CFAI Mock 2013 Q 27

Q 27: Based on correlation between CCIRP and TELLIRP, new volatility for the SCIRP is closest to:

What the ??? In the answer, they show a regression??? I thought we are done with this at level 2??


If this surprises you, then I wonder how much preparation you’ve actually done. Did you even study for Level 3? Or did you just skip straight to the Mock exam?

^not at all useful


At this point, I do not need this. If you cannot answer, please do not.

Schweser stated that this specific calculation should not be on your priority list.

But it’s on the MOCK !!!

There is a detailed explanation for this calculation in schweser note. If you put time into comprehending all that, you should get a grip. I got this question wrong too. Or maybe we can have Greenman72 explain this to us.

If you got over 70% in this mock, I wouldn’t worry too much about this calculation.

I can’t explain it, because I glossed over it. This calculation is a violation of Andy Holmes’s “Ugly Rule”. If a calculation/formula looks really ugly, then it has a relatively small chance of showing up on the exam.

But I remember reading it, and I remember seeing a LOT of multiple regression-type questions in my Level 3 studies. So I’m not really surprised to see a regression in the mock exam.

this question was bullshit. i would bet 20% answer it correctly without having guessed so not really worth stressing about and spending a day to memorize. Better off reviewing institutional IPS


Anything that takes 10 minutes to calculate is not something I’m going to memorize. I’ll look at the calculation, and make sure that I’m able to make an educated guess on game day, but I’m not going to sweat it. If it’s there, it will be one question–max.

What Study Session / reading is this?

cap mkt expectations. They required you to use the portfolio variance formula (s2=w1s1 + w2s2 +2w1w2cov etc) in order to find the variance of the error term. Then take that error term and re-apply it to the variance after re-adjusting the covariance to correlation following some new information. It was a question that required too much idiotic calculation vs. knowing a concept. Hence in such a situation educated guess is more value than spending too much time in one question out of 60.

Page 28 in Volume 3 to be precise. Equations 3a and 3b

you actually did not need to do all they did.

knowing that the correlation was zero before - you only had to add the last term …

Covariance (1, 2) = 0.25 * sqrt(0.0784) (sigma 1) * sqrt(0.1024) (sigma2) = 0.0224

Now Last Term = 2* F1 * F2 * Covariance = 2*1.02 * 1.045*0.0224 = -0.0477

New Variance = Old Variance + New Factor = 0.2704 + 0.0477 = 0.3181

Sigma = sqrt(0.3181) = 56.4%

Word. If it takes more than three minutes to calculate and fill in the bubble, you’re a net loser on this question. Better that you learn the concept, so you can eliminate one wrong answer and then take your 50-50 chance.

It actually drove home a point for me - when calculating variance (aka “self-covariance”) - you add the variance of the error term; unlike calculating covariance, when you ignore it. I in fact ignored the error variance and got the (wrong) one of the two plausible answers.

Mii = bi1^2 Var(F1) + bi2^2 Var(F2) + 2bi1 bi2 Cov(F1,F2) + Var(ε), Mij = bi1 bj1 Var(F1) + bi2 bj2 Var(F2) + (bi1 bj2 + bi2 bj1) Cov(F1,F2)

agree: Andy Holmes’s “Ugly Rule”