A Statistical Mechanics Approach to Financial Engineering: GBM and the Black-Scholes Framework

I. Introduction

The intersection of physics and finance, often termed "Econophysics," seeks to apply the laws of statistical mechanics to economic systems. Just as a physicist cannot predict the motion of a single molecule but can describe the thermodynamic properties of a gas, financial modeling aims to describe the probabilistic distribution of asset prices rather than predicting exact values.

This study focuses on NVIDIA (NVDA), a high-volatility asset, to demonstrate how randomness (entropy) in the market can be modeled using the same mathematical tools used to describe diffusion processes in physics.

II. Theoretical Background

2.1 Brownian Motion in Physics

In 1905, Albert Einstein formalized the mathematical description of Brownian motion, explaining that the random movement of pollen grains in water was caused by collisions with invisible water molecules. He proved that the mean squared displacement of a particle is proportional to time (t):

x^2 = 2Dt

where D is the diffusion coefficient.

2.2 Geometric Brownian Motion (GBM) in Finance In the 1950s, M.F.M. Osborne extended this concept to finance, proposing that stock prices follow a "log-normal" random walk. Unlike physical particles, stock prices cannot be negative. Therefore, we model the logarithm of returns, ensuring that prices remain strictly positive. This model, known as Geometric Brownian Motion (GBM), serves as the foundation for the Nobel Prize-winning Black-Scholes equation.

III. Mathematical Framework

3.1 Stochastic Differential Equations (SDE)

The evolution of a stock price $S_t$ is described by the following Stochastic Differential Equation:

dS_t = μS_t dt + σS_t dW_t

μ(Drift): The expected rate of return (analogous to velocity).

• σ(Volatility): The magnitude of random fluctuations (analogous to temperature/noise).

• dW_t: A Wiener process, representing random market shocks, where $dW_t \sim N(0, dt)$.

3.2 Itô’s Lemma and Analytical Solution

Since  is not differentiable in the classical sense, we apply Itô Calculus (the stochastic equivalent of the chain rule). The analytical solution to the SDE is:

S(t)=S_0 exp[(μ-1/2 σ^2)t+σW_t]

This equation reveals that the "most likely" price path is slightly lower than the expected drift due to the term-1/2 〖σ^2〗_, a geometric drag effect caused by volatility.

IV. Computational Implementation

4.1 Monte Carlo Simulation Method

To solve this equation numerically, we employ the Monte Carlo method. By generating thousands of random paths (parallel universes), we can construct a probability distribution for the stock's future price.

4.2 Algorithm Design (Box-Muller Transform)

Standard computer random number generators produce uniform distributions. To simulate natural market shocks (Gaussian distribution), we implement the Box-Muller transform, which converts two uniform random numbers (u, v) into a standard normal variable (Z):

Z=√(-2ln(u)) cos(2πv)

4.3 Implementation in C

The simulation is implemented in C for high-performance computing, mimicking the low-latency requirements of algorithmic trading systems.

/* * Stochastic Asset Pricing Model using Monte Carlo Simulation * Author: Kunwoo Park, CUNY Queens College * Method: Geometric Brownian Motion with Box-Muller Transform */ #include <stdio.h> #include <stdlib.h> #include <math.h> #include <time.h> #define SIMULATIONS 10000 // Number of potential market scenarios #define TRADING_DAYS 252 // Trading days in 1 year #define PI 3.14159265358979323846 // Function: Box-Muller Transform // Generates a Standard Normal Random Variable Z ~ N(0,1) double generate_gaussian_noise() { double u = (double)rand() / (double)RAND_MAX; double v = (double)rand() / (double)RAND_MAX; // Prevent domain error for log(0) if (u == 0) u = 0.0001; return sqrt(-2.0 * log(u)) * cos(2.0 * PI * v); } int main() { // 1. Market Parameters (Hypothetical Data for NVDA) double S0 = 135.0; // Current Price ($) double mu = 0.15; // Expected Annual Drift (15%) double sigma = 0.30; // Annual Volatility (30%) double T = 1.0; // Time Horizon (1 Year) double dt = T / TRADING_DAYS; // Time Step (Daily) // Seed Random Number Generator srand((unsigned int)time(NULL)); printf("--- Monte Carlo Simulation Report ---\n"); printf("Asset: NVIDIA Corp (Hypothetical)\n"); printf("Initial Price: $%.2f | Simulations: %d\n\n", S0, SIMULATIONS); double sum_final_prices = 0.0; // 2. Monte Carlo Loop for (int i = 0; i < SIMULATIONS; i++) { double current_price = S0; for (int day = 0; day < TRADING_DAYS; day++) { // Generate Random Shock double Z = generate_gaussian_noise(); // Discretized GBM Formula (Euler-Maruyama Method) // S(t+1) = S(t) * exp( (mu - 0.5*sigma^2)dt + sigma*sqrt(dt)*Z ) double drift = (mu - 0.5 * sigma * sigma) * dt; double diffusion = sigma * sqrt(dt) * Z; current_price = current_price * exp(drift + diffusion); } sum_final_prices += current_price; } // 3. Results Analysis double expected_price = sum_final_prices / SIMULATIONS; printf(">> Prediction Results:\n"); printf(" Expected Price after 1 Year: $%.2f\n", expected_price); printf(" (Theoretical Expectation: $%.2f)\n", S0 * exp(mu * T)); return 0; }

V. Discussion and Limitations

5.1 The "Fat Tail" Phenomenon

The GBM model assumes that market returns follow a Normal Distribution. However, empirical studies by Benoit Mandelbrot indicate that financial markets exhibit "Fat Tails" (Leptokurtosis). This means that extreme events (e.g., the 2008 Financial Crisis or the 2020 COVID-19 Crash) occur far more frequently than the Gaussian model predicts.

5.2 Future Research: Stochastic Volatility

To address this limitation, future work will incorporate Stochastic Volatility models, such as the Heston Model. Unlike GBM, where volatility ($\sigma$) is constant, the Heston Model treats volatility itself as a random process that reverts to a mean, providing a more realistic depiction of market dynamics.

VI. Conclusion

This study successfully implemented a Monte Carlo simulation of Geometric Brownian Motion in C to model the probabilistic future of stock prices. The results demonstrate that while physics-based equations provide a powerful framework for understanding market trends, they represent an idealized system. For a physicist entering the field of quantitative finance, understanding the deviation between these "ideal gas" models and the complex reality of human-driven markets is the key to building robust predictive algorithms.