基於機器學習與AH溢價指數漲跌的預測

Prediction of the AH Premium Index Based on Machine Learning

米麗曼(Li-Man Mi)

指導教授 ： 呂育道

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201801174

摘要

機器學習為達到人工智慧之方法，在股票預測中，機器學習方法具有非線性逼近能力、自學性與複雜系統控制能力等優勢，為此本文使用前饋神經網路模型與支持向量機模型。AH溢價指數是同時於A股市場和H股市場上市的中國大陸公司的A股相對H股的溢價或折讓，既可反映A股市場也可反映H股市場的上市公司表現。本研究利用機器學習與計量經濟模型方法，以AH溢價指數漲跌為研究標的。AH溢價指數漲跌之預測實證結果顯示，機器學習方法比計量經濟模型方法預測準確度更高。根據1,465種模型組合對比分析顯示，多層前饋神經網路模型以70%準確率略高於支持向量機模型67%準確率與計量經濟模型57%準確率。此外，本文還針對回溯天數、數據選取、資料前置處理、隱藏層數、隱藏節點數、激活函數、損失函數、核函數對模型預測準確率影響進行探討。

關鍵字

機器學習 ； 前饋神經網路 ； 支持向量機 ； 計量經濟模型 ； AH溢價指數 ； 股票指數預測

並列摘要

Machine learning is an approach of artificial intelligence. It exhibits nonlinearity in approximation, self-learning and the capability in solving complex problems. The AH Premium Index is designed to reflect the absolute price premium (or discount) of AH Companies. This thesis aims to predict AH premium index’s rise or fall the next day with multilayer feedforward neural networks (MFNNs), support vector machines (SVMs), and econometric models. It varies activation functions, loss functions, kernel functions, moving window, hidden layers, input variables and data preprocessing for controlling purpose. Among 1,465 parameter combinations, the MFNN model is found to have the highest performance (70%) followed by the SVM model (67%) and the econometric model (57%). The result indicates that machine learning techniques are better than econometric models in predicting AH premium index’s rise or fall the next day.

並列關鍵字

Machine learning ； Multilayer feedforward neural network ； Support vector machine ； Econometric model ； AH premium index ； Stock market index prediction

參考文獻

Altman, E. I., Marco, G., & Varetto, F. (1994). Corporate Distress Diagnosis: Comparisons Using Linear Discriminant Analysis and Neural Networks (the Italian Experience). Journal of Banking & Finance, 18(3), 505–529. https://doi.org/10.1016/0378-4266(94)90007-8

Bailey, W. (1994). Risk and Return on China’s New Stock Markets: Some Preliminary Evidence. Pacific-Basin Finance Journal, 2(2–3), 243–260. https://doi.org/10.1016/0927-538X(94)90019-1

Budhani, N., Jha, C. K., & Budhani, S. K. (2014). Prediction of Stock Market Using Artificial Neural Network. In 2014 International Conference of Soft Computing Techniques for Engineering and Technology (ICSCTET) (pp. 1–8). IEEE. https://doi.org/10.1109/ICSCTET.2015.7371196

Burrell, P. R., & Folarin, B. O. (1997). The Impact of Neural Networks in Finance. Neural Computing & Applications, 6(4), 193–200. https://doi.org/10.1007/BF01501506

Chakravarty, S., Sarkar, A., & Wu, L. (1998). Information Asymmetry, Market Segmentation and the Pricing of Cross-listed Shares: Theory and Evidence From Chinese A and B Shares. Journal of International Financial Markets, Institutions and Money, 8(3–4), 325–356. https://doi.org/10.1016/S1042-4431(98)00041-9