QuarcHome
Model catalogueBack to the model index

Value at Risk (VaR) by historical simulation

Estimates a loss quantile by repricing under historical moves at a defined horizon.

How it works

Historical VaR applies historical joint risk-factor moves to the current portfolio and ranks the resulting losses. This preserves the market co-movements present in the chosen sample. Repricing can account for nonlinear instruments if the valuation engine supports the scenarios and conventions.

Worked example

With 100 equally weighted loss scenarios and the empirical quantile defined as the smallest loss whose cumulative probability reaches 95%, 95% VaR is the 95th smallest loss. If that loss is 12, five higher-ranked losses show how much worse the tail can be. The degree of loss above 12 is not supplied by VaR alone.

Validation and limitations

Document history length, overlapping horizons, non-trading dates and any valuation gaps. Check forecast coverage and assess whether the recent sample reflects the current risk regime. Compare filtered historical simulation when volatility changes materially.

Research sources