Change-point-based hybrid stochastic differential equation models: the case of WTI crude oil prices

This study models the fluctuations of West Texas Intermediate (WTI) crude oil prices between ۰۱.۰۳.۲۰۱۹ and ۱۳.۰۳.۲۰۲۳ using Geometric Brownian Motion (GBM) and Cox–Ingersoll–Ross (CIR) stochastic differential equations (SDEs). The model parameters are estimated using the Quasi-Maximum Likelihood Estimation (QMLE) method, and the compatibility of both models is validated through the chi-square goodness-of-fit test. Due to abrupt price changes, two significant change points (CPs), on May ۱۴, ۲۰۲۰, and February ۲۴, ۲۰۲۲, are estimated using the QMLE method. These CPs correspond to major global events, namely the COVID-۱۹ pandemic and Russia’s military intervention in Ukraine. The GBM and CIR models, redefined based on these CPs, are combined into a hybrid SDE model, which demonstrates superior performance compared to other models, as evaluated by MAPE and RMSE criteria. The hybrid SDE model more accurately predicts sudden price fluctuations, thereby improving the model’s accuracy. These findings indicate that CP estimation significantly improves the effectiveness of SDE models, enabling more reliable modeling of abrupt changes in economic time series data. Consequently, the integration of CPs contributes to improving the accuracy of economic forecasts and facilitates more reliable predictions of future market behavior, especially in response to geopolitical and economic disruptions

کلیدواژه ها:

Geometric Brownian motion ، change point estimation ، Cox–Ingersoll–Ross ، Hybrid model ، Stochastic differential equation