傳統量化交易與機器學習選股模型多著重在靜態橫截面特徵或單一時間序列，且往往假設金融市場維持穩定，忽略了市場中「概念飄移（Concept Drift）」與高雜訊特性。同時，過去文獻在回測設計上常忽略真實市場的流動性限制，導致模型在實戰中的泛化與變現能力受限。
為解決上述問題，本研究提出一整合時空動態表徵與適應性元學習之新型量化交易框架「時空動態元學習超圖神經網路（STM-HGNN）與梯度提升樹(XGBoost)融合模型」。該框架於內部建構Transformer時序編碼器(Temporal Encoder)以擷取個股歷史動能，並結合超圖神經網路（Hypergraph Neural Network）捕捉超越傳統成對關係（Pairwise）的產業板塊外溢效果。在模型訓練上，導入元學習（Meta-Learning）機制：外部迴圈(Outer Loop)捕捉市場長期底層規律，內層迴圈(Inner Loop)則處理突發市場事件的適應。此外，本研究採用專注於頂部預測之排序學習(Learning to Rank)，在最終排序股票階段透過倒數排名融合(Reciprocal Rank Fusion)技術將 STM-HGNN 及 XGBoost 二者之預測排名結果進行非線性整合，有效消除單一模型之偏誤。
本研究以上市、上櫃股票為對象，在設定日均成交值限制以嚴格排除流動性幻覺且無前視偏誤（Look-ahead Bias）的條件下，進行2019年至2024年實證回測。實證結果顯示，本研究所提之模型具備卓越的選股能力。在扣除嚴格的交易摩擦成本後，仍創造了極高的年化報酬率與優異的夏普比率，其績效表現顯著超越台灣加權股價指數(TAIEX)。總結而言，本研究不僅提出一個具備理論與實戰價值的選股框架，亦為有效應對金融時間序列之概念飄移現象、有效捕捉台股跨產業之連動與資金輪動特徵，提供了深具開創性的學術洞見。

Traditional quantitative trading and machine learning-based stock prediction models primarily focus on static cross-sectional features or single time series. These models often assume that financial markets remain stable, ignoring the "Concept Drift" and high-noise characteristics inherent in the market. Meanwhile, previous literature frequently overlooks real-world market liquidity constraints in backtesting designs, limiting the models' generalization and practical profitability in live trading.
To address these issues, this study proposes a novel quantitative trading framework integrating spatio-temporal dynamic embeddings and adaptive meta-learning: the "Spatio-Temporal Meta-Learning Hypergraph Neural Network (STM-HGNN) and XGBoost Ensemble Model." Internally, this framework constructs a Transformer Temporal Encoder to extract the historical momentum of individual stocks and incorporates a Hypergraph Neural Network to capture industry sector spillover effects that transcend traditional pairwise relationships. For model training, a Meta-Learning mechanism is introduced: the Outer Loop captures the long-term underlying patterns of the market, while the Inner Loop handles the rapid adaptation to sudden market events. Furthermore, this study adopts Learning to Rank, focusing on top-rank optimization strategy. In the final stock ranking stage, the Reciprocal Rank Fusion (RRF) technique is utilized to non-linearly integrate the predicted rankings of both STM-HGNN and XGBoost, effectively eliminating the bias of a single model.
Targeting stocks listed on the Taiwan Stock Exchange (TWSE), this study conducts an empirical backtest from 2019 to 2024 under strict conditions—implementing average daily trading value thresholds to rigorously eliminate liquidity illusion and ensuring no Look-ahead Bias. Empirical results demonstrate that the proposed model possesses exceptional stock selection capabilities. After deducting strict trading friction costs, it still generates an extremely high annualized return and an outstanding Sharpe ratio, significantly outperforming the Taiwan Capitalization Weighted Stock Index (TAIEX). In conclusion, this study not only proposes a stock selection framework with both theoretical and practical value but also provides pioneering academic insights into dealing with Concept Drift in financial time series and effectively capturing cross-industry correlations and capital rotation characteristics in the Taiwan stock market.

一、中文文獻
1. 鄭玉海 (2025)。圖神經網路於台灣股市長期股票排名預測之應用：基於橫截面數據與優化排序方法。未出版之碩士論文，國立政治大學。
二、英文文獻
1. Almgren, R., & Chriss, N. (2000). Optimal execution of portfolio transactions. Journal of Risk, 3(1), 5–40. https://doi.org/10.21314/jor.2001.041
2. Borisov, V., Leemann, T., Seßler, K., Haug, J., Pawelczyk, M., & Kasneci, G. (2021). Deep neural networks and tabular data: A survey. arXiv. https://doi.org/10.48550/arXiv.2110.01889
3. Burges, C. J. C. (1998). A tutorial on support vector machines for pattern recognition. Data Mining and Knowledge Discovery, 2(2), 121–167. https://doi.org/10.1023/A:1009715923555
4. Burges, C., Shaked, T., Renshaw, E., Lazier, A., Deeds, M., Hamilton, N., & Hullender, G. (2005). Learning to rank using gradient descent. In Proceedings of the 22nd International Conference on Machine Learning (ICML '05) (pp. 89–96). https://doi.org/10.1145/1102351.1102363
5. Cao, Z., Qin, T., Liu, T.-Y., Tsai, M.-F., & Li, H. (2007). Learning to rank: From pairwise approach to listwise approach. Proceedings of the 24th International Conference on Machine Learning (pp. 129–136). https://doi.org/10.1145/1273496.1273513
6. Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785–794). https://doi.org/10.1145/2939672.2939785
7. Cormack, G. V., Clarke, C. L., & Buettcher, S. (2009). Reciprocal rank fusion outperforms Condorcet and individual rank learning methods. Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 758–759). https://doi.org/10.1145/1571941.1572114
8. Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33(1), 3–56. https://doi.org/10.1016/0304-405X(93)90023-5
9. Feng, Y., You, H., Zhang, Z., Ji, R., & Gao, Y. (2019). Hypergraph neural networks. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 3558–3565. https://doi.org/10.1609/aaai.v33i01.33013558
10. Finn, C., Abbeel, P., & Levine, S. (2017). Model-agnostic meta-learning for fast adaptation of deep networks. Proceedings of the 34th International Conference on Machine Learning, 70, 1126–1135. https://arxiv.org/abs/1703.03400
11. French, R. M. (1999). Catastrophic forgetting in connectionist networks. Trends in Cognitive Sciences, 3(4), 128–135. https://doi.org/10.1016/S1364-6613(99)01294-2
12. Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies, 33(5), 2223–2273. https://doi.org/10.1093/rfs/hhaa009
13. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
14. Jordan, K., Jin, Y., Boza, V., You, J., Cesista, F., Newhouse, L., & Bernstein, J. (2024). Muon: MomentUm Orthogonalized by Newton-Schulz (arXiv:2410.11974). arXiv. https://arxiv.org/abs/2410.11974
15. Kumbure, M. M., Lohrmann, C., Luukka, P., & Porras, J. (2022). Machine learning techniques and data for stock market forecasting: A literature review. Expert Systems with Applications, 197, 116659. https://doi.org/10.1016/j.eswa.2022.116659
16. Liu, T.-Y. (2009). Learning to rank for information retrieval. Foundations and Trends in Information Retrieval, 3(3), 225–331. https://doi.org/10.1561/1500000016
17. López de Prado, M. (2018). Advances in financial machine learning.
18. Lu, J., Liu, A., Dong, F., Gu, F., Gama, J., & Zhang, G. (2019). Learning under concept drift: A review. IEEE Transactions on Knowledge and Data Engineering, 31(12), 2346–2363. https://doi.org/10.1109/TKDE.2018.2876857
19. Matsunaga, D., Suzumura, T., & Takahashi, T. (2019). Exploring graph neural networks for stock market predictions with rolling window analysis (arXiv:1909.10660). arXiv. https://arxiv.org/abs/1909.10660
20. McCloskey, M., & Cohen, N. J. (1989). Catastrophic interference in connectionist networks: The sequential learning problem. Psychology of Learning and Motivation, 24, 109–165. https://doi.org/10.1016/S0079-7421(08)60536-8
21. Narendra Babu, C., & Eswara Reddy, B. (2015). A moving-average filter based hybrid ARIMA–ANN model for forecasting time series data. Applied Soft Computing, 27, 325–337. https://doi.org/10.1016/j.asoc.2014.11.051
22. Qiao, Y., Xia, Y., Li, X., Li, Z., & Ge, Y. (2023). Higher-order graph attention network for stock selection with joint analysis (arXiv:2306.15526). arXiv. https://arxiv.org/abs/2306.15526
23. Sezer, O. B., Gudelek, M. U., & Ozbayoglu, A. M. (2020). Financial time series forecasting with deep learning: A systematic literature review: 2005–2019. Applied Soft Computing, 90, 106181. https://doi.org/10.1016/j.asoc.2020.106181
24. Thrun, S., & Pratt, L. (1998). Learning to learn. Springer.
25. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30. https://proceedings.neurips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html
26. Veličković, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., & Bengio, Y. (2018). Graph attention networks. arXiv preprint. https://arxiv.org/abs/1710.10903
27. Wen, Q., Zhou, T., Zhang, C., Chen, W., Ma, Z., Yan, J., & Sun, L. (2022). Transformers in time series: A survey. arXiv preprint. https://arxiv.org/abs/2202.07125
28. Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., & Yu, P. S. (2021). A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1), 4–24. https://doi.org/10.1109/TNNLS.2020.2978386
29. Yoo, J., Soun, Y., Park, Y. C., & Kang, U. (2021). Accurate multivariate stock movement prediction via data-axis transformer with multi-level contexts. Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 2037–2045). https://doi.org/10.1145/3447548.3467297
30. Zhao, L., Kong, S., & Shen, Y. (2023). DoubleAdapt: A meta-learning approach to incremental learning for stock trend forecasting (arXiv:2306.09862). arXiv. https://arxiv.org/abs/2306.09862