Diagnosing the prognosticative potential of technical analysis versus fundamental analysis in predicting stock markets: an empirical investigation using artificial neural networks

Purpose The decision to select between technical analysis and fundamental analysis has long been a matter of great concern among researchers and practitioners, driven by the goal of making accurate stock market predictions. To address this, this paper aims to compare the relative predictive potential of technical analysis against fundamental analysis under a standardised forecasting framework to empirically investigate the superiority of technical indicators over macroeconomic variables for forecasting stock indices. Design/methodology/approach To attain the objective, the present research: a) considered two distinct input datasets – TI dataset (technical indicators) and MAC dataset (macroeconomic indicators); b) employed artificial neural networks (ANNs) to develop prediction models, namely, ANN-TI and ANN-MAC, corresponding to both datasets and Grid Search for selecting their optimal parameters; and c) compared the out-of-sample forecasting performance of the ANN-TI against that of ANN-MAC for forecasting four world major stock indices – NSE Nifty 50, CSI 300, FTSE 250, and NASDAQ. Moreover, the SHAP (SHapley Additive exPlanations) technique is utilised for interpretability of feature significance. Findings The findings evidenced the superior prognosticative potential of technical indicators over macroeconomic indicators for predicting stock market indices owing to significantly better performance of ANN-TI than ANN-MAC in terms of relatively lower values of loss metrics. Originality/value To the best of the authors’ knowledge, the current research is the first that empirically evaluates the technical analysis against fundamental analysis with the lens of its relative predictive potential for stock indices under standardised conditions. Further, it expands the capacity of researchers and practitioners for informed and strategical decision-making by promoting machine learning (ML)-powered technical analysis and ML-powered fundamental analysis.