Title page for etd-0519120-085335
 論文名稱Title 深度學習多因子模型之應用－以台股市場為例Multi-Factor Model in Taiwan Stock Market: A Deep Learning Approach 系所名稱Department 財務管理學系Department of Finance 畢業學年期Year, semester 108 學年度 第 2 學期The spring semester of Academic Year 108 語文別Language 中文Chinese 學位類別Degree 碩士Master 頁數Number of pages 58 研究生Author 蔣維倫Wei-Lun Chiang 指導教授Advisor 召集委員Convenor 口試委員Advisory Committee 口試日期Date of Exam 2020-06-18 繳交日期Date of Submission 2020-06-19 關鍵字Keywords 深度學習、股票、量化分析、投資組合、多因子模型Quantitative analysis, Deep Learning, Multi Factor Model, Stocks, Portfolio 統計Statistics 本論文已被瀏覽 5756 次，被下載 0 次The thesis/dissertation has been browsed 5756 times, has been downloaded 0 times.
 中文摘要 本研究參照美商MSCI的 BARRA股票市場多因子線性模型建構流程，以個股基本面、價格面要素資訊，分組後以特定權重合成九大風格因子，並加入依照台灣經濟新報資料庫產業別資料形成二十八個產業因子，建構代表個股資訊的因子資料，並使用上述因子建模，達到預測個股報酬的目的。　　多因子線性模型建構後，本研究進而嘗試將模型以多層感知器(Multi-Layer Perception)框架改寫，建構深度學習多因子模型，達到要素資料的標準化、因子合成權重決定、因子值標準化、資產報酬估計等建模步驟自動化的目的。同時以多模型集成學習的機制，得到最終的個股報酬預測結果。　　實證結果顯示，使用2009至2018年上市股票資料進行回測，依模型預測報酬高低排序，使用每月篩選出預測報酬前5、10、20、30名股票依等權重、市值加權的方式建構投資組合，並持有一個月的方式，在考慮交易成本後，深度學習版本模型相較線性模型有更高的年化報酬、更低的年化風險，以及更小的最大回撤。 Abstract This research refers to Barra stock market multi-factor model construction process which published in MSCI Barra research note (2009). We constructed nine style factors and 28 industry factors which are based on stock’s fundamental information, price information, and industry code of the company, which grouped by Taiwan Stock Exchange. Each factor forms the daily frequency data of market to representing individual stock information. Those factors are used to build model to achieve the purpose of predicting individual stock returns.After construct a linear version factor model, this study further attempts to rewrite the original model with a deep learning framework to construct a deep learning multi-factor model to automate steps such as standardization of factor data, determination of weights to construct factor exposure, and standardization of factor exposure, and the estimated return of each stock. Furthermore, we also test the prediction power of the multi-model voting mechanism.By using the information of listed stock information from 2009 to 2018 is used for modeling. The empirical results show that after sorting stocks according to the model's predicted return level, and pick the top 5, 10, 20, and 30 stocks that are screened out for the predicted return each month are used to construct investment based on equal weight and market value weighting. After considering transaction costs, the MLP version model has higher annual returns, lower annual risks, and smaller maximum drawdowns than the BARRA (2009) linear version model.
 目次 Table of Contents 學位論文審定書 I摘要 IIABSTRACT III圖次 VI表次 VIII第一章 緒論 1第一節 研究動機 1第二節 研究目的 2第三節 研究架構 2第二章 文獻回顧 4第一節 資產定價及多因子模型 4第二節 深度學習 5第三節 深度學習財務預測應用 6第四節 小結 7第三章 研究方法 8第一節 實驗架構 8第二節 研究資料 10第三節 模型建構 17第四章 實證結果 25第一節 模型預測力測試 25第二節 線性模型績效 28第三節 深度學習模型績效 33表13等權重投組報酬差異T檢定 40表14市值加權投組報酬差異T檢定 40第五章 研究結論 41第一節 結論 41第二節 後續研究建議 42第六章 參考文獻 43第一節 英文文獻 43第二節 中文文獻 46附錄：風格因子列表 47
 參考文獻 References 第一節　英文文獻1. Ballings, Michel, Dirk Van den Poel, Nathalie Hespeels, and Ruben Gryp. “Evaluating Multiple Classifiers for Stock Price Direction Prediction.” Expert Systems with Applications 42, no. 20 (November 2015): 7046–56. 2. Fama, Eugene F., and Kenneth R. French. “A Five-Factor Asset Pricing Model.” Journal of Financial Economics 116, no. 1 (2015): 1–22.3. Fama, Eugene F., and Kenneth R. French. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal Of, 1993.4. Fama, Eugene F., and Kenneth R. French. “The Cross-Section of Expected Stock Returns.” The Journal of Finance 47, no. 2 (1992): 427–465.5. Fama, Eugene F., and James D. MacBeth. “Risk, Return, and Equilibrium: Empirical Tests.” Journal of Political Economy 81, no. 3 (1973): 607–636.6. Glorot, Xavier, Antoine Bordes, and Yoshua Bengio. “Deep Sparse Rectifier Neural Networks.” In Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, 315–323, 2011.7. He, Kaiming, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. “Delving Deep into Rectifiers: Surpassing Human-Level Performance on Imagenet Classification.” In Proceedings of the IEEE International Conference on Computer Vision, 1026–1034, 2015.8. Hochreiter, Sepp, and Jürgen Schmidhuber. “Long Short-Term Memory.” Neural Computation 9, no. 8 (1997): 1735–1780.9. Huck, Nicolas. “Pairs Selection and Outranking: An Application to the S&P 100 Index.” European Journal of Operational Research 196, no. 2 (2009): 819–825.10. Huck, Nicolas. “Pairs Trading and Outranking: The Multi-Step-Ahead Forecasting Case.” European Journal of Operational Research 207, no. 3 (2010): 1702–1716.11. Kingma, Diederik P., and Jimmy Ba. “Adam: A Method for Stochastic Optimization.” ArXiv Preprint ArXiv:1412.6980, 2014.12. Menchero, Jose. “Characteristics of Factor Portfolios, March 2010.” SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, March 30, 2010.13. Menchero, Jose, and B. Davis. “The Characteristics of Factor Portfolios.” Journal of Performance Measurement 15, no. 1 (2010): 52–62.14. Menchero, Jose, and Jyh-Huei Lee. “Efficiently Combining Multiple Sources of Alpha.” Journal Of Investment Management 13, no. 4 (2015): 71–86.15. Menchero, Jose, and Andrei Morozov. “Capturing Equity Risk Premia (August 2010).” SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, August 3, 2010.16. Menchero, Jose and Morozov, Alexander. “Decomposing Global Equity Cross-Sectional Volatility.” Financial Analysts Journal 67, no. 5 (2011): 58–68.17. Menchero, Jose, Andrei Morozov, and Peter Shepard. “Global Equity Risk Modeling.” In Handbook of Portfolio Construction, 439–480. Springer, 2010.18. Menchero, Jose, Andrei Morozov, and Peter Shepard. “The Barra Global Equity Model (Gem2).” MSCI Barra Research Notes 53 (2008).19. Rumelhart, David E., Geoffrey E. Hinton, and Ronald J. Williams. “Learning Representations by Back-Propagating Errors.” Nature 323, no. 6088 (1986): 533–536.20. Sezer, Omer Berat, A. Murat Ozbayoglu, and Erdogan Dogdu. “An Artificial Neural Network-Based Stock Trading System Using Technical Analysis and Big Data Framework.” In Proceedings of the Southeast Conference, 223–226, 2017.21. Sharpe, William F. “Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk.” The Journal of Finance 19, no. 3 (1964): 425–442.22. Thawornwong, Suraphan, and David Enke. “The Adaptive Selection of Financial and Economic Variables for Use with Artificial Neural Networks.” Neurocomputing 56 (2004): 205–232.23. Usmani, Mehak, Syed Hasan Adil, Kamran Raza, and Syed Saad Azhar Ali. “Stock Market Prediction Using Machine Learning Techniques.” In 2016 3rd International Conference on Computer and Information Sciences (ICCOINS), 322–327. IEEE, 2016.. 第二節　中文文獻24. 蔡彥鈞（2011）。多因子風險模型建構與其在增值指數基金上之應用－以新加坡市場為例。國立中山大學財務管理學系研究所碩士論文。25. 李政儒（2010）。多因子風險模型建構與應用於增值指數基金之績效分析－以中國大陸市場為例。國立中山大學財務管理學系研究所碩士論文。26. 蔡宗勳（2009）。多因子風險模型系統平台。國立中山大學財務管理學系研究所碩士論文。
 電子全文 Fulltext 本電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。論文使用權限 Thesis access permission：自定論文開放時間 user define開放時間 Available：校內 Campus：永不公開 not available校外 Off-campus：永不公開 not available您的 IP(校外) 位址是 34.239.176.54論文開放下載的時間是 校外不公開Your IP address is 34.239.176.54This thesis will be available to you on Indicate off-campus access is not available. 紙本論文 Printed copies 紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊，請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。開放時間 available 永不公開 not available
 QR Code