使用時間序列用於動態避險之反饋效應實證研究

Abstract

本論文採用發行於2009.09.22 至 2010.12.31期間，連接於六檔上市標的股(鴻海、國巨、群光、正崴、洋華、大聯大)的權證資料，觀察這些大量發行而繫於同一標的的權證，它們的希臘字母(Gamma)在隨著標的每日起伏的波動下，交易員因應的動態避險行為，是否對標的股本身在股價或成交量上，有著可觀察的反饋效應。 實證結果顯示，不論是在價或量方面，被解釋變數在三日移動平均振幅的擬合建模中，相對於隔日振幅、或五日均值振幅，具有較強的解釋性，並顯示了充分的一致性，而在建模與預測兩段區間中，以兩種波動率(隱含波動率/歷史波動率)估出的敏感性係數(Gamma)，乃至理論避險量(Pseudo-Gamma Potential)，他們的結構透過MSE與MAE的一致性顯現了出來，這一點與先前所提：基本交易量與價被發生在三日移動平均相互輝映，顯示了反饋效應在動態避險的解釋下，於內部結構的確有存在的空間。

In this paper we examine the interaction between hedge and feedback effect with transfer function models. Six underlying shares are used to investigate the explanatory power by their derivatives issued during 2009.09.22 and 2010.12.31. We hire two volatilities in contrast. The final shoot from the head of the trader would nearly root from the two, when he tries to trace with the up and down and eventually are forced to pull the trigger to the realized equilibrium state. Feedback effect could be discovered when we compare the models among the scenarios. Fundamental trades in the study are consistent in 3-day moving average no matter in the price or in the amount of trading in the markets. And hedge trade could be observed to shoot in one day or two.