Short-Term Stock Price Forecasting Using Geometric Brownian Motion

In these days of big data, machine learning, and AI, many researchers are showing growing interest in sophisticated models for stock price prediction or to refine basic models of stock dynamics. Reference [1] takes the opposite approach. It uses a classical model for stock price dynamics—the Geometric Brownian Motion (GBM)—and examines whether it can still be used to forecast stock prices. Specifically, the study applies four volatility measures to large-cap stocks in an emerging market to estimate volatility, then incorporates these estimates into the GBM to generate price forecasts.

The authors pointed out,

One effective method for forecasting short-term investment involves models like GBM. This study specifically applied GBM over a two-week period, focusing on the crucial aspect of volatility measurement. By examining four distinct volatility measurements, simple volatility (S), log volatility (L), high-low volatility (HL) and high-low-closed volatility (HLC), the findings indicate that simple volatility (S) yielded the closest forecast to actual stock prices, as evidenced in Table 3 and Figure 1.

Furthermore, the overall high accuracy of the forecasts generated by GBM, with most MSE, MAPE, and MAD values falling below 10% as shown in Table 4, confirms its potential as a valuable tool for short-term stock market forecasting. These results suggest that for investors and analysts focusing on short-term investment in the Malaysian stock market, utilizing GBM with a simple volatility measurement can provide a reasonably accurate basis for making timely trading decisions.

In short, and somewhat surprisingly, the simple GBM model combined with a basic volatility measure delivers the most accurate forecasts over short horizons of up to two weeks.

We note the following,

1. The forecast accuracy is limited to the short term,
2. Although four volatility measures are tested, the simplest performs best,
3. The analysis is conducted in an emerging market, and
4. The sample size is small.

Overall, this study runs counter to the current trend and suggests that simple models—both in volatility measurement and price dynamics—can still be effective. This is an interesting study and worth further examination.

Let us know what you think in the comments below or in the discussion forum.

References

[1]  FS Fauzi, SM Sahrudin, NA Abdullah, SN Zainol Abidin, SM Md Zain, Forecasting stock market prices using Geometric Brownian Motion by applying the Optimal Volatility measurement, Mathematical Sciences and Informatics Journal (2025) Vol. 6, No. 2

Post Source Here: Short-Term Stock Price Forecasting Using Geometric Brownian Motion

source https://harbourfronts.com/short-term-stock-price-forecasting-using-geometric-brownian-motion/