Will Durant

03-01-2008, 03:24 PM

Harlow creates a portfolio that is optimized to minimize LPM2. He then compares its performance to a portfolio optimized to minimize the usual standard deviation measure. And lo and behold the LPM2 portfolio is "less risky" (when measured using LPM2) than the usual mean variance portfolio. Isn't this essentially tautological (assuming that the future behaved the same as the past)?

Beyond the first part of the paper where he makes a (pretty good) case that LPM2 is a better risk measure than standard deviation, what is the point of the rest of the paper?

ETA: The LPM2 actually does better than mean variance on a standard deviation basis too (looking at Table I) although this is not mentioned in the body of the paper. I guess it makes sense after all.

Beyond the first part of the paper where he makes a (pretty good) case that LPM2 is a better risk measure than standard deviation, what is the point of the rest of the paper?

ETA: The LPM2 actually does better than mean variance on a standard deviation basis too (looking at Table I) although this is not mentioned in the body of the paper. I guess it makes sense after all.