Calculating Skewness and Kurtosis

Which equation do we use for calculating these? In the Schweser books they use 1/n as the coefficient for both. But the CFAI book uses ((n / (n-1)(n-2)) as the coefficient for Skewness and an even messier one for Kurtosis. It says that if n is large enough, it reduces to 1/n for both.

What is considered large and can we just know the simplified formula or will we need to know this more complicated one?


no need to memorise skewness/kutosis formulas just what they mean :slight_smile:

Yes read the LOS - just says to ‘explain’ measures of sample skewness and kurtosis - but do be able to define leptokurtic/mesokurtic/platykurtic etc., I saw this on a recent mock. I.e. know that kurtosis of 3 is standard, kurtosis - 3 = excess kurtosis, and whether that means lepto/meso/etc.

Thanks…I wish they wouldn’t have EOC questions that want you to use the calculation if we won’t need it.

The only real application of this throughout the 3 Levels is that you understand that a lot of the theories and techniques used in analysing markets are based around a Normal Distribution, which assumes no excess kurtosis (i.e. a 3 measure) and no skewness (i.e. right/left tail distrubtions) on returns.

It is obviously becoming more of a focus that a Normal Distribution is not necessarily a fair assumption for market returns (whichever market that may be) and therefore they are seeking that you understand that in detailed modelling and analysis of past returns, you may need to factor in or measure these amounts (which will be done by a computer on your behalf and therefore no need to actually know how to compute).

It is important that you know what it is and how to apply it though.