Neural networks reveal the benefit of requiring less formal statistical training owing to their capability of detecting complex nonlinear relationships between dependent and independent variables implicitly. Machine learning, artificial neural networks (ANN) and deep learning, on the other hand, show their upper hands in forecasting and prediction processes involved in various domains, including the analyses concerning financial indexes among other ones. Besides these advances and applications, time series models based on financial data expressed by financial theories may form as a basis for forecasting a series of data. When theories may not directly be applicable to predict the market values with external impacts, ANNs can be applied as a prediction tool for forecasting the market value series. Due to its robustness as a statistical means for examining sequential data over time, time series analyses examine financial stock market indices by identifying trends, predictive details and changing patterns toward accurate and efficient analyses in dynamic and nonlinear markets. Accordingly, the current study aims at forecasting based on the daily values of FCHI (France) and GSPC (India) indexes. Hurst exponent is applied to the dataset consisting of the Open, High, Low and Close values of the FCHI and GSPC indexes. Modeling for the purpose of the study is carried out by feed-forward backpropagation (FFBP) algorithm which is among the ANNs. The forecasting accuracy obtained from the modeling of the index values based on the past daily close attributes of 20 days is compared with respect to Mean Squared Error, and the results have validated the forecasting of indexes. As the steps, Hurst exponent is initially applied to the daily values (Open, High, Low, Close) of FCHI and GSPC indexes, leading to the generation of the dataset with more attributes. As the following step, the dataset with more attributes consisting of the FCHI and GSPC daily values are split into training dataset and test dataset for forecasting the indexes. Subsequently, the model constructed is used to compare the two datasets. The forecasting accuracy of the index values’ modeling based on the daily close values of 20 days is compared by Mean Squared Error. Lastly, modeling through the FFBP algorithm application to the new dataset generated and the FCHI and GSPC indexes enables the comparison and performing of forecasting with regard to these indexes. In addition, interpolation method is employed for managing missing data in the dataset. This approach is intended to a new mode of scheme and implementation tactics in an extended form to reveal the interaction between time series analysis, predictive modeling and indices. Thus, the models to be constructed can be evaluated regarding comprehensive outlooks that are concerned with decision-making, investment strategies’ optimization and risk mitigation in various domains with accurate, reliable and data-driven aspects.