Forecasting Stock Returns Using State-Space Models and Time-Varying Parameters

Citation: Abdelmadjid Laouedj, "Forecasting Stock Returns Using State-Space Models and Time-Varying Parameters", Universal Library of Business and Economics, Volume 03, Issue 01.

Copyright: This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

The article examines the application of state-space models with time-varying parameters for forecasting stock returns in a non-stationary, regime-switching financial environment. The relevance of the study stems from the fact that observed returns are a hybrid of rare structural breaks and day-to-day noise, which results in traditional regression schemes with fixed coefficients rapidly deteriorating out of sample, especially in sectors characterised by high regulatory and technological uncertainty, such as pharmaceuticals. The aim of the study is to provide a conceptual and methodological justification for a forecasting approach that, from the outset, treats parameter instability as a norm rather than an exception. The scientific novelty lies in a coherent specification of a state-space model in which expected returns, factor loadings and, in extensions, volatility and market regimes are treated as latent states evolving according to probabilistic laws and estimated via recursive filtering, with explicit control of uncertainty, coefficient shrinkage, and time-based validation of the end-to-end pipeline under transaction cost constraints. The main conclusions demonstrate that this approach shifts the focus from guessing a number to constructing an adaptive, properly calibrated representation of risk premia and the range of outcomes, where the key resource is not a universal formula but a procedure for robustly updating inferences in a changing environment. The article is intended for researchers and practitioners in financial markets, as well as pharmaceutical managers and participants in continuing education programmes who rely on return forecasts when assessing risk and the robustness of decisions.

Keywords: Stock Return Forecasting, State-Space Models, Time-Varying Parameters, Kalman Filter.

https://doi.org/10.70315/uloap.ulbec.2026.0301008