( 您好!臺灣時間:2021/05/07 03:01
字體大小: 字級放大   字級縮小   預設字形  


論文名稱(外文):A Study on Stock Price Forecasting by Deep Learning
外文關鍵詞:deep learningneural networkstock price forecastingsocial network.
  • 被引用被引用:6
  • 點閱點閱:3360
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:54
  • 收藏至我的研究室書目清單書目收藏:1
預測股價的漲、跌趨勢,長久以來一直都是學者與財經專家所感興趣的主題。但是大量的雜訊與非線性的資料,讓評估過程產生許多不確定性,因此我們希望透過近年來盛行於人工智慧領域的深度學習,尋找出有效的方法來解決此問題。本研究收集並萃取股票市場交易資料,做為深度學習所需的訓練資料集,藉此建立一套以技術分析指標為基礎的深度學習模型,並使用近年來因社群網路崛起而蓬勃發展網路新聞,以Word2vec類神經網路演算法,來建立財經新聞關鍵字的詞嵌入 (Word Embedding) 模型,以預測股票交易利多與利空的趨勢,最後我們將決策模型分成:趨勢向上、利空出盡、利多出盡、趨勢向下四類,讓使用者可以借助人工智慧的技術快速掌握第一手的投資資訊,輔助其決策,提高投資收益。
It has long been an interesting research direction for scholars and financial experts to conduct the prediction of stock price. Nevertheless, a bunch of noise together with non-linear data have created a tough hurdle in the assessment process with many uncertainties. Therefore, in this study, we will try to find an effective way to solve this problem through the deep learning technology, a prevalence of artificial intelligence in recent years. We first collect and extract the stock market data to provide the training data set based on technical analysis in the deep learning model. As the social networks are becoming popular in recent years, and the financial news are delivered in real time, we will try to add Word2vec neural network algorithm in our approach to conduct word-embedding model through the establishment of financial news keywords to predict the trend of stock pricing. Our approach divides the decision-making model into four categories: the trends up, negative news diminished, positive news diminished, and the trends down, such that users can quickly grasp the first-hand investment information, to assist their decision-making process and promote the ROI (Return on Investment).
中文摘要 i
誌 謝 iii
目 錄 iv
圖 目 錄 vi
表 目 錄 viii
一、緒論 1
1.1研究背景與動機 1
1.2研究目的 2
二、文獻探討 3
2.1股票交易 3
2.1.1當日沖銷 3
2.1.2高頻交易 3
2.1.3三大法人 3
2.2技術指標分析 4
2.2.1布林帶 4
2.2.2 隨機指標 5
2.3社群網路新聞分析 6
2.3.1Jieba斷詞 6
2.3.2tf-idf簡介 8
2.3.3Word2vec 10
2.4深度學習 (Deep Learning) 14
2.4.1類神經網路 14
2.4.2CUDA簡介 15
2.4.3Keras簡介 16
三、研究方法 18
3.1系統架構 18
3.2系統模組 20
3.2.1資料蒐集模組 20
3.2.2技術指標計算模組 28
3.2.3文字分析模組 30
3.2.4股價預測模組 32
3.3決策模型 33
四、實驗建置與結果評估 34
4.1系統環境與建置 34
4.2模組實作 36
4.2.1資料收集模組 36
4.2.2技術指標計算模組 37
4.2.3文字分析模組 38
4.2.4股價預測模組 41
4.3模擬投資 48
五、結論與未來研究 54
參考文獻 55
附錄一:網路爬蟲-證券交易所股票交易資料 58
附錄二:網路爬蟲程式碼-Unews財經新聞 61
附錄三:網路爬蟲-GoogleSearch API與PTT內容抓取(以台積電為例) 64
附錄四:Jieba斷詞與word2vec訓練 67
附錄五:股價預測訓練模組與訓練測試結果 69
附錄六:模擬投資 72

[1]Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G.S., Davis, A., Dean, J., Devin, M., Ghemawat, S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L., Kudlur, M., Levenberg, J., Mane, D., Monga, R., Moore, S., Murray, D., Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Talwar, K., Tucker, P., Vanhoucke, V., Vasudevan, V., Viegas, F., Vinyals, O., Warden, P., Wattenberg, M., Wicke, M., Yu, Y., and Zheng, X., “TensorFlow: Large-scale Machine Learning on Heterogeneous Systems,” Proceedings of the 12th USENIX conference on Operating Systems Design and Implementation, 2016, pp. 265-283.
[2]Abu-Mostafa, Y. S., and Atiya, A. F. “Introduction to Financial Forecasting,” Proceedings of the Applied Intelligence, 1996, pp. 205-213.
[3]Akita, R., Yoshihara, A., Matsubara, T., and Uehara, K., “Deep Learning for Stock Prediction Using Numerical and Textual Information,” Proceedings of the 2016 IEEE/ACIS 15th International Conference on Computer and Information Science (ICIS), June 2016, pp. 1-6.
[4]Atsalakis, G. S., and Valavanis, K. P., “Surveying Stock Market Forecasting Techniques–Part II: Soft Computing Methods,” Proceedings of the Expert Systems with Applications, Vol. 36, Issue 3, Part 2, April 2009, pp. 5932-5941.
[5]Bellman, R., “Dynamic Programming and Lagrange Multipliers,” Proceedings of the National Academy of Sciences, August 1956, pp. 767-769.
[6]Bollinger, J. , Bollinger on Bollinger Nands, McGraw Hill Professional, 2001.
[7]CUDA wikipedia,Last access: 2017-3-16. (available online at https://zh.wikipedia.org/wiki/CUDA)
[8]Di Persio, L., and Honchar, O., “Artificial Neural Networks Approach to The Forecast of Stock Market Price Movements,” Proceedings of the International Journal of Economics and Management Systems, Vol. 1, 2016, pp. 158-162
[9]Doeringer, W., Dykeman, D., Karjoth, G., Nassehi, M., and Sharma, M., “Variable Length Data Sequence Backtracking a Trie Structure,” U.S. Patent No. 5,787,430. Washington, DC: U.S. Patent and Trademark Office.
[10]Eddy, S. R., and “Hidden Markov Models,” Proceedings of the Current opinion in structural biology, Vol. 6, Issue 3, June 1996, pp. 361-365.
[11]Fukushima, K., Miyake, S., and Ito, T., “Neocognitron: A Neural Network Model for A Mechanism of Visual Pattern Recognition,” Proceedings of the IEEE Transactions on Systems, Man, and Cybernetics, Vol. SMC-13, Issue 5, Sept. 1983, pp. 826-834.
[12]Ichinose, K., and Shimada, K., “Stock Market Prediction from News on The Web and A New Evaluation Approach in Trading,” Proceedings of the 2016 5th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI), Jult 2016, pp. 77-81.
[13]Jieba github Last access: 2016-12-13. (available online at https://github.com/fxsjy/jieba)
[14]Keras Last access:2016-3-24 (available online at https://keras.io/)
[15]LeCun, Y., Bengio, Y., and Hinton, G., “Deep Learning,” Proceedings of the Nature, May 2015, pp. 436-444.
[16]Luhn, H. P., “A Statistical Approach to Mechanized Encoding and Searching of Literary Information,” Proceedings of the IBM Journal of research and development, Vol. 1, Issue 4, Oct. 1957, pp. 309-317.
[17]Martinez, L. C., da Hora, D. N., Palotti, J. R. D. M., Meira, W., and Pappa, G. L., “From an Artificial Neural Network to A Stock Market Day-Trading System: A Case Study on The BM&F BOVESPA,” Proceedings of the 2009 International Joint Conference on Neural Networks, July 2009, pp. 2006-2013.
[18]Mittermayer, M. A., “Forecasting Intraday Stock Price Trends with Text Mining Techniques,” Proceedings of the 37th Annual Hawaii International Conference on System Sciences, Jan. 2004, pp. 1-10.
[19]Nickolls, J., and Dally, W. J., “The GPU Computing Era,” Proceedings of the IEEE Micro, Vol. 30, Issue 2, March 2010, pp.56-69.
[20]Owens, J. D., Houston, M., Luebke, D., Green, S., Stone, J. E., and Phillips, J. C., “GPU Computing,” Proceedings of the IEEE, Vol. 96, Issue 5, May 2008, pp. 879-899.
[21]Papineni, K., “Why Inverse Document Frequency?,” Proceedings of the second meeting of the North American Chapter of the Association for Computational Linguistics on Language technologies, June 2001, pp. 1-8.
[22]Robertson, S., “Understanding Inverse Document Frequency: on Theoretical Arguments for IDF,” Proceedings of the Journal of documentation, Vol. 60, Issue 5, 2004, pp. 503-520.
[23]Salton, G., and Buckley, C., “Term-weighting approaches in automatic text retrieval,” Proceedings of the Information Processing & Management, Vol. 24, Issue 5, Jan. 1988, pp. 513-523.
[24]Yan, D., Zhou, G., Zhao, X., Tian, Y., and Yang, F., “Predicting Stock Using Microblog Moods,” Proceedings of the China Communications, Vol. 13, Issue 8, Aug. 2016, pp. 244-257.
[25]台灣證券交易所,Last access: 2016-11-9. (available online at http://www.tse.com.tw/ch/index.php)

第一頁 上一頁 下一頁 最後一頁 top
系統版面圖檔 系統版面圖檔