Modeling and Forecasting Stock Closing Prices: A case study of L’Oreal listed on the French stock exchange

Abstract

Introduction: Financial forecasting has long sought to model stock price dynamics through statistical and econometric approaches. This study focuses on predicting L'Oréal’s daily stock closing prices using a linear regression framework, grounded in the assumption that fundamental market indicators can effectively capture short-term price variations.
Objective: The aim is to evaluate the predictive capacity of a multivariate linear regression model based on fundamental variables—Open, High, Low prices, and trading Volume—over a 25-year historical dataset.
Method: Daily OHLCV data from 2000 to 2025 were obtained from Yahoo Finance. The dataset was divided into training (80 %) and testing (20 %) subsets. Model performance was evaluated using Mean Squared Error (MSE = 1,238) and the coefficient of determination (R² = 0,9987). Market regime shifts, such as the 2008 and 2020 financial crises, were included to assess model robustness.
Results: The model achieved high predictive accuracy with a Mean Absolute Percentage Error (MAPE = 0,55 %). Among predictors, High and Low prices were the most influential (β = 0,845 and 0,723, respectively), while Volume showed no statistical significance (p > 0,05). Residual analysis revealed minor deviations from normality but no signs of autocorrelation.
Conclusions: The results demonstrate that linear regression remains a valid and interpretable method for forecasting blue-chip stock prices. The model’s precision suggests potential for integration into algorithmic trading systems. However, incorporating volatility-based adjustments is recommended to enhance stability during market turbulence. Future studies could compare linear and nonlinear models across sectors to assess generalizability.