本研究之目的為檢視台灣股票市場中投資人之日內群聚現象之特性，並分析群聚對未來股票報酬影響。投資策略方面，觀察加入投資人群聚行為之群聚動能策略對未來股價報酬是否具有預測性。本研究利用逐筆委託檔與成交檔資料，採用Patterson and Sharma(2005)所提出的Runs test檢定日內群聚現象；交易策略方面採用Jegadeesh and Titman(1993)和Kang(2005)提出的日內動能策略建構方式，並將市場分為向上趨勢和向下趨勢，探討投資人在不同市場情況下群聚特性為何。 實際交易策略方面，利用一般投資者可取得之成交檔資料，採行Finucane(2000)方法判斷交易資料為買方趨動或是賣方趨動，衡量出投資者群聚行為後，再加入價格動能策略中，驗證交易策略的可行性。 實證發現，投資人在一般情況下，日內各區間群聚行為皆不顯著；在市場為向上趨勢中，投資人日內群聚行為皆為顯著；在市場為向下趨勢時，投資人在日內各區間群聚行為皆不具顯著性。交易策略方面，任何市場情況下，在持有期J=30、60、120分鐘下，價格動能策略和群聚動能策略，具有持續動能現象，比較加入群聚前後的策略報酬，群聚動能策略投資報酬率大於未加入群聚因子前的價格動能策略投資報酬率，且在前期具有顯著正報酬差異。實際交易策略中，成交檔顯示資訊和委託檔顯示資訊呈現一致，表示在現實中，投資者可利用成交檔資訊得到和委託檔一致之資訊。 The purpose of the research is to observe the investors intraday herding behaviors in Taiwan stock market and to analyze the influence on stock price in the future. On the aspect of investment strategies, we observe that if it has been anticipation in stock price by considering the herding factor in the future. We employ the intraday trade data and the measure of herding by Patterson and Sharma(2005) ─ Runs test, and on the aspect of investment strategies, we follow the intraday momentum strategies of Jegadeesh and Titman(1993) and Kang(2005), and to find the difference of the herding behavior in the different market. In realities of investment strategies, we employ the strike order which investor can be possess easily. We employ the method of Finucane(2000) to determine each strike data which is belonging to buy-initiated or sell-initiated. After measuring the herding value, we employ the herding value into the price momentum strategies to verify the executable possible in real. In general situation, the result shows that herding isn’t significance in intraday , but it is significance in the bull market . In the aspect of investment strategies, it is momentum continuous in the holding period of 30, 60 and 120 minutes in price momentum strategies and herding-momentum strategies in any market situation. The difference portfolio return between the price momentum strategies and the herding-momentum strategies isn’t significance. In real investment strategies, the information is the same of order data and strike data, and it shows that investors can observe the information like order data showing by employing the strike data.
