本研究主要透過文字探勘技術與機器學習的模型，將情緒詞庫作為模型的變數之一，藉此強化模型的分析能力，並採用誤分類率作為模型準確率的判斷依據。
將三種不同的情緒詞庫落後1至5期的變數帶入模型中，分別採用機器學習方法中的梯度提升決策樹、羅吉斯迴歸、類神經網路及支援向量機，可發現其誤分類率的穩定度與準確度都不盡相同，為了選取一個較為穩定的變數，將各方法所算出的誤分類率再計算其變異數，最後採用誤分類率及變異數相對小的情緒指數Opinion Lexicon落後5期之類神經網路模型作為預測模型，進行投資策略。
本研究模型的目標變數採用道瓊貴金屬指數，因其與黃金價格具高度相關性，而在進行投資策略時採用COMEX黃金期貨2017年12月交割的期貨價格，並分別以買進持有及放空持有的策略作為基準報酬率與預測模型下的投資策略結果進行比較，實證結果顯示Opinion Lexicon落後5期情緒指數之類神經網路模型進行投資策略，買進策略下可維持一定程度的報酬率，而放空策略下可避免投資損失擴大。

In this study, the sentiment lexicon is taken as one of the variables of the model mainly through text mining technology and the machine learning model to reinforce the analytic ability of the model, and the misclassification rate is used as the basis for the accuracy of the model.

The three different sentiment lexicon lagged from 1 to 5 variables into the model, respectively. Using machine learning methods about Gradient Boosting Decision Tree , Decision Tree, Logistic Regression, Neural Networks and Support Vector Machines, we can find that misclassification rate stability and accuracy are not the same. In order to select a more stable variable, each method calculates the misclassification rate and variance, and finally take the sentiment index Opinion Lexicon lagged 5,using the neural network model as the forecast model.

The target variable for this research model uses the Dow Jones Precious Metals Index, which is highly correlated with the gold price .This study chose the futures price delivered by COMEX Gold Futures in December 2017 when making the investment strategy, and buys and sells the futures respectively .The holding strategy is compared with the investment strategy results under the forecasting model. Empirical results show that the Opinion Lexicon lagged 5 using the neural network model. When at the long strategy, it can maintain a certain rate of return; while under the short strategy, investment loss can be avoided.

黃婉婷，投資人情緒應用於原油、金及美元指數期貨動態關聯之研究，東海大學碩士論文，2011。
呂瑞秋、陳淑華，黃金價格與股價指數、石油價格及波動率指標關係之研究」，台灣銀行季刊第六十三卷第二期，2012。
張秀瑜，糧食、石油與黃金的期貨價格之關聯性分析，國立交通大學管理學院(經營管理學程)碩士論文，2012。
陳世榮，社會科學研究中的文字探勘應用：以文意為基礎的文件分類及其問題，《人文及社會科學集刊》第二十七卷第四期，2015。
謝昌儒，機構與個別投資人情緒指標對黃金價格之影響，國立中央大學財務金融學系碩士論文，2016。
陳怡安，文本探勘技術之應用-發掘新聞情緒並探討其和恐慌指數之關聯性，國立成功大學財務金融研究所在職專班碩士論文，2017。
Baur ,Dirk G. and McDermott ,Thomas K., “ Is Gold a Safe Haven? International Evidence,” Journal of Banking & Finance , pp. 1886-1898, 2009.
Baur ,Dirk G. and Lucey ,Brian M , “Is Gold a Hedge or a Safe Haven? An
Analysis of Stocks, Bonds and Gold,” The Financial Review ,pp.217–229, 2010.
Bollen , Johan and Huina Mao and Xiaojun Zeng, “Twitter mood predicts the stock market,” Journal of Computational Science ,pp. 1–8, 2011.
COMEX Gold Futures data (2016/12/27~2017/5/22) , https://in.investing.com/commodities/gold.,2017.
Driel ,Marc A van, Jorn Bruggeman, Gert Vriend, Han G Brunner and Jack A M Leunissen, “ A text-mining analysis of the human phenome,” European Journal of Human Genetics, pp535-542, 2006.
Fama ,Eugene F, “Efficient Capital Markets: A Review of Theory and Empirical Work,” The Journal of Finance, pp. 383-417, 1970.
Hu , Minqing and Bing Liu, “Mining and Summarizing Customer Reviews, ” Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining,pp.168-177,2004.
Han, B., “Investor Sentiment and Option Prices, ” The Review of Financial Studies, Volume 21, Issue 1, pp. 387–414, 2008.
Hardle, W.K. and Lee, Y.J. and Schafer D. and Yeh Y. R, “ Variable Selection and Oversampling in the Use of Smooth Support Vector Machines for Predicting the Default Risk of Companies,” Journal of Forecasting J. Forecast, pp. 512–534 ,2009.
Hood ,Matthew and Farooq Malik, “Is gold the best hedge and a safe haven under changing stock market volatility?” Review of Financial Economics ,pp.47–52, 2013.
Liu ,Bing and Hu, Minqing and Cheng, Junsheng, “ Opinion Observer: Analyzing and Comparing Opinions on the Web, “ Proceeding of the 14th international conference on World Wide Web, pp.342-351,2005.
Loughran ,Tim and Mcdonald ,Bill, “ When Is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks,” The Journal of Finance. Vol. LXVI, NO. 1, 2011.
Lee A. Smales, “News sentiment in the gold futures market,” Journal of Banking & Finance ,pp.275–286 , 2014.
Malkiel,Burton Gordon “ A Ramdom Wak Down Wall Street,” W.W.Norton&Company Inc , pp. 4-6 ,1973,
Mittermayer ,Marc-Andre, “ Forecasting Intraday Stock Price Trends with Text Mining Techniques,” Proceedings of the 37th Hawaii International Conference on System Sciences, 2004.
Narayan, Paresh Kumar and Narayan, Seema and Zheng,Xinwei, “Gold and oil futures markets: Are markets efficient?“ Applied Energy, pp.3299-3303,2010.
Rohan Christie–David, Mukesh Chaudhry, and Timothy W. Koch, “Do Macroeconomics News Releases Affect Gold and Silver Prices?” Journal of Economics and Business ,pp.405–421 , 2000.
Schumaker, Robert P. and Chen , Hsinchun, “Textual Analysis of Stock Market Prediction Using Breaking Financial News: The AZFinText System,” ACM Transactions on Information Systems, Vol. 27, No. 2, Article 12, 2009.
Samanta ,Subarna K. and Zadeh ,Ali H. M.,“Co-Movements of Oil, Gold, the US Dollar, and Stocks,” Modern Economy, pp. 111-117,2011.
Wilson ,Theresa , Janyce Wiebe and Paul Hoffmann, “Recognizing Contextual Polarity in Phrase-Level Sentiment Analysis,” Proceeding of the Conference on Human Language Technology and Empirical Methods in Natural Language, pp.347-354, 2005.
Yoon, Byungun and Park,Yongtae, “ A text-mining-based patent network: Analytical tool for high-technology trend,” The Journal of High Technology Management Research, pp. 37-50,2004.
Zhang , Yue-Jun and Wei,Yi-Ming, “ The crude oil market and the gold market: Evidence for cointegration, causality and price discovery,” Resources Policy, pp.168-177,2009.
Zhang, J. ,Chen C. Y. , Hardle , W.K and Bommes, E, “Distillation of News Flow Into Analysis of Stock Reactions,” Journal of Business & Economic Statistics, pp.547-563 , 2016.