以Google Trends關鍵字搜尋量為基礎之台灣股市交易行為研究

The Study of TAIEX Trading Behavior based on Search Volume for Keywords on Google Trends

廖恩志(En-Chih Liao)

指導教授 ： 范敏玄

共同指導教授 ： 陳牧言

國立臺中科技大學/資訊與流通學院/資訊管理系碩士班/碩士(2014年)

https://doi.org/10.6826/NUTC.2014.00108

摘要

預測股市走勢一直是投資人所努力的方向，不論是基本分析或者是技術分析，所想要達到的目的無非就是了解股市的變化規則，以作為進出股市的依據。本研究以網際網路搜尋引擎的搜尋量Google Trends為分析對象，透過Google Trends的搜尋量與台灣加權股價指數研究分析之間的相關性。利用Google Trends所提供的關鍵字搜尋量，進行相關性檢定與單根檢定，再將所得到的關鍵字分別進行實驗一-機器學習與實驗二-搜尋趨勢中做分析，經過實證分析後，發現在實驗一中的類神經網路表現優於支援向量機與決策樹，進而挑選類神經網路作為與實驗二的搜尋趨勢方法做比較，透過報酬值計算的比較分析，發現搜尋趨勢的報酬值優於類神經網路的報酬值。因此，本論文發現以台灣50指數公司名稱作為搜尋關鍵字與台灣加權股價指數的漲跌存在關聯性。

關鍵字

Google Trends ； 台灣股市 ； 搜尋量 ； 類神經網路

並列摘要

Investors have always tried to predict the stock market trend. Whether it is fundamental analysis or technical analysis, what they want is nothing more than to understand the rule of changes in the stock market and use it as a basis to trade in the stock market. This study used the amount of Internet search on Google Trend and analyzed the correlation between the search volume on Google Trend and Taiwan Weighted Stock Index. The keyword search volume provided by Google Trend was used in the correlation test and the unit root test. Then the keywords obtained were analyzed in two experiments – first, machine learning, and second, search trend. After empirical analysis, it was found that neural network in experiment one performed better than support vector machine and decision trees. Therefore, neural network was selected to compare with the search trend in the second experiment. Through comparative analysis of calculation of return values, it was found that the return value in search trend is higher than that of the neural network. Therefore, this paper revealed that there was a correlation between using company names of Taiwan 50 Index as search keywords and the rise and fall of TAIEX index.

並列關鍵字

Google Trends ； TAIEX ； search volume ； Artificial Neural Network

參考文獻

林美珍. (2012). 行為財務學. 臺北市:華泰文化.

張茹敏. (2013). 人民幣與美國總體經濟之實證研究. 臺灣大學國際企

楊智凱. (2011). 台灣加權指數與上海股價指數關連之研究. 中興大學

Aouadi, A., Arouri, M., & Teulon, F. (2013). Investor attention and stock market activity: Evidence from France. Economic Modelling, 35, 674-681.

Choi, H., & Varian, H. (2012). Predicting the present with google trends.Economic Record, 88(s1), 2-9.