Looks like both are same but downside deviation usese some threshold while semivariance uses mean as threshold.
Is that correct?
Thanks.
can someone confirm?
Yeah I believe you are correct, but we don’t need to know how to calculate semivariance for the test. Just the fact that it’s not as good as variance for the 4 reasons given
thx
If the threshold is a recent average return, then we call the downside deviation the semivariance.
What are these 4 reasons again?
Downside deviation (aka, lower semideviation) is the square root of lower semivariance.
I think JSobes is referring to the four points below that describe deficiencies associated w/ semivariance.
- The calculation of semivariance is computationally challenging for large portfolios. 2. To the extent that investment returns are symmetric (normal distribution of returns described by mean and standard deviation/variance/covariance), semivariance is proportional to variance and therefore contains no additional information. 3. To the extent that returns may not be symmetric (i.e., if the portfolio contains options w/ asymmetric payouts), return asymetries are very difficult to forecast and may not be a good forecaset of future risk. 4. The estimation of downside risk takes into account only half the data (the data to the far left of the return mean) - we may lose some statistical accuracy.
CFAI Volume 4 Reading 23 page 231 for further reference.