本研究首先使用機器學習的模型，比較主成份分析（PCA）與自動編碼器（Autoencoder）兩種方式做降維後的資料擬合之結果，並且以測試集的R^2衡量表現，結果顯示，經過自動編碼器預訓練後的資料能更大幅度的提升模型性能。下一步，使用前面訓練好的模型作為弱學習器，以簡單平均的方式做集成，比較三個弱學習器與集成後的預測表現，再以模型預測結果作為買賣訊號來建構外匯投資組合，同時，加入利差策略與動量策略作為比較基準，觀察投資組合的績效表現，根據實驗結果，集成後的投組明顯優於個別機器學習模型，而機器學習模型又優於傳統策略。

In this study, we compare two different techniques which are principal component analysis (PCA) and autoencoder(AE) for reducing the dimensionality of data prior to modeling, and deploy machine learning models for data fitting to observe their results. Then, we measure performance by R^2 on the test set. The results showed that the data pre-trained by AE can greatly improve the model performance. The next step is to use previously trained models as weak learners to combine them by simple average method and compare its result to weak learners. After that, we adopt the result of model prediction as a trading signal to construct a foreign exchange portfolio. Moreover, we also add traditional strategies which are carry strategy and momentum strategy as the benchmarks to observe the portfolio performance. According to the experimental results, the composite is better than all weak learners, and all weak learners are better than the traditional strategies.

[1] 石川、劉洋溢、連祥斌(2020)。因子投資與實踐。北京：電子工業出版社。

[2] Alkhatib, K., Najdat, H., Hmeidi, L., Shatnawi, M. (2013). Stock Price Prediction Using K-Nearest Neighbor(kNN) Algorithm. International Journal of Business, Humanities and Technology, 3(3).

[3] Andersen, T.G., Bollerslev, T., Diebold, F.X., & Vega, C. (2002). Micro Effects of Macro Announcements: Real-Time Price Discovery in Foreign Exchange. The American Economic Association, 93, 38-62.

[4] Asness, C.S., Moskowitz, T.J., & Pedersen, L. (2013). Value and Momentum Everywhere. Journal of Finance, 68, 929-885.

[5] Baldi, P., & Hornik, K. (1989). Neural networks and principal component analysis: Learning from examples without local minima. Neural networks, 2, 53-58.

[6] Basak, S., Kar, S., Saha, S., Khaidem, L., & Dey, S.R. (2019). Prediction the direction of stock market prices using tree-based classifiers. The North American Journal of Economics and Finance, 47, 552-567.

[7] Breiman, L. (1996a). Bagging Predictors. Machine Learning, 24(2), 123–140.

[8] Breiman, L. (2001). Random Forests. Machine Learning, 45, 5–32.

[9] Breiman, L., Friedman, J., Olshen, R., & Stone, C. (1984). Classification and Regression Trees. Wadsworth.

[10] Carhart, M.M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance, 52, 57-82.

[11] Chen, T., & Carlos G. (2016). Xgboost: A scalable tree boosting system. A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794.

[12] Clemen, R.T. (1989). Combining Forecasts: A Review and Annotated Bibliography. International Journal of Forecasting, 5, 559–583.

[13] Ebrahimpour, D.H., & Kouzani, D.A. (2007). FACE RECOGNITION USING BAGGING KNN.

[14] Fama, E.F.(1984). Forward and spot exchange rates. The Journal of Monetary Economics, 14, 319-338.

[15] Fama, E.F., & French, K.R.(1992). The Cross-Section of Expected Stock Returns. Journal of Finance, 47, 427-465.

[16] Fama, E.F., & French, K.R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33, 3-56.

[17] Fama, E.F., & MacBeth, J. (1973). Risk, return and equilibrium: Empirical tests. The Journal of Political Economy, 81, 607-636.

[18] Friedman, J.H. (2001). GREEDY FUNCTION APPROXIMATION: A GRADIENT BOOSTING MACHINE. The Annals of Statistics, 29, 1189-1232.

[19] George, A. (2012). Anomaly Detection based on Machine Learning: Dimensionality Reduction using PCA and Classification using SVM. International Journal of Computer Applications, 47, 5-8.

[20] Gu, S., Kelly, B. & Xiu, D. (2020). Empirical Asset Pricing via Machine Learning. The Review of Financial Studies, 33, 2223-2273.

[21] Gu, S., Kelly, B. & Xiu, D. (2021). Autoencoder asset pricing models. Journal of Econometrics, 222, 429-450.

[22] Hasen, L.P., & Hodrick, R.J. (1980). Forward Exchange Rates as Optimal Predictors of Future Spot Rates: An Econometric Analysis. The Journal of Political Economy, 88, 829-853.

[23] Hasen, L.K., & Salamom, P. (1990). Neural network ensembles. IEEF Transactions on Pattern Analysis and Machine Intelligence, 12, 993-1001.

[24] Hiton, G.E. and Salakhutdinov, R.R. (2006). Reducing the Dimensionality of Data with Neural Networks. Science, 33, 504-507.

[25] Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 48, 65-91.

[26] Jiang, Z., Ji, R., & Chang, K.C. (2020). A Machine Learning Integrated Portfolio Rebalance Framework with Risk-Aversion Adjustment. Journal of Risk and Financial Management, 13, 155.

[27] Jurczenko, E., & Teiletche, J., (2019). Macro Factor Mimicking Portfolios. Available at SSRN 3363598.

[28] Kelly, B.T., Pruitt, S., & Su., Y. (2018). Characteristics Are Covariances: A Unified Model of Risk and Return. Available at SSRN 3032013.

[29] Kroencke, T. A., Schindler, F., & Schrimpf, A. (2014). International diversification benefits with foreign exchange investment styles. Review of Finance, 18, 1847-1883.

[30] Levy, R.A. (1967). Relative Strength as a Criterion for Investment Selection. Journal of Finance, 22, 595-610.

[31] Lustig, H., Roussanov, N., & Verdelhan, A. (2011). Common risk factors in currency markets. The Review of Financial Studies, 24, 3731-3777.

[32] Lustig, H., & Verdelhan, A. (2007). The cross-section of foreign currency risk premia and US consumption growth risk. The American Economic Review, 97, 89–117.

[33] Menkhoff, L., Sarno, L., Shmeling, M., & Schrimpf, A. (2012a). Carry Trades and Global Foreign Exchange Volatility. Journal of Finance, 67, 681-718.

[34] Menkhoff, L., Sarno, L., Schmeling, M., & Schrimpf, A. (2012b). Currency momentum strategies. Journal of Financial Economics, 106, 660-684.

[35] Okunev, J., & White, D. (2003). Do Momentum-Based Strategies Still Work in Foreign Currency Markets?, Journal of Financial and Quantitative Analysis, 38(2), 425-447.

[36] Qian, B., & Rasheed, K. (2010). Foreign Exchange Market Prediction with Multiple Classifiers. Journal of Forecasting, 29, 271-284.

[37] Qian, X.Y. (2017). Financial Series Prediction: Between Precision of Time Series Models and Machine Learning Methods. arXiv: 1706.00948v3.

[38] Simpson, M.W., Ramchander, S., & Chaudhry, M. (2005). The impact of macroeconomic surprises on spot and forward foreign exchange markets. Journal of International Money and Finance, 24, 693-718.

[39] Steele, B.M. (2009). Exact bootstrap k-nearest neighbor learners. Mach Learn, 74, 235-255.

[40] Struck, C., & Cheng, E. (2019). The Cross-Section of Returns: A Non-Parametric Approach. Available at SSRN 3494141.

[41] Waskle, S., Parashar, L., & Singh, U. (2020). Intrusion Detection System Using PCA with Random Forest Approach. International Conference on Electronics and Sustainable Communication System, 2020, 803-808.

[42] Yu, L., Lai, K.K. & Wang, S. (2008). Multistage RBF neural network ensemble learning for exchange rates forecasting. Neurocomputing, 71, 3295-3303.

[43] Zhang, D., Wang, X., Gao, L., & Gong, Y. (2021). Predict and Analyze Exchange Rate Fluctuations Accordingly Based on Quantile Regression Model and K-Nearest Neighbor. Journal of Physics: Conference Series, 12-16.

[44] Zhou, Z., & Yu, Y. (2005). Adapt Bagging to Nearest Neighbor Classifiers. Journal of Computer Science and Technology, 20, 48-54.