Value at Risk (VaR)
VaR estimates the maximum potential loss of a portfolio over a specific time period at a given confidence level (e.g., 'We have 95% confidence losses won't exceed $10M in one day').
Concept map
Learn, apply, review
Definition
VaR estimates the maximum potential loss of a portfolio over a specific time period at a given confidence level (e.g., 'We have 95% confidence losses won't exceed $10M in one day').
Use case
Used in risk management workflows, analysis, and technical interviews.
Judgment check
Useful only when the assumptions and inputs behind the metric are understood.
Deep dive
How to think about Value at Risk (VaR)
Three main calculation methods: Historical (using past returns), Variance-Covariance (parametric, assuming normal distribution), and Monte Carlo Simulation (running thousands of scenarios). VaR has limitations — it doesn't indicate losses beyond the confidence threshold (tail risk), leading to use of CVaR (Conditional VaR).
Example: A bank reports 1-day 99% VaR of $50M. This means there's only a 1% chance (1 in 100 days) of losing more than $50M tomorrow. During the 2008 crisis, many firms saw losses exceeding their VaR estimates as correlations spiked.