簡易檢索 / 詳目顯示
| 研究生： |
洪晨敏 Hong, Chen-Min |
|---|---|
| 論文名稱： |
機器學習與報酬預測：來自美國半導體產業的實證分析 Machine Learning and Return Predictability: Evidence from the U.S. Semiconductor Industry |
| 指導教授： |
徐愛恩
Tsui, Stephanie |
| 口試委員: |
朱琇妍
張景宏 |
| 學位類別： |
碩士
Master |
| 系所名稱： |
商學院 - 國際經營管理英語碩士學位學程(IMBA) International MBA Program College of Commerce(IMBA) |
| 論文出版年： | 2026 |
| 畢業學年度： | 114 |
| 語文別： | 英文 |
| 論文頁數： | 45 |
| 中文關鍵詞： | 資產定價 、機器學習 、報酬可預測性 、半導體產業 、投資組合績效 |
| 外文關鍵詞： | Asset Pricing, Machine Learning, Return Predictability, Semiconductor Industry, Portfolio Performance |
| 相關次數： | 點閱：31 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
本研究探討機器學習模型相較於傳統線性模型，是否能提升美國半導體產業股票報酬的預測能力。研究使用 2005 年至 2024 年來自 CRSP 與 Compustat 的公司層級資料，比較普通最小平方法（OLS）、Random Forest 與 XGBoost 三種模型在預測未來一個月超額報酬方面的表現，並同時以非半導體企業作為基準樣本進行比較。
實證結果顯示，所有模型在統計預測能力上皆表現有限，且機器學習模型並未顯著優於線性基準模型。然而，以投資組合績效作為評估標準時，則呈現具有經濟意義的結果，尤其在半導體樣本中更為明顯。儘管點預測能力有限，Long-only 與 Rank-weighted 投資策略仍維持相對穩定的績效，顯示模型仍具備有效的橫斷面排序能力。
此外，研究結果亦顯示，預測關係會隨市場環境而改變，且產業特定變數，尤其是與存貨相關的指標，在半導體產業中的預測重要性高於基準樣本。整體而言，本研究發現，預測模型的有效性不僅取決於模型本身的設計，也受到產業特性及預測關係穩定性的影響。本研究進一步凸顯了資產定價研究中統計預測能力與經濟實用性之間的重要差異。
This study examines whether machine learning models improve stock return predictability in the U.S. semiconductor industry relative to traditional linear models. Using firm-level data from CRSP and Compustat from 2005 to 2024, the analysis compares OLS, Random Forest, and XGBoost models in predicting one-month-ahead excess returns. Semiconductor firms are also compared with a non-semiconductor benchmark sample. The results show that all models exhibit weak statistical prediction performance, and machine learning models do not significantly outperform the linear benchmark. However, portfolio-based evaluation reveals economically meaningful results, particularly in the semiconductor sample. Long-only and rank-weighted strategies remain relatively stable despite weak point prediction, suggesting that the models retain useful cross-sectional ranking ability. The analysis also shows that predictive relationships vary across market environments and that industry-specific variables, particularly inventory-related measures, receive greater predictive importance in semiconductors than in the benchmark sample. Overall, the findings suggest that the usefulness of predictive models depends not only on model specification, but also on industry context and the stability of underlying predictive relationships. The study highlights the distinction between statistical prediction and economic usefulness in asset pricing.
1 Introduction 1
2 Literature Review 2
2.1 Asset Pricing and Machine Learning 2
2.2 Firm Characteristics and Return Predictability 4
2.3 Research Gap 4
3 Data and Sample Construction 5
3.1 Data Sources 5
3.2 Sample Selection 6
3.3 Data Cleaning and Matching 6
4 Variables Definition 7
4.1 Dependent Variable 8
4.2 Independent Variables 8
4.2.1 Industry-Specific Variables 9
4.2.2 Market-Based Variables 9
4.2.3 Accounting-Based Variables 10
4.3 Variable Construction 11
5 Methodology 11
5.1 Empirical Framework 11
5.2 Linear Models (Benchmark) 12
5.3 Machine Learning Model 12
5.4 Portfolio Construction 13
5.5 Regime Analysis 14
6 Empirical Results 15
6.1 Descriptive Statistics 15
6.2 Regression and ML Prediction Results 17
6.2.1 Linear Model (OLS) Results 17
6.2.2 Machine Learning Models Results 20
6.3 Model Comparison 22
6.4 Portfolio Performance 24
6.5 Regime Analysis (Pre vs Post 2023) 29
6.6 Factor Effectiveness Analysis 30
7 Discussion 33
7.1 Summary of Empirical Findings 33
7.2 Changes in Model Performance After 2023 33
7.3 Implications for Asset Pricing 34
7.4 Implications for Semiconductor Industry 35
8 Conclusion 35
8.1 Summary of Findings 36
8.2 Limitations 37
8.3 Suggestions for Future Research 38
Reference 39
Appendix 41
Avramov, D., Cheng, S., & Metzker, L. (2023). Machine learning vs. economic restrictions: Evidence from stock return predictability. Management Science.
Banz, R. W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics, 9(1), 3–18.
Campbell, J. Y. (2000). Asset pricing at the millennium. Journal of Finance, 55(4), 1515–1567.
Cooper, M. J., Gulen, H., & Schill, M. J. (2008). Asset growth and the cross-section of stock returns. Journal of Finance, 63(4), 1609–1651.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. Journal of Finance, 25(2), 383–417.
Fama, E. F., & French, K. R. (1992). The cross-section of expected stock returns. Journal of Finance, 47(2), 427–465.
Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189–1232.
Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. Review of Financial Studies, 33(5), 2223–2273.
Harvey, C. R., Liu, Y., & Zhu, H. (2016). …and the cross-section of expected returns. Review of Financial Studies, 29(1), 5–68.
Jegadeesh, N., & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance, 48(1), 65–91.
Kamalov, F., Gurrib, I., & Rajab, K. (2021). Forecasting stock returns using machine learning: Evidence from financial data transformations. Journal of Risk and Financial Management, 14(9), 418.
Novy-Marx, R. (2013). The other side of value: The gross profitability premium. Journal of Financial Economics, 108(1), 1–28.
Rossi, B. (2018). Forecasting stock market returns with machine learning. International Journal of Forecasting, 34(3), 498–514.
Shiller, R. J. (1981). Do stock prices move too much to be justified by subsequent changes in dividends? American Economic Review, 71(3), 421–436.
Titman, S., Wei, K. C. J., & Xie, F. (2004). Capital investments and stock returns. Journal of Financial and Quantitative Analysis, 39(4), 677–700.
Welch, I., & Goyal, A. (2008). A comprehensive look at the empirical performance of equity premium prediction. Review of Financial Studies, 21(4), 1455–1508.
Russell Investments. (2024). The Magnificent Seven: Market Concentrations and Complications. Russell Investments.
BNP Paribas Wealth Management. (2024). Time to Fade the Magnificent 7? BNP Paribas Wealth Management.
全文公開日期 2031/07/08