Assessing Models using Monte Carlo Simulations

We establish a framework for assessing the validity of a given model using Monte Carlo simulations and inferences based on sampling distributions. Using this framework, we show that geometric brownian motion alone cannot generate a majority of the patterns in the distribution of stock returns and wealth creation. Our paper represents an often overlooked departure from the traditional way of validating asset pricing models, in which implications are derived, parameters calibrated, and magnitudes compared to empirical data. Instead, we seek to leverage the power of large numbers by conducting numerous simulations and assessing the probability that they contain our realized stock market.