Article Preview
Top1. Introduction
The ability of artificial neural network (ANN) in forecasting the daily market stock exchange rate was investigated (Moghaddam et al., 2016; Hedayati et al., 2016). An international exchange market is one of the most invested markets in the world with an average daily trade volume of $1.8 trillion (Ahmed and Hokey, 2009). Researchers have shown that many commonly cited signals have had very weak and erratic correlations with actual subsequent returns, even at long investment horizons (Joseph et al., 2012). Large changes have taken place over the while in the market places, involving the use of significant communication and commercial platforms that empowered the number of investors entering the markets. Traditional capital market theory has also changed, and methods of financial analysis have improved (Laurent, 1979). Stock-return forecasting has drawn the attention of researchers for several years and typically conveys an idea that essential information generally available in the past has some predictive connections to future stock returns or indices. The samples of such information include: economic variables, exchange rates, industry and sector-specific information, and individual corporates financial statements. This is opposed to the general ideal of the efficient market hypothesis which states that all available information affecting the current stock value is constituted by the market before the general market builds trades based on it. Hence, it is hard to forecast future returns for a reason that stocks already reflect all known data. This is still an empirical problem because there is a contradictory evidence that markets are not totally efficient, and that is possible to guess the future returns with results that are better than random by means of generally available information such as time-series data on financial and economic variables (David, 1988).These studies assumed that variables e.g. interest rates, monetary-growth rates, changes in industrial production, and inflation rates are statistically necessary for predicting some of the stock returns. However, most of the studies simply mentioned that effort to capture the conjunction between the current information and also the stock returns depends on simple linear regression assumptions despite the fact that there's no proof for the linear connection between the stock returns and the monetary and economic variables. Since there existing significant residual variance of the actual stock return from the prediction of the regression function, it is possible that nonlinear models should be used to demonstrate this residual variance and turn out additional reliable predictions of the stock value movements. The stock probably leads to a downturn in the housing market because business’ share prices have engaged very high relatives to their outcomes. Analysts who made that point of view in the past earned high price ratios have usually followed by not really fast growth in stock prices (Pu, 2000). Neural network is widely used to overcome such problems as a nonlinear modelling technique. Neural network offers a novel technique that does not require pre-specification during the modelling process since they independently learn the inherent relationship between the variables (Amin et al, 2018). This is particularly useful in protection investment and other financial fields where much was proposed and little was known about the type of the processes specifying asset prices. Neural networks propose different infrastructure models and learning platforms, in which, classification and level estimation in both multi-layer feed-forward neural networks and recurrent neural network were briefly reviewed. As a consequence, as shown in this study, using different learning models in artificial neural network would forecast time-series of future value more efficiently. Current studies and previous studies have had forecasted markets in specific periods or forecasting data in specific markets.
Figure 1.
Time Series of the Dow Jones Industrial Average (DJIA)