使用VPIN預測台灣市場日內權益波動度

Abstract

本文藉由選取台灣90家電子類股的上市公司做為樣本，並用2016年至2017年每天之日內交易資料，使用VPIN（Volume-Synchronized Probability of Informed Trading）當作指標來預測日內未來的波動度，並判斷VPIN的有用與否，以及將公司依特徵做分類，使用多項羅吉斯迴歸來檢驗擁有何種性質的公司VPIN越有預測短期未來波動度的效果；最後發現成交量越多、P/B ratio越高、公司對大盤超額波動的程度越大以及外資持股比率越高的公司，VPIN越有能力預測短期未來的波動度。

This thesis employs the intraday trading data of 90 exchange listed companies in the electronics sector on Taiwan Stock Exchange from 2016 to 2017 as the sample to examine whether VPIN (Volume-Synchronized Probability of Informed Trading) can be a useful indicator to forecast the intraday future volatility. In addition, this thesis examine what kind of companies can make the VPIN more likely to be an effective leading indicator to forecast intraday future volatility by conducting a multinomial logistic regression of the usefulness of the VPIN over the attributes of examined companies. Finally, this thesis finds that the VPIN performs better in predicting intraday future volatility for those companies with higher trading volume, P/B ratio, excess volatility and institutional investments.