<p>Macroeconomic policies rely on the ability to forecast future values of macroeconomic and financial variables, which traditional econometric models often struggle to capture effectively. To address this challenge, the current study employs artificial neural network optimised by hybrid firefly and particle swarm optimisation (ANN-HFPSO) to predict the financial stress index (FSI) of the Eurozone for a sample period from January 1995 to August 2023. The ANN-HFPSO model is compared with autoregressive model (AR), univariate generalised autoregressive conditional heteroskedasticity (GARCH), multivariate GARCH (M-GARCH), artificial neural network optimised by firefly algorithm (ANN-FA) and artificial neural network optimised by particle swarm optimisation (ANN-PSO). Accuracy measures are used to evaluate the models’ performance, while Taylor and contour diagrams are employed to visualise the metrics. The findings show that the ANN-HFPSO model significantly improves the predictive accuracy of AR by approximately 38.47% for the training sample and around 39.55% for the testing sample. ANN-FA is the second-best model in predicting FSI in the Eurozone, as it improves the accuracy of AR by 33.49% and 36.38% for the training and testing samples respectively. The findings can be used by policymakers in designing macroeconomic policy and by investors in making investment decisions.</p>