FRM Part I · FRM Exam Part I · Hypothesis Testing
A bank's 99% one-day VaR model is backtested over 1,000 days, yielding 16 exceptions. Using the normal approximation to the binomial with p = 0.01, the analyst tests H0: p = 0.01 at the 1% significance level, two-sided (critical value 2.576). Which statement is correct?
Do not reject the model. Expected exceptions are 10 and the standard deviation is about 3.15, so z is roughly 1.9, below the 2.576 two-sided critical value at 1% significance. The 16 observed exceptions do not provide sufficient evidence that the model is inadequate.
- Az is about 1.90, so do not reject; the model is not shown to be inadequateCorrect
- Bz is about 6.00, so reject
- Cz is about 1.90, so reject because it exceeds 1.645
- Dz is about 1.60, so reject the model at 1% significance
Explanation
Expected exceptions = 1,000 x 0.01 = 10. Standard deviation = sqrt(1,000 x 0.01 x 0.99) = sqrt(9.9) = 3.146. z = (16 - 10)/3.146 = 1.91, about 1.90, below 2.576, so do not reject. Comparing with 1.645 uses the wrong critical value for a two-sided 1% test; z of 6 results from omitting the division by the standard deviation.
Did you get it right without looking?
One question tells you little. A timed set on Hypothesis Testing shows your real accuracy, how long you take and where you lose marks.
More Hypothesis Testing questions
- A risk manager estimates the mean of a return series and obtains a 95% confidence interval whose half-width is 1.20% based on 50 observation…
- Daily excess returns of a strategy have a known population standard deviation of 3.0%. A sample of 100 days gives a mean excess return of 0.…
- Using the setting of a 99% VaR backtest over 250 days (expected exceptions 2.5, standard deviation 1.57 under the null), a risk manager obse…
- A risk analyst computes the mean of 64 independent daily returns on a portfolio and obtains a sample mean of 0.30% with a sample standard de…
- A sample of 100 monthly excess returns on a fund has a mean of 0.80% and a sample standard deviation of 4.00%. Using the normal approximatio…
- A risk manager tests H0: sigma^2 = 4 against H1: sigma^2 > 4 for daily P&L (normal) using 21 observations. The sample variance is 6.5. The c…