The number of exceptions should be equal to the VaR significance level. In the context of backtesting a VaR model, an exception is an instance where the actual loss exceeds the predicted loss. The VaR significance level, on the other hand, is the probability that the actual loss will exceed the predicted loss. Therefore, for a model to be accurate, the number of exceptions should be equal to the VaR significance level. This is because the VaR model is designed to predict the maximum loss that will not be exceeded with a certain level of confidence. Therefore, if the number of exceptions is equal to the VaR significance level, it means that the model's predictions are accurate. However, it's important to note that due to the limited sample size of the backtesting period and the specific confidence level used, it's unrealistic to expect the model-predicted number of exceptions to be found in every sample. Therefore, the observed number of exceptions may not always be the same as that predicted by the model, but this doesn't necessarily mean that the model is flawed. Instead, a threshold must be established at which the model is rejected.

Choice A is incorrect. The number of exceptions should not be greater than the VaR significance level. If the number of exceptions is greater than the VaR significance level, it indicates that the model is underestimating risk and therefore, it's not accurate.

Choice B is incorrect. The number of exceptions should not be less than the VaR significance level. If there are fewer exceptions than expected at a given confidence level, this suggests that our model may be overestimating risk which again points to an inaccurate model.

Choice D is incorrect. As explained above, for an accurate VaR model, the number of exceptions should ideally match with the VaR significance level and hence 'None of above' does not hold true in this context.

Things to Remember

• Backtesting is a risk management technique used to test the accuracy of financial models by comparing predicted outcomes with actual outcomes.
• Value at Risk (VaR) is a measure used to estimate the potential loss in value of a portfolio over a defined period for a given confidence interval.