Kurtosis

The CFAI book has a formula to compute excess kurtosis, however, in the schweser notes they just say it is kurtusis- 3 with a kurtosis of 4.39 (example on pg 305) using the formula gives excess kurtosis of .05, which we coul interpret as mesokurtic , but using 4.39-3 = 1.39 that would be leptokurkik, so it is a relative difference…what do you guys say??? I think it’d take like 5 min to do a question like that…does anybody know if they even test this? cheers

Unlikely, but if the LOS says “COMPUTE KURTOSIS” then it is fair game.

Could you paste the formula too for reference? Odds are there’s a slip in interpretation in this one but then again…schweser has been known to err.

No way they ask you to calculate kurtosis. Zero chance.

All you have to know is that (Kurtosis - 3) = Excess Kurtosis. Normal distributions have Kurtosis of 3 and excess of 0.

they didint ask ME to calculate it. but its pretty simple if you think about it… just a bunch of Ns 3’s 2’s and 1’s… looks like a count formula. what they WILL do tho is try to trick you when they give you kurtosis and excess kurtosis. for example, they will ask something like, excess kurtosis = 3 they’ll trick you like that i imagine. note i didint violate anything…

I’ve just come upon this also, but the CFAI and Schweser formulas are reconciled like this: CFAI formula for sample kurtosis is = [n(n+1)/(n-1)(n-2)(n-3)] * sum(x-X)^4 / s^4 and then it says to subtract [3(n-1)^2 / (n-2)(n-3)] from the sample kurtosis formula above in order to obtain excess kurtosis. Then CFAI actually says that when n gets to be very large, you can use 1/n * sum(x-X)^4 / s^4 to calculate sample kurtosis, and it says that you can just subtract 3 to get the excess kurtosis. Which is what Schweser uses. Quite a pain to find this out - shows that you really do need to do the CFAI end of chapter problems to make sure Schweser didn’t miss anything. Which it also did with semivariance in reading 7. Blah!

CFAI will not test whether you know how to punch keys on calculator to calculate kurtosis/skewness. Know the general idea that for normal distribution kurtosis is 3.

That’s what I’m thinking, thanks… I wasn’t even gonna bother with this anymore , there is a lot of material to go over. Good luck everyone