Entropy corrected geometric Brownian motion

Rishabh Gupta,Ewa A. Drzazga-Szczȩśniak,Sabre Kais,Dominik Szczȩśniak
2024-03-16
Abstract:The geometric Brownian motion (GBM) is widely employed for modeling stochastic processes, yet its solutions are characterized by the log-normal distribution. This comprises predictive capabilities of GBM mainly in terms of forecasting applications. Here, entropy corrections to GBM are proposed to go beyond log-normality restrictions and better account for intricacies of real systems. It is shown that GBM solutions can be effectively refined by arguing that entropy is reduced when deterministic content of considered data increases. Notable improvements over conventional GBM are observed for several cases of non-log-normal distributions, ranging from a dice roll experiment to real world data.
Data Analysis, Statistics and Probability,Statistical Finance
What problem does this paper attempt to address?