CFA Level I · CFA Level I Exam · Simulation of Financial Asset Prices and Returns
An analyst runs a bootstrap using 5 years of returns that include no market crisis. The resulting estimate of the 1% worst-case loss is most likely:
The estimate is most likely understated. Bootstrap draws only from observed returns, so if the five-year sample contains no crisis, severe losses cannot appear in any resample. The method inherits the limitations of the data and cannot reveal tail risks absent from it.
- Aunderstated because extreme outcomes absent from the sample cannot be drawnCorrect
- Boverstated because resampling with replacement inflates the variance
- Cunbiased because resampling removes the dependence on the sample period
Explanation
Bootstrap can only reproduce observed values, so if the sample lacks crisis returns, tail losses will be understated. Resampling with replacement does not systematically inflate variance. The result still depends heavily on the sample, so it is not free of sample-period dependence.
Did you get it right without looking?
One question tells you little. A timed set on Simulation of Financial Asset Prices and Returns shows your real accuracy, how long you take and where you lose marks.
More Simulation of Financial Asset Prices and Returns questions
- Compared with historical simulation, a parametric Monte Carlo simulation of asset returns is most likely to:
- In a bootstrap procedure applied to a sample of 60 monthly returns, each resample is most likely constructed by:
- An analyst has 60 monthly returns and wants to estimate the standard error of the sample median using a bootstrap. Which approach is most ap…
- An analyst uses Monte Carlo simulation to estimate the value of a portfolio in one year. Which step is most likely performed first?
- An analyst increases the number of Monte Carlo trials from 1,000 to 100,000 to value a derivative. This change will most likely:
- An analyst wants to estimate the sampling distribution of the median daily return from a sample of 250 observations without assuming any par…