Estimating Future Stock Return for the pharmaceutical sector using Bayesian Vector Auto regression modeling and artificial Intelligence

The process of forecasting stock returns in the pharmaceutical industry is quite complex and risky because of such issues as the decision making process by regulatory authorities, the results of the research and development, and the variability in the macroeconomic conditions. The paper constructs a decision-support model by evaluating two modern forecasting models namely Bayesian Vector Auto regression (BVAR) model and Artificial Neural Network- Radial Basis Function (ANN-RBF) to predict stock returns of three stock exchange listed pharmaceutical companies based on Amman Stock Exchange. The given framework is based on the methodology where the two models are trained using time series data during the time frame 20192024 and the predictions are made during 20252030. The results show that BVAR, which includes the incorporation of time-dependences of macroeconomic variables, gives better forecasts of returns that possess the features of stability and linearity but ANN-RBF has better forecasts of returns that have volatility and nonlinearity. These findings prompt the need to consider such characteristics in the forecasting studies in order to produce meaningful results. In addition to methodological inputs, the research provides feasible guidelines to investors, corporate leaders, and other policy makers that seek to cope with the risk, make long-term investments, and guide financial decision-making in the pharmaceutical firms. Also, the research incorporates methods of applied mathematics, artificial intelligence, and finance, which supports the importance of forecasting models as evidence-based decision-making tools in conditions with high risks.