以深度學習建構股價預測模型:以台灣股票市場為例

Constructing stock price forecast model with deep learning: Evidence from Taiwan Stock Market

林逸青(Yi-Ching Lin)

指導教授 ： 謝孟芬

共同指導教授 ： 徐旺興(Wang-Hsing Hsu)

國立臺中科技大學/商學院/財務金融研究所碩士班/碩士(2019年)

摘要

機器學習與深度學習模型近年來在巨量資料分析和科技金融方面取得了顯著的成效。對於需要收集大量數據、分析大數據的時代，使用深度學習處理資料及分析數據已是一種優勢。本文應用了深度學習在股票市場市進行分析與預測，此方法的應用能從大量的原始數據中提取特徵而不仰賴於預測模型的先預測模式。這使的深度學習對股票市場預測具有潛在的預期性，本研究使用了深度神經網絡模型中的長期短期神經記憶網路模型(Long Short Term Memory Network, LSTM)去對2000-2018年間52個金融相關指標做分析，本篇與過往文獻最大的差異在於過往文獻大多研究針對於股票市場隔日或未來的漲跌趨勢做預測，而本文更進一步的探討到了股價指數漲跌指數的預測。本篇研究結果表明，深度學習中的長期短期神經記憶網路模型可以有效地用於台灣股票市場的預測。

關鍵字

深度學習 ； 大數據 ； 股票市場預測 ； 多層神經網絡

並列摘要

Machine learning and deep learning have achieved remarkable results in big data analysis and Fintech in recent years. For the era of collecting mass data and analyzing big data, using deep learning to process data and analyze data is an advantage.This paper applies deep learning that can extract features from a large of raw data without relying on the predictive model's pre-prediction mode to analyzes and forecasting in stock market. Such an architecture makes deep learning potentially predictive of stock market forecasts. This study used the Long Short Term Memory Network model (LSTM) to analyzing Taiwan stock price that data include 52 financial related indicators between 2000 and 2018.The biggest difference between this study and the past literature that most of the literature studies are aimed at forecasting the trend of stock market on the next day and this study further explores the forecast of the stock price index. The results of this paper show that the LSTM in deep learning can be effectively used in the forecast of Taiwan's stock market.

並列關鍵字

Deep Learning ； Big Data ； Stock market prediction ； Multilayer neural network

參考文獻

Andrew, A., and Bekaert, G. (2006). Stock return predictability: Is it there? Review of Financial studies, Vol.20, 651-707.

Arévalo, A., Niño, J., Hernández, G., and Sandoval, J. (2016). High-frequency trading strategy based on deep neural networks. In International conference on intelligent computing, 424-436.

Blau, B. M. (2018). Exchange rate volatility and the stability of stock prices. International Review of Economics & Finance, Vol 58, 299-311

Campbell, J.Y., and Thompson, S. B. (2008). Predicting excess stock returns out of sample: Can anything beat the historical average? Review of Financial Studies, Vol 21, 1509 - 1531

Cao, J., Li, Z., and Li, J. (2019). Financial time series forecasting model based on CEEMDAN and LSTM, Physica A:Statistical Mechanics and its Applications, Vol.519 ,127-139