Value at Risk: Your Portfolio's Loss Risk in a Single Number
“With 95 % probability you won’t lose more than €8,000 over the next 20 trading days” — that is exactly what a Value at Risk (VaR) of €8,000 at 95 % confidence and a one-month horizon says for a €100,000 portfolio. VaR compresses loss risk into a single number, making it the standard metric in institutional risk management. For private investors it’s useful — as long as you know what the number doesn’t tell you.
Confidence, horizon, method: the three dials
A VaR without confidence level and time horizon is worthless. Common choices are 95 % or 99 % confidence over one day, ten days or one month. Important: at 95 % the threshold is breached on average every twentieth day — that is not a model failure, it is the definition. Three calculation methods are established:
- Historical simulation: actual past returns (e.g. 250 trading days) are sorted; the VaR is the corresponding percentile. No distribution assumption, but only as good as the observation window.
- Variance-covariance method: VaR is derived from volatility and correlations under a normality assumption. Fast, but it systematically underestimates the fat tails of real equity returns.
- Monte Carlo simulation: thousands of random paths based on a model. Flexible, but model-dependent.
Like the Sharpe ratio, parametric VaR is built on standard deviation — both metrics inherit its weaknesses. When comparing different VaR figures, also make sure confidence level and horizon match — a one-day VaR at 99 % and a one-month VaR at 95 % are simply not comparable.
Concentration risks, ETF overlap and look-through analysis – free with MoneyPeak.
What VaR doesn’t tell you — and how to use it anyway
The central criticism: VaR says nothing about how bad things get beyond the threshold. Whether the worst 5 % of cases mean a €9,000 or a €35,000 loss, the metric can’t distinguish — which is exactly why professionals add Expected Shortfall (CVaR), the average loss of the worst cases. On top of that, VaR models are calibrated on calm market phases and tend to fail precisely when needed — in crashes, correlations spike and the diversification that kept VaR low evaporates.
For private investors in practice:
- Use VaR as a comparison metric (portfolio A vs. B, before/after reallocations), not as a guarantee.
- Read it alongside maximum drawdown: VaR describes normal operations, drawdown the worst case.
- Translate it into euros: a monthly VaR (95 %) of 8 % on a €100,000 portfolio means that in one out of 20 months you will lose more than €8,000. If that number makes you nervous, you have an allocation problem, not a measurement problem.
MoneyPeak calculates risk metrics like volatility and loss potential directly from your real portfolio — not from a model portfolio.
Frequently asked questions
What does a VaR of 5 % at 95 % confidence mean concretely?
That your portfolio will, with 95 % probability, not lose more than 5 % within the chosen period — and conversely, in roughly one out of 20 cases it will lose more.
Is a low VaR a sign of a safe portfolio?
Only to a degree. VaR measures normal operations based on historical data. It captures extreme events (tail risk) and the correlation spikes of a crash poorly.
What is the difference between VaR and Expected Shortfall?
VaR states the loss threshold that will not be exceeded with a given probability. Expected Shortfall (CVaR) states how large the loss is on average when the threshold is breached — making it the more honest measure of extreme risk.
Concentration risks, ETF overlap and look-through analysis – free with MoneyPeak.