NNARX

It is well documented that using linear models to forecast plenty of financial observations due to their nonlinearity is not satisfactory. Therefore, in this paper, the technical analysis indicators are forecasted using Neural Network Auto-Regressive model with eXogenous inputs (NNARX). Then the effect of different factors (economic, systematic risk, company's properties and corporate governance) on their forecasting error (eRSI, eMA1, eMA2 and eMACD) was investigated. For this purpose, required data were collected using the removal sampling method for 323 companies listed on the Tehran Stock Exchange from 2014-2020. In addition, the mean absolute percentage error (MAPE) was applied to measure the error of forecasting technical analysis indicators. NNARX and dynamic panel data models (GMM) were used to study the effective factors on the error of forecasting technical analysis indicators. Results indicated that the error of forecasting technical analysis indicators is less than 0.1 and has sound accuracy. Also, the company's size and corporate governance indicators didn't significantly affect the error of forecasting technical analysis indicators. In addition, financial leverage doesn't significantly affect eRSI and eMACD but has a significant inverse effect on eMA1 and eMA2. On the other hand, return on assets has a significant inverse effect on eRSI, eMA1, eMA2 and eMACD. Also, economic recession and prosperity, inflation fluctuations, exchange rate fluctuations and systemic risk have a significant positive effect on eRSI, eMA1, eMA2 and eMACD.