Notice of Retraction Financial time series prediction based on Echo State Network

Notice of Retraction After careful and considered review of the content of this paper by a duly constituted expert committee, this paper has been found to be in violation of IEEE's Publication Principles. We hereby retract the content of this paper. Reasonable effort should be made to remove all past references to this paper. The presenting author of this paper has the option to appeal this decision by contacting TPII@ieee.org. Neural networks have been popular in time series prediction in financial area for their advantages in handling nonlinear systems. However, conventional neural networks are confronted with the problem to include time property and the risk of local convergence. This paper presents a study of using a novel recurrent neural network-Echo State Network (ESN) to predict next value in a financial time series. In order to prove the predictability of financial time series, we attain the Hurst exponent through rescaled range (R/S) analysis first. In feature selection, technical analysis is utilized to extract underlying information in the time series and principal component analysis (PCA) helps to filter noise and obtain faithful representation of principle components. The data of six major stock indices in the world, DJIA, S&P 500, NASDAQ, HSI, FTSE 100 and NIKKEI 225, are employed to test our model. For comparison purpose, other neural networks such as back-propagation network and Elman network are also considered. Experiment results demonstrate that ESN performs much better than other neural networks in forecasting the next closing price and the combination of PCA with technical analysis improves the prediction accuracy a little comparing with a model using only raw price data. Our preliminary study also suggests that ESN is effective in short-term financial time series prediction and worth further investigation.