A Comparative Study of Methods Used in the Analysis and Prediction of Financial Data

The main goal of this paper is to present a complete pipeline solution to the problem of forecasting financial data - the closing price of two stock indices. Firstly, we explore wavelet decomposition as a powerful method for reducing noise. In the second step we analyze several approaches for feature extraction. For the final phase we consider two prediction models based on long short-term memory cells -single layer and stacked-, together with a deep feed-forward neural network architecture. Our focus is on the second and third pipeline stages, for which we compare and discuss different strategies. We are concerned with emphasizing the differences in performance for recurrent and feed-forward neural networks respectively. Also, we analyze the effect of adding a higher order feature extraction phase before prediction, and compare stacked auto-encoders to principal component analysis in this sense.

[1]  Jinsong Leng,et al.  Modelling and Analysis on Noisy Financial Time Series , 2014 .

[2]  Stéphane Mallat,et al.  A Theory for Multiresolution Signal Decomposition: The Wavelet Representation , 1989, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Li Tang,et al.  PANK-A financial time series prediction model integrating principal component analysis, affinity propagation clustering and nested k-nearest neighbor regression , 2018 .

[4]  Ahmad Sharmi Abdullah,et al.  A Stock Market Trading System Using Deep Neural Network , 2017, AsiaSim 2017.

[5]  Stephen J. Wright,et al.  Hogwild: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent , 2011, NIPS.

[6]  Victor Chang,et al.  An innovative neural network approach for stock market prediction , 2018, The Journal of Supercomputing.

[7]  Vincent Tam,et al.  A comparative study of a recurrent neural network and support vector machine for predicting price movements of stocks of different volatilites , 2017, 2017 IEEE Symposium Series on Computational Intelligence (SSCI).

[8]  Yulei Rao,et al.  A deep learning framework for financial time series using stacked autoencoders and long-short term memory , 2017, PloS one.

[9]  Jacinta Chan Phooi M’ng,et al.  Forecasting East Asian Indices Futures via a Novel Hybrid of Wavelet-PCA Denoising and Artificial Neural Network Models , 2016, PLoS ONE.

[10]  Swathi Nibhanupudi.  SIGNAL DENOISING USING WAVELETS , 2003 .

[11]  Yuk Ying Chung,et al.  A novel stacked denoising autoencoder with swarm intelligence optimization for stock index prediction , 2017, 2017 International Joint Conference on Neural Networks (IJCNN).