Page tree
Skip to end of metadata
Go to start of metadata

Backtesting is a statistical method for evaluating the forecast power of a stochastic model. It is based on a daily comparison of the potential risk exposure determined using the risk model and the value change of the portfolio included in this model calculation over the underlying holding period.
As for backtesting purposes, a sample portfolio is defined, i.e. a compilation of several deals in accordance with certain criteria, mostly risk transactions that are analysed together. In this case, the sample portfolio reflects the business structure of the entire bank. Deals representative of each sector are analysed using the portfolio to validate the quality of the forecast made by the risk model.
This sample portfolio is evaluated in an analysis using the market data for the posting date of the portfolio and the market data for the current posting date. The difference is formed from the results of these present values and indicates the efficiency of the risk model applied.

  • No labels