Sortino and SemiVariance

  1. Sortino considers options? 2. And what about SemiVar limits’ impact to Sortino? So, there is limit of Sharpe, and limit of semivariance, and they kind of point out short comings of each other… that’s really where my confusion comes from.

If a ratio relies on the normality of returns, then it won’t work for options which have asymmteric returns.

Why Sortino is good for options then?

Any one is expert on Sortino? Bigwilly? Mo34?

Sortino is good b/c it doesnt assume normality b/c the denominator is Downside Deviation.

Downside deviation is similar to semi-variance, thus I thought still same as sd type?

what’s your email? If convenient to you…thanks!

sortino = R-MAR/sqrt(semivariance)

Downside deviation is similar to variance but it only considers one side, so it wont be ‘skewed’ by ‘positively skewed’ distributions


no Std deviation considers all the returns. Downside deviation and semi deviation(variance) are the same measure. They only considers returns below a minimum return requirement, hence the MAR. Email is bigwilly11189 @ G M A I L . C O M but I dont check it all the time.

That’s the point: if Sortino is using sqr(semi-variance) it will get all the short-comings of semi-variance: 1. Yes, it ignores those good (positive) deviations. (good) 2. Still it assumes normal distribution and impacted by skewness and kurtosis of those semi-variances. In short semi-variance is good, but it is still “variance” type of measurement…

No it doesnt assume normal distribution. Forget semi-variance I could have been wrong, but thought it was the same as semi-deviation but I guess not.

sortino uses downside devaition… .downside deivation isn’;t susceptible to the same shortcomings as stan dev and variance…

comp-sci kid is right Be careful with statements like ‘assumes normal distribution’. Remember the normal distribution is just a pretty graph. A calculation can’t ‘assume a normal distribution’. I can divide by whatever number I like without assuming anything! I could make the ‘elem100 distribution’ which could be ( Return - Min acceptable return ) / (Downside deviation + 13 pink elephants). My measure would be a useful way of determining risk adjusted performance, so long as I used it consistently. I am not assuming anything. Its the trying to work out a probability or % either side of a number in a distribution in which you are assuming that probability calculation has relation to the (already known) normal distribution, because you use a Z-table, or whatnot, based on the special normal distribution.

^Seriously totally uncalled for “be careful with statements like ‘assumes…’” Are you kidding me? For this test purposes you SHOULD ASSUME, so stop trying to get technical for a 6 Grade question. Ok, sorry I’m a little on edge lately.

sorry man, I didnt mean it like that! i just wanted to clarify what assuming normal means

Its ok, like I said I’m easily aggitated lately. No problems.

bigwilly Wrote: ------------------------------------------------------- > No it doesnt assume normal distribution. Forget > semi-variance I could have been wrong, but thought > it was the same as semi-deviation but I guess not. bw, I am still having foxiegroup’s thoughts but I think I need more reading regarding “sortino doesn’t assume normal distribution”. Do you have CFAI/Schweser page reference for this? Thanks. - sticky

I think for the fact that it doesnt consider retursn above the Target or MAR return then it can’t be a normal distribution b/c its ignoring returns greater than X%, so it doesnt consider both sides of the normal distribution. But at the same time I think it considers Normality in the end that you are considering… Its almost as if it basis its theory off a normal distribution, but isnt normal b/c it only considers return below a certain point… Now I’m confused. So I apologize b/c I would have bet my life on the fact taht I read that or heard that or learned that before, but now I cant find it.