Some misconceptions about Monte Carlo for investment markets simulation.

I recently saw a post claiming that Monte Carlo simulations are useless for investment markets, because “they produce time independent simulations.”

Sure, if you use some naive approach that generates independent samples, but why would we do that given our knowledge of stylized market facts?

In the article below, Laura Kristensen and I introduce a new approach for Time- and State-Dependent Resampling.

The Python case study clearly illustrates that it is possible to produce simulations with time series dependencies.

The benefit of resampling methods is their ability to capture the cross-sectional dependencies, no matter how complex they are.

By performing clever data preprocessing and conditioning on time and state, we can generate realistic paths for very high-dimensional data: