機器學習與人工智慧的技術能夠應於金融交易之決策，並獲得創新的交易策略，本研究則希望發掘文字探勘應用於金融交易之決策領域之可能。文字探勘將非結構化資料轉化為結構化資料以利使用者進行後續分析，具有將文字間的隱藏訊息轉化為數據的能力，本研究希望藉由採用新聞文本之分析來建構台灣股票市場之程式交易策略。在系統設計上，我們先運用文字轉向量方法將個股實施分類，再運用量化及資料分析方法將切割後的文本轉換為情緒溫度，最後再依據各則新聞的輿論溫度加總在動能策略之上決定做多或放空。為檢測該系統的可行性，我們選用台灣加權股價指數中佔比前十大之上市股票自2010年01月04日至2018年07月27日的日資料以及同段時間的鉅亨網台股新聞資料予以回測。實證發現，套用本研究所量化之新聞溫度計的策略會有比起未套用新聞溫度計的策略有較佳之報酬與風報比。

The technique of machine learning and artificial intelligence applies in the area of financial decisions, which helps us build innovative trading strategies. This research aims to discover ways how text mining tempts to gain application in trading. Text mining transforms the unstructured data into structured ones, which gains the ability to turn word into analyzable data. This research tries to analyze stock news to construct algorithm trading strategies in Taiwan stock market. In terms of system design, we divide the stocks into several groups by word2vec, then by text mining we can gain access to the mood of stock market. Based on it, constructing a beta strategy by determining whether the stock is bear or bull will be the work of stock thermometer. To test the feasibility of the research, the research back-tested the ten biggest stocks of the Taiwan stock market (2010/1/4-2018/7/27) and the stock news of Anue. Our findings illustrate that strategies using the news data tend to perform better and have higher profit than those who didn’t.

[1]　李春安, 羅進水, & 蘇永裕. (2006). 動能策略報酬, 投資人情緒與景氣循環之研究. 財務金融學刊, 14(2), 73-109.
[2]　Alostad, H., & Davulcu, H. (2015). Directional prediction of stock prices using breaking news on Twitter. In 2015 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT) (Vol. 1, pp. 523-530). IEEE.
[3]　Chan, L. K., Jegadeesh, N., & Lakonishok, J. (1996). Momentum strategies. The Journal of Finance, 51(5), 1681-1713.
[4]　Chan, K., Hameed, A., & Tong, W. (2000). Profitability of momentum strategies in the international equity markets. Journal of Financial and Quantitative analysis, 35(2), 153-172.
[5]　Gidofalvi, G., & Elkan, C. (2001). Using news articles to predict stock price movements. Research Report. Department of Computer Science and Engineering, University of California, San Diego.
[6]　Jegadeesh, N., & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. The Journal of Finance, 48(1), 65-91.
[7]　Mao, Y., Wei, W., & Wang, B. (2013, August). Twitter volume spikes: analysis and application in stock trading. In Proceedings of the 7th Workshop on Social Network Mining and Analysis (p. 4).
[8]　Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781.
[9]　Schumaker, R., & Chen, H. (2006). Textual analysis of stock market prediction using financial news articles. Americas Conference on Information Systems (AMCIS) 2006 Proceedings, 185.
[10]　Schumaker, R. P., & Chen, H. (2009). A quantitative stock prediction system based on financial news. Information Processing & Management, 45(5), 571-583.
[11]　網際網路資料，Word2vec是如何得到詞向量的？民106年12月25日，取自：https://www.zhihu.com/question/44832436
[12]　網際網路資料，詞嵌入(Word embeddings)的基本概念。民107年8月20日。取自：https://www.kesci.com/home/project/5b7a359e31902f000f55152f
[13]　網際網路資料，Jieba分詞的原理。民106年12月03日，取自：https://www.itread01.com/content/1512314575.html