within-cluster resampling. The spots and futures of the Taiwan Weighted Index (TWI) are adopted to demonstrate that the futures hedge effectiveness can be significantly improved.