What common mistakes show up in a Monte Carlo VaR implementation?

Question

AcadiFi · Accepted Answer

Common mistakes include using inconsistent return definitions, applying weights incorrectly, mixing arithmetic and log returns without care, using a stale or invalid correlation matrix, and failing to revalue nonlinear positions. Monte Carlo VaR is flexible, but that flexibility creates many places to hide modeling errors.

Suppose Northbridge Quant Risk simulates three equity returns and uses Cholesky decomposition to impose correlation. If the analyst multiplies simulated log returns by simple portfolio weights and treats the result as an exact simple portfolio return, the loss distribution may be distorted. If the correlation matrix is not positive definite, the simulation may fail or require an adjustment that changes dependence.

The FRM mindset is to trace the calculation from inputs to P/L. Are the risk factors defined correctly? Are volatilities and correlations estimated over a consistent horizon? Are weights updated or frozen? Are derivatives repriced or approximated? A Monte Carlo engine is not validated merely because it produces thousands of scenarios. It is valid only if those scenarios translate into economically coherent portfolio losses. On the exam, also state the horizon, confidence level, data source, modeling assumption, and validation step, because those choices usually explain why two reasonable VaR answers differ materially. Practice more in our FRM question bank.

What common mistakes show up in a Monte Carlo VaR implementation?

Master Part I with our FRM Course

Related Questions

Practice Questions