Semivariance - Schweser ComprehQ 1L

L

libba

New member
Joined
Jun 18, 2026
Messages
0
Reaction score
0
I may have misread something here.
For the solution of Comprehensive Questions from Schweser Book1 question 1.L. the use of n = 4, being the number of values below the mean opposed to 6 being the sample size, for the semivariance equation; however in the CFAI book example they use the full sample size as n for (n-1). Can anyone please explain?
 
semi variance is defined as the variance of observations below the mean.

Since mean was 4.4 all those observations whose value was less than the mean --> 3,4, 4.3 and 4.1 would be considered.

CP
 
thank you CP, i understand this however the example in the cfai text uses n as the sample size and not the number of values below the mean.

eg. from text

returns = 16.2, 20.3, 9.3, -11.1 and -17.0
mean = 3.54
below mean - -11.1 and -17.0

compute semivar - (-11.1-3.54)^2 + (-17.0-3.54)^2 = 636.2212

with n-1 = 5-1 semi var = 636.2212/(5-1) = 159.0553

notice use of sample size 5 for n

eg from schweser

returns - 3, 4, 4.3, 5, 4.10,6
mean = 4.4
below mean = 3, 4, 4.10, 4.3

compute semi var - (3-4.4)^2 + (4-4.4)^2 + (4.10-4.4)^2 + (4.3-4.4)^2 / (4-1) = 0.0074

notice use of n = 4 opposed to sample size of 6
 
I am not sure about that. The 2007 text book which I have, says for the same problem sqrt(636.2212) = 25.2 percent.

CP
 
not sure, when i doubt, id go with CFAI. schweser has an errata section on its site, fyi
 
not sure the CFA errata which asks to use the entire sample to calculate the semi variance instead of only the actual # of items that fall below the mean is the right erratum posted.

My CFAI 2007 text for the same problem had only the items below the mean as the number used to calculate the semi variance. which is, I believe, the right way, given the definition of semi variance.


CP
 
i understand the logic however this is from the 2009 curriculum, im guessing it would be the most up-to-date.

thanks for assistance
 
page numbers in your text book, designated reading as well as the same, only the errata seems to have been put in wrongly...

send them a mail and find out....

say there is a confusion... and ask.

CP
 
It hardly matters - what is semi-variance trying to estimate anyway? In any event, those two statistics are asymptotically just multiples of each other. The semi-variance is such a namby-pamby statistic that it's no wonder they can't get a consistent definition.
 
@JDV-Just cos I can't wrap my head around the math of it...how is it a multiple of variance?

@Mel- Yeah I remember those...remember FREAKING out about pureplay!!Unlever, relever, unilever...
 
You must log in or register to reply here.
Back
Top