本研究利用台灣股票價量資料與國際大盤指數資料進行特徵工程，並使用XGBoost演算法來訓練模型。在XGBoost中又分別使用分類預測模型與迴歸預測模型來預測五日內的最大跌幅。透過不同模型的堆疊下，取出交叉篩選後的最佳結果。
在模型預測結果分析中，兩種單模型篩選下預測結果皆勝過全樣本中的平均最大跌幅，以及在超過10個百分比跌幅情況下所佔的比例。顯示模型的預測能提升對於最大跌幅的控制。分類預測模型能將平均最大跌幅從比較基準的
-6%提升到-8%，而迴歸模型能提升至-6%。在準確率方面，分類模型能從比較基準的6% 提升至35%，迴歸模型則能提升至20%。

This study uses Taiwan stock price and volume data and international market index data for feature engineering, and uses the XGBoost algorithm to train models. Classification prediction models and regression prediction models were used to predict the largest down range in five days. In XGBoost, the classification prediction model and the regression prediction model are used to predict the maximum down range in five days. Then stack different models to get the results.
Results shows that both models outperformed the average maximum down range in the full sample, and the proportion of the full sample in the case of more than 10 percent down range. It means that models can improve the control of the maximum down range. In the average maximum down range, compared to the benchmark -3%, the classification model can improve it to -8% with the change of the threshold value; the regression model can improve it to -6%. In terms of precision rate, compared to the benchmark 6%, the classification model can improve it to 35% with the change of the threshold value; the regression model can improve it to 20%.

Contents
論文審定書 i
摘要 ii
Abstract iii
Contents iv
List of Figures v
List of Tables vi
Chapter 1 Introduction 1
1.1 Research Background 1
1.2 Research Purpose 2
1.3 Research Construction and Process 3
Chapter 2 Literature Review 4
2.1 Technical Indicators and Chip Indicators 4
2.2 Machine Learning in Stock Prediction 5
Chapter 3 Methodology 8
3.1 Research Design 8
3.2 Data Collection 11
3.3 Feature Engineering 11
3.3.1 Prediction Target 11
3.3.2 Taiwan Stock Price and Volume Data 12
3.3.3 Taiwan Stock Market Chip Data 15
3.3.4 International Market Index Data 16
3.3.5 Standardizing 16
3.4 Algorithm 17
3.4 K-Fold Cross Validation 20
Chapter 4 Empirical Results 22
4.1 Model Analysis 22
4.1.1 Classfication Model Analysis 23
4.1.2 Regression Model Analysis 24
4.1.3 Combined Model Analysis 25
4.1.4 Accumulation of Correlation Coefficient Plot 29
4.2 Portfolio Backtesting 31
Chapter 5 Conclusion and Recommendation 35
Reference 36

Reference
1. Douglas Foster, F., David R. Gallagher, and Adrian Looi. ‘Institutional Trading and Share Returns’. Journal of Banking & Finance 35, no. 12 (1 December 2011): 3383–99.
2. Alhashel, Bader, Fahad Almudhaf, and J. Andrew Hansz. ‘Can Technical Analysis Generate Superior Returns in Securitized Property Markets? Evidence From East Asia Markets’. SSRN Scholarly Paper. Rochester, NY, 2018.
3. Tsai, Chih-Fong, Yuah-Chiao Lin, David C. Yen, and Yan-Min Chen. ‘Predicting Stock Returns by Classifier Ensembles’. Applied Soft Computing, The Impact of Soft Computing for the Progress of Artificial Intelligence, 11, no. 2 (1 March 2011): 2452–59.
4. Basak, Suryoday, Saibal Kar, Snehanshu Saha, Luckyson Khaidem, and Sudeepa Roy Dey. ‘Predicting the Direction of Stock Market Prices Using Tree-Based Classifiers’. The North American Journal of Economics and Finance 47, no. C (2019): 552–67.
5. Suryoday, B., Saibal, K., Snehanshu, S., Luckyson, K., & Sudeepa, R. (2019). Predicting the direction of stock market prices using tree-based classifiers. The North American Journal of Economics and Finance, 47, 552–567.
6. Sharma, Nonita, and Akanksha Juneja. ‘Combining of Random Forest Estimates Using LSboost for Stock Market Index Prediction’. In 2017 2nd International Conference for Convergence in Technology (I2CT), 1199–1202, 2017.
7. Chen, Tianqi, and Carlos Guestrin. “XGBoost: A Scalable Tree Boosting System.” In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–94. KDD ’16. New York, NY, USA: Association for Computing Machinery, 2016.
8. Kara, Yakup, Melek Acar Boyacioglu, and Ömer Kaan Baykan. “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 38, no. 5 (May 1, 2011): 5311–19.
9. Sortino, Frank A., and Lee N. Price. “Performance Measurement in a Downside Risk Framework.” The Journal of Investing 3, no. 3 (August 31, 1994): 59–64.
10. Zhang, Xiaoyun, and Wanyi Chen. “Stock Selection Based on Extreme Gradient Boosting,” 8926–31, 2019.
11. 陳彥豪。「外資與投信法人持股比率變化對股價報酬率影響之研究-以上市 電子股為例」。碩士論文，國立中山大學財務管理學系研究所，2002。
12. 范聖培。「三大法人之買賣超行為對股價短期報酬之研究」。碩士論文，國立中央大學財務金融學系在職專班，2014。
13. 陳子軒。「運用基本面、技術面及籌碼面建構台股深度學習交易策略」。碩士論文，國立中山大學金融創新產業碩士專班，2018。
14. 黃祺敦。「運用當日籌碼面變數預測隔日股價方向」。碩士論文，國立中正大學國際經濟研究所，2012。