According to the curriculum does it mean that for HY bond we can not assume normal distribution?
ss 5.2
Tail risk is the risk that there are more actual events in the tail of a probability distribution than probability models would predict. As an example of tail risk, consider that during a 25-year period ending in August 2008, monthly total returns for the widely used Bloomberg Barclays US Corporate Index had a standard deviation of 2.1%. During the last four months of 2008, however, the monthly total returns of the Bloomberg Barclays Corporate Index were –7.8%, –6.4%, 4.1%, and 6.8%, sequentially. The absolute values of these monthly returns, respectively, were 3.7, 3.0, 2.0, and 3.2 standard deviations from the mean of the preceding 25-year period. With a normal distribution, it would have been almost impossible for a risk model based on historical returns to foresee the possibility of such a return pattern (Institute 231)
Institute, CFA. 2018 CFA Program Level III Volume 4 Fixed Income and Equity Portfolio Management. CFA Institute, 07/2017. VitalBook file.
Use the analytical method when you expect that the returns will have a normal distribution, and you do not have historical data which you believe will reasonably represent the future.
Use the historical method when you have historical data that you believe will reasonably represent the future, especially if you expect that the returns will not have a normal distribution.
certainly true from a theoretical perspective but raw historical data is such a poor estimator of the future especially on something with non normal distributions that I think the only time you’d actually suggest using it is in some theoretical construct on an exam. There just isn’t enough history for the law of large numbers to work especially on a non normal distribution. I could see why people who actually work in the field would find deciding between two effectively bad choices difficult.
