A deep learning framework for financial time series using stacked autoencoders and long-short term memory
By Wei Bao, Jun Yue, Yulei Rao. 2017
1,013 citations recorded by the source corpus.
The application of deep learning approaches to finance has received a great deal of attention from both investors and researchers. This study presents a novel deep learning framework where wavelet transforms (WT), stacked autoencoders (SAEs) and long-short term memory (LSTM) are combined for stock price forecasting. The SAEs for hierarchically extracted deep features is introduced into stock price forecasting for the first time. The deep learning framework comprises three stages. First, the stock price time series is decomposed by WT to eliminate noise. Second, SAEs is applied to generate deep high-level features for predicting the stock price. Third, high-level denoising features are fed into LSTM to forecast the next day’s closing price. Six market indices and their corresponding index futures are chosen to examine the performance of the proposed model. Results show that the proposed model outperforms other similar models in both predictive accuracy and profitability performance.
Subfields: Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Management Science and Operations Research, Time-Series & Forecasting
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Xiangnan He
- Xiang Wang
- Tat-seng Chua
- Fuli Feng
- Weiwei Jiang
- M. Ugur Gudelek
- Omer Berat Sezer
- A. Ozbayoglu
In Electrical and Electronic Engineering
- Weiwei Jiang
- A. Ozbayoglu
- M. Ugur Gudelek
- Omer Berat Sezer
- J. Joshua Yang
- Qiangfei Xia
- John Paul Strachan
- Zhongrui Wang
In Management Science and Operations Research
- Weiwei Jiang
- M. Ugur Gudelek
- Omer Berat Sezer
- A. Ozbayoglu
- Xiangnan He
- Xiang Wang
- Tat-seng Chua
- Stefan Zohren
In Time-Series & Forecasting
- Weiwei Jiang
- M. Ugur Gudelek
- Omer Berat Sezer
- A. Ozbayoglu
- Stefan Zohren
- Bryan Lim
- Stephen J. Roberts
- El-Sayed M. El-Alfy
Citations in the corpus
- Financial Time Series Forecasting with Deep Learning : A Systematic Literature Review: 2005-2019 (2019, methodology): “The authors of [123] used stacked AEs to generate deep features using OCHL of the stock prices, technical indicators and macroeconomic conditions to feed to LSTM to predict the future stock prices.”
- Enhancing Time-Series Momentum Strategies Using Deep Neural Networks (2019, background): “While numerous papers have investigated the use of machine learning for financial time series prediction, they typically focus on casting the underlying prediction problem as a standard regression or classification task [23, 24, 25, 12, 26, 19, 27] – with regression models forecasting expected retur”
- Deep Learning for Financial Applications : A Survey (2020, methodology): “[36] used technical indicators as the input into Wavelet Transforms (WT), LSTM and Stacked Autoencoders (SAEs) for the forecasting of stock prices.”
- Enhancing the momentum strategy through deep regression (2019, methodology): “Notably, Bao et al. (2017) used deep neural networks to make stock price forecasting and Heaton et al.”
- DeepLOB: Deep Convolutional Neural Networks for Limit Order Books (2018, background): “markets, the LSTM has been popular in recent years, [20], [28], [42]–[47] all utilising LSTMs to analyse financial data.”
- Evaluation of bidirectional LSTM for short-and long-term stock market prediction (2018, methodology): “[17] to propose a prediction approach using LSTM.”
- Unsupervised Pre-training of a Deep LSTM-based Stacked Autoencoder for Multivariate Time Series Forecasting Problems (2019, background): “In such a case, a lower error reconstruction will be obtained, and thereby, better generalization is expected 53 .”
- Temporal Relational Ranking for Stock Prediction (2018, methodology): “• LSTM [4]: This method is the vanilla LSTM, which operates on the sequential data including closing prices and moving averages of 5, 10, 20, and 30 days, to obtain a sequential embedding; and then a FC layer is used to make prediction of the return ratio.”
- Applications of deep learning in stock market prediction: recent progress (2020, methodology): “(2018) LSTM WT+SAEs+LSTM (Bao et al., 2017) Xu & Cohen (2018) StockNet Random Guess, ARIMA, RF, TSLDA (Nguyen & Shirai, 2015), HAN (Hu et al.”
- CNNpred: CNN-based stock market prediction using a diverse set of variables (2019, background): “, 2015; Fischer & Krauss, 2018), autoencoder (AE) (Bao et al., 2017) and convolutional neural network (CNN) (Gunduz et al.”
- Stock Market Embedding and Prediction: A Deep Learning Method (2018, background): “Bao, Yue & Rao (2017) [18] proposes a prediction framework for financial time series data that integrates Wavelet transformation, Stacked Autoencoders, and LSTM.”
- Bankruptcy prediction using imaged financial ratios and convolutional neural networks (2019, background): “Since then, non-linear prediction models using15 logistic regression (Ohlson, 1980), neural networks (Odom and Sharda, 1990; Brédart, 2014), support vector machines (SVM) (Shin, Lee, and Kim, 2005; Li and Sun, 2009), and AdaBoost (Alfalo, Garćıa, Gámez, and Elizondo, 2008; Ramakrishnan, Mirzaei,…”