臺灣博碩士論文加值系統

The aim of this thesis is to evaluate machine learning models for the purpose of predicting asset prices. The models are limited to neural networks, specifically the multilayer perceptron, the recurrent neural network, and the pi-sigma network. The asset used as forecasting target is the EURUSD. The thesis discusses and compares the differences between the neural network architectures, and uses a custom fitness function during the models’ training. Results show that the three models can be used as forecasting tools to build profitable trading strategies. To enhance the validity of the results, the performance of the three models are compared to the discoveries of other researchers, and among the three neural networks, the pi-sigma network displays the best forecasting performance, in line with what other studies show as well. Besides comparing the performance of the three models, the thesis also discusses how the forecasts can be improved by combining several machine learning models. The thesis also discusses potential reasons for why some scholars achieve dissatisfying forecasting results when testing similar types of neural networks, and provides some suggestions of how results might be improved.

1. Introduction 6
1.1. Research Methodology 7
2. Literature Review 9
2.1. Forecasting Methods 10
2.2. Economic Theory 11
2.3. Criticism of Technical Analysis 13
2.4. Machine Learning 13
2.4.1. Neural Networks 14
2.4.2. Overfitting 15
2.5. Neural Network Architectures 15
2.5.1. Multilayer Perceptron 16
2.5.2. Recurrent Neural Network 18
2.5.3. Pi-Sigma Network 19
3. Analysis 22
3.1. Preparing the Dataset 22
3.2. The Naive Trading Strategy 24
3.3. Training the Neural Networks 25
3.4. Trading Strategy, Transaction Costs, and Trading Volume 26
3.5. Training and Trading Results 28
3.5.1. Comment 28
3.6. Comparing the Results 29
4. Discussion 33
4.1. Importance of Fitness Function 33
4.2. Importance of Careful Training 34
4.3. Machine Learning for Predicting Asset Prices 36
5. Conclusion 38
6. Appendix 40
7. References 41

Bank for International Settlements. (2016). Triennial Central Bank Survey - Foreign exchange turnover in April 2016. Retrieved July 10, 2018, from Bank for International Settlements: https://www.bis.org/publ/rpfx16fx.pdf
Bishop, C. M. (2006). Pattern Recognition and Machine Learning. New York: Springer.
Björklund, S., & Uhlin, T. (2017). Artificial Neural Networks for Financial Time Series Prediction and Portfolio Optimization. Master of Science Thesis in Industrial Engineering and Management, Department of Management and Engineering, Linköping University.
Brock, W., Lakonishok, J., & LeBaron, B. (1992). Simple Technical Trading Rules and the Stochastic Properties of Stock Returns. Journal of Finance, 47(5), pp. 1731-1764. doi:10.2307/2328994
Castellano, G., Fanelli, A. M., & Pelillo, M. (1997). An Iterative Pruning Algorithm for Feedforward Neural Networks. IEEE Transactions on Neural Networks, 8(3), pp. 519-531. doi:10.1109/72.572092
Dunis, C. L., & Huang, X. (2002). Forecasting and Trading Currency Volatility: An Application of Recurrent Neural Regression and Model Combination. Journal of Forecasting, 21(5), pp. 317-354. doi:10.1002/for.833
Elman, J. L. (1990). Finding Structure in Time. Cognitive Science, 14(2), pp. 179-211. doi:10.1016/0364-0213(90)90002-E
European Central Bank. (2018). Statistical Data Warehouse - Reference rates. Retrieved June 30, 2018, from European Central Bank: http://sdw.ecb.europa.eu/browse.do?node=9691296
Fama, E. F. (1965). Random Walks in Stock Market Prices . Financial Analysts Journal, 21(5), pp. 55-59. doi:10.2469/faj.v21.n5.55
Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of Finance, 25(2), pp. 383-417. doi:10.2307/2325486
Ghazali, R. (2007). Higher Order Neural Networks for Financial Time Series Prediction. Doctor of Philosophy Thesis in Computer Science, School of Computing & Mathematical Sciences, Liverpool John Moores University. doi:10.24377/LJMU.t.00005879
Harrald, P. G., & Kamstra, M. (1997). Evolving Artificial Neural Networks to Combine Financial Forecasts. IEEE Transactions on Evolutionary Computation, 1(1), pp. 40-52. doi:10.1109/4235.585891
Hsu, P.-H., & Kuan, C.-M. (2005). Re-Examining the Profitability of Technical Analysis with White’s Reality Check and Hansen’s Spa Test. Journal of Financial Econometrics, 3(4). doi:10.2139/ssrn.685361
Hull, J. C. (2015). Risk Management and Financial Institutions. New Jersey: John Wiley & Sons.
Ince, H., & Trafalis, T. B. (2006). A Hybrid Model for Exchange Rate Prediction. Decision Support Systems, 42(2), pp. 1054-1062. doi:10.1016/j.dss.2005.09.001
Jensen, M. C., & Benington, G. A. (1970). Random Walks and Technical Theories: Some Additional Evidence. Journal of Finance, 25(2), pp. 469-482. doi:10.1111/j.1540-6261.1970.tb00671.x
Kaastra, I., & Boyd, M. (1996). Designing a Neural Network for Forecasting Financial and Economic Time Series. Neurocomputing, 10(3), pp. 215-236. doi:10.1016/0925-2312(95)00039-9
Ling, S. S., Leung, F. H., Lam, H.-K., Lee, Y.-S., & Tam, P. K. (2003). A Novel Genetic-Algorithm-Based Neural Network for Short-Term Load Forecasting. IEEE Transactions on Industrial Electronics, 50(4), pp. 793-799. doi:10.1109/TIE.2003.814869
Lo, A. W. (2004). The Adaptive Markets Hypothesis: Market Efficiency from an Evolutionary Perspective. Journal of Portfolio Management, 30(5), pp. 15-29. doi:10.3905/jpm.2004.442611
Malkiel, B. G. (2005). Reflections on the Efficient Market Hypothesis: 30 Years Later. The Financial Review, 40(1), pp. 1-9. doi:10.1111/j.0732-8516.2005.00090.x
Murphy, J. J. (1999). Technical Analysis of the Financial Markets: A Comprehensive Guide To Trading Methods And Applications. New York: New York Institute of Finance.
Park, C. H., & Irwin, S. H. (2007). What Do We Know About the Profitability of Technical Analysis? Journal of Economic Surveys, 21(4), pp. 786-826. doi:10.1111/j.1467-6419.2007.00519.x
Prechelt, L. (2012). Early Stopping - But When? In G. Montavon, G. B. Orr, & K.-R. Müller, Neural Networks: Tricks of the Trade. Berlin: Springer.
Record, N. (2003). Currency Overlay. West Sussex: John Wiley & Sons.
Russell, S., & Norvig, P. (2009). Artificial Intelligence: A Modern Approach. Essex: Pearson.
Saunders, M., Thornhill, A., & Lewis, P. (2009). Research Methods for Business Students. New York: Prentice Hall.
Sermpinis, G., Laws, J., Karathanasopoulos, A., & Dunis, C. (2012). Forecasting and Trading the EUR/USD Exchange Rate with Gene Expression and Psi Sigma Neural Networks. Expert Systems with Applications, 39(10), pp. 8865-8877. doi:10.1016/j.eswa.2012.02.022
Sermpinis, G., Stasinakis, C., & Dunis, C. (2014). Stochastic and Genetic Neural Network Combinations in Trading and Hybrid Time-Varying Leverage Effects. Journal of International Financial Markets, Institutions & Money, 30, pp. 21-54. doi:10.1016/j.intfin.2014.01.006
Sermpinis, G., Theofilatos, K., Karathanasopoulos, A., Georgopoulos, E. F., & Dunis, C. (2013). Forecasting Foreign Exchange Rates with Adaptive Neural Networks using Radial-Basis Functions and Particle Swarm Optimization. European Journal of Operational Research, 225(3), pp. 528-540. doi:10.1016/j.ejor.2012.10.020
Shin, Y., & Ghosh, J. (1991). The Pi-Sigma Network: An Efficient Higher-Order Neural Network for Pattern Classification and Function Approximation. Seattle International Joint Conference on Neural Networks. doi:10.1109/IJCNN.1991.155142
Tenti, P. (1996). Forecasting Foreign Exchange Rates using Recurrent Neural Networks. Applied Artificial Intelligence, 10(6), pp. 567-582. doi:10.1080/088395196118434
Vecci, L., Piazza, F., & Uncini, A. (1998). Learning and Approximation Capabilities of Adaptive Spline Activation Function Neural Networks. Neural Networks, 11(2), pp. 259-270. doi:10.1016/S0893-6080(97)00118-4
Zhang, G., Hu, M. Y., Patuwo, E., & Indro, D. C. (1999). Artificial Neural Networks in Bankruptcy Prediction: General Framework and Cross-Validation Analysis. European Journal of Operational Research, 116(1), pp. 16-32. doi:10.1016/S0377-2217(98)00051-4