Comparative Analysis of Geometric Brownian Motion for Stock Price Forecasting across US and Emerging Market Stocks

Abstract

This study evaluates the performance of the Geometric Brownian Motion (GBM) model for short-term stock price forecasting across developed and emerging markets. Using a dataset of 31 assets, including major U.S. equities and stocks from India, Brazil, South Korea, and Southeast Asia, we estimate drift and volatility parameters and generate forecasts through Monte Carlo simulation. The predictive accuracy of GBM is assessed using Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE), and correlation measures, and is benchmarked against a random walk model.

The results suggest that GBM provides reasonable forecasts for relatively stable assets, where price dynamics are less affected by sudden volatility shifts. However, its performance deteriorates in highly volatile or noisy market conditions, where the model fails to capture clustering effects and abrupt price movements. In several cases, GBM does not outperform the random walk benchmark, raising questions about its practical effectiveness as a forecasting tool.

Overall, the findings indicate that while GBM remains a useful baseline model due to its simplicity and analytical tractability, its applicability is limited in environments characterized by high volatility and structural instability. These results highlight the importance of model selection and suggest that more flexible approaches may be required for accurate forecasting in complex market conditions.