本研究以臺灣上市營造建設產業公司股票為示範例，應用機器學習、深度學習和自然語言處理模型等技術，結合財務多特徵因子，搭配多空投資的對沖交易策略模擬實證，創建自適應性風險的中短期投資組合決策模型。本研究透過自動化爬蟲程式，爬取基本財務報表資料、每日股市行情、籌碼資訊，以及財經新聞網站之新聞文本數據。經資料預處理後，應用XGBoost分類模型，優先對基本財務報表資料與歷史行情資料預測下一季別營造建設產業中被低估或高估的公司，以此建構每季財報區間的買多與賣空候選投資組合。接續採用基於BERT雙向自然語言處理模型所建構之情緒因子分數、歷史價格資料所計算之技術分析指標、三大法人進出場之交易數據與其他財務指標，透過滑動窗口結合LSTM迴歸模型預測每季『買多與賣空』候選投資組合內各公司未來每日的股價變化，以實現每日持股權重動態調整之交易策略。分析實證顯示，經初篩過的多空候選投資組合搭配財務多面向特徵所建立的動態權重調整模型為一穩健可行的方法。此外，於回測結果指出，依集成式學習模型方式為基礎建構的自適應風險對沖交易策略，不論總體市場趨勢，皆可使報酬率走勢穩定成長且擁有較低的波動風險，故於市場發生劇烈變化之際，應用集成式學習投資組合決策系統的多空投資對沖交易策略，可在降低波動性風險下創造相對穩定之超額報酬。本研究呈現之分析流程架構與研究成果除可用於營造建設產業，亦可配合投資人之自身專業領域，擴充涵蓋其它產業類別，預期能成為大型投資機構或投資經理人中短期交換持股、交易決策訊號判斷的輔助模組。

This research utilizes machine learning, deep learning, and natural language processing models to create an adaptive risk portfolio decision-making model for short-term investments. The analysis framework and process are demonstrated using the stocks of listed construction companies in Taiwan. This study utilizes an automated crawler program to obtain fundamental financial data, daily stock market information, chip data, and sentiment data. Based on financial statements and historical market data, preprocessed data is then used in an XGBoost classification model to predict undervalued or overvalued companies in the next quarter. Based on this, candidate investment portfolios for long and short positions are constructed for the next financial reporting quarter.
Moreover, the study proceeds to utilize sentiment factor scores based on a bidirectional natural language processing model (BERT), technical analysis indicators computed from historical price data, trading data from institutional investors, and other financial indicators. By employing a sliding window in conjunction with an LSTM regression model, the study forecasts the future daily stock price variations of individual companies within the long and short candidate investment portfolios for each quarter. This approach enables the realization of a trading strategy that dynamically adjusts the weights of stock holdings on a daily basis.
The empirical analysis reveals that the dynamically adjusted weighting model, constructed using pre-screened long and short candidate investment portfolios and incorporating various Multi-feature, is a robust and viable approach. Additionally, the backtesting results indicate that the adaptive risk-hedging trading strategy, based on an ensemble learning model, consistently achieves stable growth in return trends and exhibits favorable risk performance irrespective of the overall market trends.
Moreover, during periods of substantial market changes, the long-short hedging trading strategy implemented by the ensemble learning investment portfolio decision system can generate relatively stable excess returns while reducing volatility risks. The proposed analysis framework and research findings are not limited to the construction industry but can be extended to other industry sectors, catering to the specific expertise of investors. This research has the potential to serve as an auxiliary module for short-term trading decision signal judgments for large investment institutions or fund managers.

[1] J.-S. Chou, N.-M. Nguyen, and C.-P. Chang, "Intelligent candlestick forecast system for financial time-series analysis using metaheuristics-optimized multi-output machine learning," Applied Soft Computing, vol. 130, p. 109642, 2022.
[2] A. Thakkar and K. Chaudhari, "Fusion in stock market prediction: A decade survey on the necessity, recent developments, and potential future directions," Information Fusion, vol. 65, pp. 95-107, 2021/01/01/ 2021.
[3] I. K. Nti, A. F. Adekoya, and B. A. Weyori, "A systematic review of fundamental and technical analysis of stock market predictions," Artificial Intelligence Review, vol. 53, no. 4, pp. 3007-3057, 2020.
[4] E. Beyaz, F. Tekiner, X.-j. Zeng, and J. Keane, "Comparing technical and fundamental indicators in stock price forecasting," in 2018 IEEE 20th International Conference on High Performance Computing and Communications; IEEE 16th International Conference on Smart City; IEEE 4th International Conference on Data Science and Systems (HPCC/SmartCity/DSS), 2018: IEEE, pp. 1607-1613.
[5] K.-Y. Shen and G.-H. Tzeng, "Combined soft computing model for value stock selection based on fundamental analysis," Applied Soft Computing, vol. 37, pp. 142-155, 2015.
[6] L. Zhang, Y. Yin, and Y. You, "Portfolio Models Based on Fundamental Analysis Using Learning to Rank," in Proceedings of the Fifteenth International Conference on Management Science and Engineering Management: Volume 2 15, 2021: Springer, pp. 352-365.
[7] P. Agarwal, S. Bajpai, A. Pathak, and R. Angira, "Stock market price trend forecasting using machine learning," International Journal for Research in Applied Science and Engineering Technology, vol. 5, pp. 1673-1676, 2017.
[8] M. Sedighi, H. Jahangirnia, M. Gharakhani, and S. Farahani Fard, "A novel hybrid model for stock price forecasting based on metaheuristics and support vector machine," Data, vol. 4, no. 2, p. 75, 2019.
[9] X. Gabaix, P. Gopikrishnan, V. Plerou, and H. E. Stanley, "Institutional investors and stock market volatility," The Quarterly Journal of Economics, vol. 121, no. 2, pp. 461-504, 2006.
[10] M. T. Bohl and J. Brzeszczyński, "Do institutional investors destabilize stock prices? evidence from an emerging market," Journal of International Financial Markets, Institutions and Money, vol. 16, no. 4, pp. 370-383, 2006/10/01/ 2006.
[11] R. W. Sias and L. T. Starks, "Return autocorrelation and institutional investors," Journal of Financial Economics, vol. 46, no. 1, pp. 103-131, 1997/10/01/ 1997.
[12] J. Long, Z. Chen, W. He, T. Wu, and J. Ren, "An integrated framework of deep learning and knowledge graph for prediction of stock price trend: An application in Chinese stock exchange market," Applied Soft Computing, vol. 91, p. 106205, 2020/06/01/ 2020.
[13] W. Souma, I. Vodenska, and H. Aoyama, "Enhanced news sentiment analysis using deep learning methods," Journal of Computational Social Science, vol. 2, no. 1, pp. 33-46, 2019.
[14] M. G. Sousa, K. Sakiyama, L. de Souza Rodrigues, P. H. Moraes, E. R. Fernandes, and E. T. Matsubara, "BERT for stock market sentiment analysis," in 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI), 2019: IEEE, pp. 1597-1601.
[15] A. Picasso, S. Merello, Y. Ma, L. Oneto, and E. Cambria, "Technical analysis and sentiment embeddings for market trend prediction," Expert Systems with Applications, vol. 135, pp. 60-70, 2019/11/30/ 2019.
[16] Y. Li, K. Fu, Y. Zhao, and C. Yang, "How to make machine select stocks like fund managers? Use scoring and screening model," Expert Systems with Applications, vol. 196, p. 116629, 2022.
[17] E. F. Fama, "Efficient capital markets: II," The Journal of Finance, vol. 46, no. 5, pp. 1575-1617, 1991.
[18] W. F. De Bondt and R. H. Thaler, "Do security analysts overreact?," The American Economic Review, pp. 52-57, 1990.
[19] R. F. Stambaugh and Y. Yuan, "Mispricing factors," The review of financial studies, vol. 30, no. 4, pp. 1270-1315, 2017.
[20] R. F. Stambaugh, J. Yu, and Y. Yuan, "The short of it: Investor sentiment and anomalies," Journal of Financial Economics, vol. 104, no. 2, pp. 288-302, 2012.
[21] A. W. Lo, "The adaptive markets hypothesis: Market efficiency from an evolutionary perspective," Journal of Portfolio Management, Forthcoming, 2004.
[22] M. Usmani, S. H. Adil, K. Raza, and S. S. A. Ali, "Stock market prediction using machine learning techniques," in 2016 3rd International Conference on Computer and Information Sciences (ICCOINS), 15-17 Aug. 2016 2016, pp. 322-327, doi: 10.1109/ICCOINS.2016.7783235.
[23] R. Khalfaoui, S. Ben Jabeur, S. Hammoudeh, and W. Ben Arfi, "The role of political risk, uncertainty, and crude oil in predicting stock markets: evidence from the UAE economy," Annals of Operations Research, pp. 1-31, 2022.
[24] Y. Ma, Y. Wang, W. Wang, and C. Zhang, "Portfolios with return and volatility prediction for the energy stock market," Energy, vol. 270, p. 126958, 2023/05/01/ 2023.
[25] X. Zhang and W. Chen, "Stock Selection Based on Extreme Gradient Boosting," in 2019 Chinese Control Conference (CCC), 27-30 July 2019 2019, pp. 8926-8931, doi: 10.23919/ChiCC.2019.8865781.
[26] A. Dezhkam and M. T. Manzuri, "Forecasting stock market for an efficient portfolio by combining XGBoost and Hilbert–Huang transform," Engineering Applications of Artificial Intelligence, vol. 118, p. 105626, 2023/02/01/ 2023.
[27] W. Chen, H. Zhang, M. K. Mehlawat, and L. Jia, "Mean–variance portfolio optimization using machine learning-based stock price prediction," Applied Soft Computing, vol. 100, p. 106943, 2021/03/01/ 2021.
[28] E. Chong, C. Han, and F. C. Park, "Deep learning networks for stock market analysis and prediction: Methodology, data representations, and case studies," Expert Systems with Applications, vol. 83, pp. 187-205, 2017/10/15/ 2017.
[29] Y. Lin, Y. Yan, J. Xu, Y. Liao, and F. Ma, "Forecasting stock index price using the CEEMDAN-LSTM model," The North American Journal of Economics and Finance, vol. 57, p. 101421, 2021.
[30] S. Borovkova and I. Tsiamas, "An ensemble of LSTM neural networks for high‐frequency stock market classification," Journal of Forecasting, vol. 38, no. 6, pp. 600-619, 2019.
[31] P. Ghosh, A. Neufeld, and J. K. Sahoo, "Forecasting directional movements of stock prices for intraday trading using LSTM and random forests," Finance Research Letters, vol. 46, p. 102280, 2022/05/01/ 2022.
[32] X. Chen, X. Ma, H. Wang, X. Li, and C. Zhang, "A hierarchical attention network for stock prediction based on attentive multi-view news learning," Neurocomputing, vol. 504, pp. 1-15, 2022/09/14/ 2022.
[33] M. G. Sousa, K. Sakiyama, L. d. S. Rodrigues, P. H. Moraes, E. R. Fernandes, and E. T. Matsubara, "BERT for Stock Market Sentiment Analysis," in 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI), 4-6 Nov. 2019 2019, pp. 1597-1601, doi: 10.1109/ICTAI.2019.00231.
[34] K. Mishev, A. Gjorgjevikj, I. Vodenska, L. T. Chitkushev, and D. Trajanov, "Evaluation of sentiment analysis in finance: from lexicons to transformers," IEEE access, vol. 8, pp. 131662-131682, 2020.
[35] Q. Song, A. Liu, and S. Y. Yang, "Stock portfolio selection using learning-to-rank algorithms with news sentiment," Neurocomputing, vol. 264, pp. 20-28, 2017/11/15/ 2017.
[36] R. C. Grinold and R. N. Kahn, "The efficiency gains of long–short investing," Financial Analysts Journal, vol. 56, no. 6, pp. 40-53, 2000.
[37] J. Miffre, "Long-short commodity investing: A review of the literature," Journal of Commodity Markets, vol. 1, no. 1, pp. 3-13, 2016/03/01/ 2016.
[38] B. I. Jacobs, K. N. Levy, and D. Starer, "Long-short portfolio management: An integrated approach," Journal of Portfolio Management, vol. 25, pp. 23-32, 1999.
[39] S. Almahdi and S. Y. Yang, "An adaptive portfolio trading system: A risk-return portfolio optimization using recurrent reinforcement learning with expected maximum drawdown," Expert Systems with Applications, vol. 87, pp. 267-279, 2017/11/30/ 2017.
[40] M. Noma, "Value investing and financial statement analysis," Hitotsubashi Journal of Commerce and Management, pp. 29-46, 2010.
[41] R. A. Kamble, "Short and long term stock trend prediction using decision tree," in 2017 International Conference on Intelligent Computing and Control Systems (ICICCS), 2017: IEEE, pp. 1371-1375.
[42] Y. Yao, S. Cai, and H. Wang, "Are technical indicators helpful to investors in china’s stock market? A study based on some distribution forecast models and their combinations," Economic Research-Ekonomska Istraživanja, vol. 35, no. 1, pp. 2668-2692, 2022.
[43] J. D. Piotroski and D. T. Roulstone, "The influence of analysts, institutional investors, and insiders on the incorporation of market, industry, and firm‐specific information into stock prices," The accounting review, vol. 79, no. 4, pp. 1119-1151, 2004.
[44] T. Chen and C. Guestrin, "Xgboost: A scalable tree boosting system," in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016, pp. 785-794.
[45] J. Nobre and R. F. Neves, "Combining principal component analysis, discrete wavelet transform and XGBoost to trade in the financial markets," Expert Systems with Applications, vol. 125, pp. 181-194, 2019.
[46] D. Zhu, "Predicting stock volatility based on weighted fusion model of XGBoost and LightGBM," in 2022 3rd International Conference on Computer Vision, Image and Deep Learning & International Conference on Computer Engineering and Applications (CVIDL & ICCEA), 2022: IEEE, pp. 65-67.
[47] S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Computation, vol. 9, no. 8, pp. 1735-1780, 1997.
[48] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, "Bert: Pre-training of deep bidirectional transformers for language understanding," arXiv preprint arXiv:1810.04805, 2018.
[49] A. Vaswani et al., "Attention is all you need," Advances in Neural Information Processing Systems, vol. 30, 2017.
[50] W. Wang, W. Li, N. Zhang, and K. Liu, "Portfolio formation with preselection using deep learning from long-term financial data," Expert Systems with Applications, vol. 143, p. 113042, 2020/04/01/ 2020.
[51] J.-S. Chou, C.-Y. Liu, H. Prayogo, R. R. Khasani, D. Gho, and G. G. Lalitan, "Predicting nominal shear capacity of reinforced concrete wall in building by metaheuristics-optimized machine learning," Journal of Building Engineering, vol. 61, p. 105046, 2022.
[52] Y.-F. Chen and S.-H. Huang, "Sentiment-influenced trading system based on multimodal deep reinforcement learning," Applied Soft Computing, vol. 112, p. 107788, 2021/11/01/ 2021.
[53] D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," arXiv preprint arXiv:1412.6980, 2014.
[54] A. U. Haq, A. Zeb, Z. Lei, and D. Zhang, "Forecasting daily stock trend using multi-filter feature selection and deep learning," Expert Systems with Applications, vol. 168, p. 114444, 2021.
[55] H. Son, C. Hyun, D. Phan, and H. J. Hwang, "Data analytic approach for bankruptcy prediction," Expert Systems with Applications, vol. 138, p. 112816, 2019.