學術產出-Theses
Article View/Open
Publication Export
-
題名 利用股票市場圖形與機器學習配置最佳投資組合
Stock Chart Pattern with Machine Learning to Construct the Optimal Portfolio作者 何聿涵
He, Yu-Han貢獻者 黃泓智
Huang, Hong-Chih
何聿涵
He, Yu-Han關鍵詞 機器學習
自動編碼器
XGBoost
股票圖形
Machine Learning
AutoEncoder
stock chart pattern
XGBoost日期 2020 上傳時間 2-Sep-2020 11:51:21 (UTC+8) 摘要 近年隨著電腦技術及硬體設備的進步，人工智慧廣泛應用於各領域當中，因此本研究將利用其中圖形辨識之技術，搭配機器學習，期望創造高於台灣加權指數之報酬。本研究利用股票市場最常見之蠟燭圖與成交量圖作為資料庫資料，並對圖形進行兩階段之壓縮降維，分別為自動編碼器及主成分分析，成功萃取共500個資料特徵。將降維後的資料輸入進XGBoost模型中，預測未來20天股票股價，並利用交叉驗證以防止模型過度擬合，最終選取10檔股票作為投資組合。最後本文透過實證分析，分別對疫情發生前後做回測，回測期間為2012年至2019年底與2012年至2020年5月底。在疫情前，回測結果年化報酬率為25.2%，年化夏普比率為1.44；涵蓋疫情後，雖最大回撤率變動劇烈，但年化報酬率仍有20.6%及年化夏普比率1.17不錯之報酬，兩段期間皆優於台灣加權指數。
In recent years, with the advancement of computer technology and hardware equipment, artificial intelligence is widely used in various fields. Therefore, this study will use the technology of pattern recognition and machine learning to create a reward higher than the Taiwan Capitalization Weighted Stock Index(TAIEX).This study uses the most common candle charts and volume charts in the stock market as database data, and performs two-stage compression and dimensionality reduction on the stock chart pattern, which are AutoEncoder and principal component analysis, and successfully reduce total of 500 data features.Input the data after dimensionality reduction into the XGBoost model, predict the stock price in the next 20 days, and use cross-validation to prevent the model from overfitting, and finally select 10 stocks as the investment portfolio to construct the optimal portfolio.Finally, this study evaluates the investment portfolio through empirical analysis. The backtesting period is from 2012 to the end of 2019 and 2012 to the end of May 2020. Before COVID-19, the investment portfolio deliver an annual return rate was 25.2% and the annualized Sharpe ratio was 1.44. After the epidemic was covered, although the maximum drawdown rate changed drastically, the annualized return rate was still 20.6% and the annualized Sharpe ratio was 1.17. Both of periods are better than TAIEX.參考文獻 Bollen, J., H. Mao, and X. Zeng, 2011, Twitter mood predicts the stock market: Journal of computational science, v. 2, p. 1-8.Ding, X., Y. Zhang, T. Liu, and J. Duan, 2015, Deep learning for event-driven stock prediction: Twenty-fourth international joint conference on artificial intelligence.Freund, Y., 1995, Boosting a weak learning algorithm by majority: Information and computation, v. 121, p. 256-285.Freund, Y., and R. E. Schapire, 1995, A desicion-theoretic generalization of on-line learning and an application to boosting: European conference on computational learning theory, p. 23-37.Friedman, J. H., 2001, Greedy function approximation: a gradient boosting machine: Annals of statistics, p. 1189-1232.Hebb, D. O., 1949, The organization of behavior: a neuropsychological theory, J. Wiley; Chapman & Hall.Hunt, E. B., J. Marin, and P. J. Stone, 1966, Experiments in induction.Kara, Y., M. A. Boyacioglu, and Ö. K. Baykan, 2011, Predicting direction of stock price index movement using artificial neural networks and support vector machines: The sample of the Istanbul Stock Exchange: Expert systems with Applications, v. 38, p. 5311-5319.Kryzanowski, L., M. Galler, and D. W. Wright, 1993, Using artificial neural networks to pick stocks: Financial Analysts Journal, v. 49, p. 21-27.Pearson, K., 1901, LIII. On lines and planes of closest fit to systems of points in space: The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, v. 2, p. 559-572.Quinlan, J. R., 1986, Induction of decision trees: Machine learning, v. 1, p. 81-106.Rosenblatt, F., 1958, The perceptron: a probabilistic model for information storage and organization in the brain: Psychological review, v. 65, p. 386.Samuel, A. L., 1959, Some studies in machine learning using the game of checkers: IBM Journal of research and development, v. 3, p. 210-229.Schapire, R. E., 1990, The strength of weak learnability: Machine learning, v. 5, p. 197-227.Sezer, O. B., and A. M. Ozbayoglu, 2018, Algorithmic financial trading with deep convolutional neural networks: Time series to image conversion approach: Applied Soft Computing, v. 70, p. 525-538.Simonyan, K., and A. Zisserman, 2014, Very deep convolutional networks for large-scale image recognition: arXiv preprint arXiv:1409.1556.Valiant, L. G., 1984, A theory of the learnable: Communications of the ACM, v. 27, p. 1134-1142.陳暐文, 2019, 利用深度學習圖形辨識技術建置最適投資策略-以台灣股票市場為例. 描述 碩士
國立政治大學
風險管理與保險學系
107358028資料來源 http://thesis.lib.nccu.edu.tw/record/#G0107358028 資料類型 thesis dc.contributor.advisor 黃泓智 zh_TW dc.contributor.advisor Huang, Hong-Chih en_US dc.contributor.author (Authors) 何聿涵 zh_TW dc.contributor.author (Authors) He, Yu-Han en_US dc.creator (作者) 何聿涵 zh_TW dc.creator (作者) He, Yu-Han en_US dc.date (日期) 2020 en_US dc.date.accessioned 2-Sep-2020 11:51:21 (UTC+8) - dc.date.available 2-Sep-2020 11:51:21 (UTC+8) - dc.date.issued (上傳時間) 2-Sep-2020 11:51:21 (UTC+8) - dc.identifier (Other Identifiers) G0107358028 en_US dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/131518 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 風險管理與保險學系 zh_TW dc.description (描述) 107358028 zh_TW dc.description.abstract (摘要) 近年隨著電腦技術及硬體設備的進步，人工智慧廣泛應用於各領域當中，因此本研究將利用其中圖形辨識之技術，搭配機器學習，期望創造高於台灣加權指數之報酬。本研究利用股票市場最常見之蠟燭圖與成交量圖作為資料庫資料，並對圖形進行兩階段之壓縮降維，分別為自動編碼器及主成分分析，成功萃取共500個資料特徵。將降維後的資料輸入進XGBoost模型中，預測未來20天股票股價，並利用交叉驗證以防止模型過度擬合，最終選取10檔股票作為投資組合。最後本文透過實證分析，分別對疫情發生前後做回測，回測期間為2012年至2019年底與2012年至2020年5月底。在疫情前，回測結果年化報酬率為25.2%，年化夏普比率為1.44；涵蓋疫情後，雖最大回撤率變動劇烈，但年化報酬率仍有20.6%及年化夏普比率1.17不錯之報酬，兩段期間皆優於台灣加權指數。 zh_TW dc.description.abstract (摘要) In recent years, with the advancement of computer technology and hardware equipment, artificial intelligence is widely used in various fields. Therefore, this study will use the technology of pattern recognition and machine learning to create a reward higher than the Taiwan Capitalization Weighted Stock Index(TAIEX).This study uses the most common candle charts and volume charts in the stock market as database data, and performs two-stage compression and dimensionality reduction on the stock chart pattern, which are AutoEncoder and principal component analysis, and successfully reduce total of 500 data features.Input the data after dimensionality reduction into the XGBoost model, predict the stock price in the next 20 days, and use cross-validation to prevent the model from overfitting, and finally select 10 stocks as the investment portfolio to construct the optimal portfolio.Finally, this study evaluates the investment portfolio through empirical analysis. The backtesting period is from 2012 to the end of 2019 and 2012 to the end of May 2020. Before COVID-19, the investment portfolio deliver an annual return rate was 25.2% and the annualized Sharpe ratio was 1.44. After the epidemic was covered, although the maximum drawdown rate changed drastically, the annualized return rate was still 20.6% and the annualized Sharpe ratio was 1.17. Both of periods are better than TAIEX. en_US dc.description.tableofcontents 第一章、 緒論 1第一節、 研究動機與研究背景 1第二節、 研究目的 2第三節、 研究流程 2第二章、 文獻回顧 4第一節、 機器學習之文獻回顧 4第二節、 機器學習應用於股票預測之文獻回顧 5第三章、 研究方法 7第一節、 圖形資料建構 7第二節、 圖形特徵萃取及降維 9一、 自動編碼器(Autoencoder) 9二、 主成分分析(Principal Component Analysis；PCA) 11第三節、 機器學習模型建構—XGBOOST 12第四節、 投資策略及績效指標 14第四章、 實證結果分析 16第一節、 實證分析樣本來源 16第二節、 最終預測值選擇 16一、 回測時段(疫情前)：2012年1月1日至2019年12月31日 17二、 回測時段(含疫情)：2012年1月1日至2020年05月31日 18第三節、 資產配置方式 20第五章、 結果與未來建議 23第一節、 結論 23第二節、 未來建議 23參考文獻 25 zh_TW dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0107358028 en_US dc.subject (關鍵詞) 機器學習 zh_TW dc.subject (關鍵詞) 自動編碼器 zh_TW dc.subject (關鍵詞) XGBoost zh_TW dc.subject (關鍵詞) 股票圖形 zh_TW dc.subject (關鍵詞) Machine Learning en_US dc.subject (關鍵詞) AutoEncoder en_US dc.subject (關鍵詞) stock chart pattern en_US dc.subject (關鍵詞) XGBoost en_US dc.title (題名) 利用股票市場圖形與機器學習配置最佳投資組合 zh_TW dc.title (題名) Stock Chart Pattern with Machine Learning to Construct the Optimal Portfolio en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) Bollen, J., H. Mao, and X. Zeng, 2011, Twitter mood predicts the stock market: Journal of computational science, v. 2, p. 1-8.Ding, X., Y. Zhang, T. Liu, and J. Duan, 2015, Deep learning for event-driven stock prediction: Twenty-fourth international joint conference on artificial intelligence.Freund, Y., 1995, Boosting a weak learning algorithm by majority: Information and computation, v. 121, p. 256-285.Freund, Y., and R. E. Schapire, 1995, A desicion-theoretic generalization of on-line learning and an application to boosting: European conference on computational learning theory, p. 23-37.Friedman, J. H., 2001, Greedy function approximation: a gradient boosting machine: Annals of statistics, p. 1189-1232.Hebb, D. O., 1949, The organization of behavior: a neuropsychological theory, J. Wiley; Chapman & Hall.Hunt, E. B., J. Marin, and P. J. Stone, 1966, Experiments in induction.Kara, Y., M. A. Boyacioglu, and Ö. K. Baykan, 2011, Predicting direction of stock price index movement using artificial neural networks and support vector machines: The sample of the Istanbul Stock Exchange: Expert systems with Applications, v. 38, p. 5311-5319.Kryzanowski, L., M. Galler, and D. W. Wright, 1993, Using artificial neural networks to pick stocks: Financial Analysts Journal, v. 49, p. 21-27.Pearson, K., 1901, LIII. On lines and planes of closest fit to systems of points in space: The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, v. 2, p. 559-572.Quinlan, J. R., 1986, Induction of decision trees: Machine learning, v. 1, p. 81-106.Rosenblatt, F., 1958, The perceptron: a probabilistic model for information storage and organization in the brain: Psychological review, v. 65, p. 386.Samuel, A. L., 1959, Some studies in machine learning using the game of checkers: IBM Journal of research and development, v. 3, p. 210-229.Schapire, R. E., 1990, The strength of weak learnability: Machine learning, v. 5, p. 197-227.Sezer, O. B., and A. M. Ozbayoglu, 2018, Algorithmic financial trading with deep convolutional neural networks: Time series to image conversion approach: Applied Soft Computing, v. 70, p. 525-538.Simonyan, K., and A. Zisserman, 2014, Very deep convolutional networks for large-scale image recognition: arXiv preprint arXiv:1409.1556.Valiant, L. G., 1984, A theory of the learnable: Communications of the ACM, v. 27, p. 1134-1142.陳暐文, 2019, 利用深度學習圖形辨識技術建置最適投資策略-以台灣股票市場為例. zh_TW dc.identifier.doi (DOI) 10.6814/NCCU202001231 en_US
