本研究主要為預測未來一個月的澳幣走勢，使用深度學習模型長短期記憶(Long short-term memory, LSTM)以及機器學習模型等演算法作為預測模型，透過匯率資料、總體經濟資料以及原物料資料作為主要變數，共365個變數，預測下個月是否澳幣為升值與貶值作為目標，來對澳幣匯率走勢做趨勢預測。而本研究利用LSTM演算法的特殊設計，可將過往的資訊傳輸至下一個時間訓練中，其模型設計較過往傳統的機器學習更加符合時間序列資料，而最終我們也發現LSTM在預測績效上較其他預測方法更為優越。另外本研究也納入了Shapley Additive Explanations value概念，過往在預測準確度與可解釋性之間，經常會因模型過於複雜而無從解釋，而這個概念能透過數學式子將各個特徵對於預測值的影響做衡量，將可幫助我們對於模型的了解，透過較複雜的模型更可以了解，影響澳洲匯率的主要特徵為哪些因子，更能讓我們在後續調整模型時，能有更明確的方向。

In this study, we use deep learning model, such as Long short-term memory (LSTM), and a machine learning model to predict the trend of the Australian dollar in the next month. Furthermore, we use exchange rate data, macroeconomic data and raw material data as the main variables, with total of 365 variables to predict whether the Australian dollar will appreciate or depreciate in the next month as the target to make a trend forecast of the Australian dollar. The special design of LSTM algorithm can transfer the past information to the next time training, and its model design is more consistent with the time series data than the traditional machine learning. In addition, we also find that LSTM outperforms other prediction methods in terms of prediction performance. Additionally, this study also includes the concept of SHAP (Shapley Additive Explanations) value. In previous studies, it was found that there is a substitution relationship between accuracy and interpretability, which is often unexplained due to the complexity of the model. SHAP value can measure the impact of each feature on the forecast value through mathematical equations, which will help us understand the model, and can understand which variables are the main features that affect the Australian exchange rate, so that we can have a clearer direction when adjusting the model later.

Table of Contents
論文審定書 i
摘要 ii
Abstract iii
Table of Contents v
1. Introduction 1
1.1 Study Background 1
1.2 Study Purpose 3
1.3 Study Process 4
2. Literature Review 5
2.1 Exchange Rate Trend Analysis 5
2.2 Machine learning 6
2.3 Shapley Additive Explanations (SHAP) 8
3. Data Acquisition and Preprocessing 10
3.1 Technical Indicators 13
3.2 Macroeconomic Indicators 28
4. Methodology and Results 32
4.1 Recurrent Neural Network (RNN) 32
4.2 Long Short-Term Memory (LSTM) 36
4.3 Support vector machine (SVM) 39
4.5 eXtreme Gradient Boosting (XGBoost) 40
4.6 Random Forest 42
4.7 SHAP Value 43
4.8 Model Framework 44
4.8.1 Machine Learning Model 44
4.8.2 LSTM 45
4.8.3 Target Variable and Trading Strategies 46
4.8.4 Model measurement 47
4.9 Results 48
4.10 SHAP 52
5. Conclusion 55
References 56

Archer, M. D. (2010). Getting Started in Currency Trading: Winning in Today's Forex Market (Getting Started In.....).
Breiman, L. (1996). Bagging predictors. Machine learning, 24(2), 123-140.
Breiman, L. (2001). Random forests. Machine learning, 45(1), 5-32.
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.
Freund, Y., Schapire, R., & Abe, N. (1999). A short introduction to boosting. Journal-Japanese Society For Artificial Intelligence, 14(771-780), 1612.
Hu, Z., Zhao, Y., & Khushi, M. (2021). A survey of forex and stock price prediction using deep learning. Applied System Innovation, 4(1), 9.
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in neural information processing systems, 30.
Murphy, J. J. (1999). Technical analysis of the financial markets: A comprehensive guide to trading methods and applications. Penguin.
Özorhan, M. O., Toroslu, İ. H., & Şehitoğlu, O. T. (2019). Short-term trend prediction in financial time series data. Knowledge and Information Systems, 61(1), 397-429.
Wolpert, D. H. (1992). Stacked generalization. Neural networks, 5(2), 241-259.
Yıldırım, D. C., Toroslu, I. H., & Fiore, U. (2021). Forecasting directional movement of Forex data using LSTM with technical and macroeconomic indicators. Financial Innovation, 7(1), 1-36.
Yong, Y. L., Ngo, D. C., & Lee, Y. (2015). Technical indicators for forex forecasting: a preliminary study. International Conference in Swarm Intelligence.
Zhang, Y., & Hamori, S. (2020). The predictability of the exchange rate when combining machine learning and fundamental models. Journal of Risk and Financial Management, 13(3), 48.