PREDIKSI NILAI ASET MENGGUNAKAN MODEL GEOMETRIC BROWNIAN MOTION DAN MODEL VARIANCE GAMMA

Abstract

The basic assumption in the Black-Scholes-Merton model is that assets that generate returns are normally distributed. Asset price movements fluctuate greatly so the data is not normally distributed. This paper proposes a forecasting method using the Variance Gamma (VG) model. The return data of traded assets in Indonesia shows excess kurtosis and tails in the return distribution so the performance of the geometric Brownian motion (GBM) model is less appropriate for use. One of the appropriate models for data that shows excess kurtosis and tails is the VG model. The VG model has three parameters to control volatility, skewness, and kurtosis. We compare the results of this study with the geometric Brownian motion (GBM) model. The accuracy of the model uses the Mean Absolute Percentage Error (MAPE). In this study, the VG asset model has a MAPE value of 3.08%, while the GBM model has a MAPE value of 13.20%. These results conclude that the VG asset model is more accurate in predicting asset prices compared to the GBM asset model.