What's new

# Exam FeedbackFRM Part 2 (May 2015) Exam Feedback

#### JDGutzmann

##### Member
i faced the same problem as well and was doubting if the none of the answers was correct because i obtained:
1- e ^ (-0.042*2/0.6) = 13.05%

However, the answer provided in the question was 13.5% and the rests were much lower than this percentage....
Either GARP made a typo or I worked in a wrong way.

I got:
1 - (1 - 0.042 / 0,6 ) ^ 2 = 0.1351

Which uses not the continous but discrete PD(!)
I have to say that, despite the statement in the cover about all interest rates and the like to be continously compounded, some were definitely not, and not marked as such. I am positive that the question about two ways to average forward rates also used discretely compounded rates. I think GARP wasn't consistent on this.

#### saran11

##### New Member
might
I remember the text specified something like the new stock had the same variance and same expected return of the others in the portfolio. So SR does not change, RAROC either, 14 PC might be - but why not 15 then?
The fact that 12 PC now explain only for 65% instead of 85% means to me that the stock has a lower correlation with the portfolio's incumbent risk factors, so it is likely to reduce the amount of the diversified VaR, once added to it.
So I went for a DVaR decrease. Anyone agrees?

Might be wrong. My reasoning, the question said the modified portfolio has same return but has 65% components of the old portfolio. I remember not seeing same variance. So thought variance increased. and opted low sharpe ratio. Might be wrong.

#### Ryan S

##### Member
Subscriber
And another question on "if correlation of the of assets in the pool increases what could result" I chose the CVAR decreases. I might be wrong.

Default correlation is an increasing function with UL, it does not impact EL. Therefore CVaR should increase, I believe.

#### robin3301

##### New Member
Default correlation is an increasing function with UL, it does not impact EL. Therefore CVaR should increase, I believe.
Can you remember other options?

#### saran11

##### New Member
Default correlation is an increasing function with UL, it does not impact EL. Therefore CVaR should increase, I believe.

Yeah I agree, there can be 2 cases if the probability of default is less than the significance of var and when the probability of default is greater than significance of var.

cvar 95% of 100 assets with 0.1 probability of default each with 0 correlation is greater than cvar 95% of same portfolio with 1 correlation.

i guess both the answers are wrong. might be the other 2 options.

Last edited:

#### saran11

##### New Member
Can you remember other options?

i think u marked between the other 2 options. I dont remember the other options.

#### Alberto G

##### New Member
Subscriber
might

Might be wrong. My reasoning, the question said the modified portfolio has same return but has 65% components of the old portfolio. I remember not seeing same variance. So thought variance increased. and opted low sharpe ratio. Might be wrong.

Hi saran11,

I'm quite confident about the "same variance" was reported too, but maybe somebody else can confirm it... if so DVaR decrease would be the only one that make sense to me.
But if I was mistaken and no variance info was given, how would you solve it for a variance increase? The mere fact that 12 risk factors now explain for less variance % means that assets are less correlated, not necessarily that absolute variance is more than before.

As far as I remember, the question was about a portfolio rebalancing decision, so same amount invested but less risk factor correlations (with possibly same individual variance) >> less portfolio variance >> higher SR and decrease in DVaR.

Does it make sense?

A.

#### robin3301

##### New Member
Hi saran11,

I'm quite confident about the "same variance" was reported too, but maybe somebody else can confirm it... if so DVaR decrease would be the only one that make sense to me.
But if I was mistaken and no variance info was given, how would you solve it for a variance increase? The mere fact that 12 risk factors now explain for less variance % means that assets are less correlated, not necessarily that absolute variance is more than before.

As far as I remember, the question was about a portfolio rebalancing decision, so same amount invested but less risk factor correlations (with possibly same individual variance) >> less portfolio variance >> higher SR and decrease in DVaR.

Does it make sense?

A.
i also remember same varience

#### saran11

##### New Member
Hi saran11,

I'm quite confident about the "same variance" was reported too, but maybe somebody else can confirm it... if so DVaR decrease would be the only one that make sense to me.
But if I was mistaken and no variance info was given, how would you solve it for a variance increase? The mere fact that 12 risk factors now explain for less variance % means that assets are less correlated, not necessarily that absolute variance is more than before.

As far as I remember, the question was about a portfolio rebalancing decision, so same amount invested but less risk factor correlations (with possibly same individual variance) >> less portfolio variance >> higher SR and decrease in DVaR.

Does it make sense?

A.

Yep, I guessed on sharpe. I didnt have a concrete reasoning. I missed same variance part I guess.

#### Goldfish3

##### New Member
I definately remember a question that was word for word exactly the same as either the past exams (2010 - 2014) or schweser questions. But cant specifically remember for the life of me. And no this is not the QQ. It was a theoretical.

#### saran11

##### New Member
Also, I understand that bond prices cant follow black scholes options pricing because they dont have constant valatility
along the maturity.
Can someone shed some light on why option that says " bond prices can not take log normal" is wrong. I marked this(first option) option not reading other options.

#### vsism

##### New Member
I
Also, I understand that bond prices cant follow black scholes options pricing because they dont have constant valatility
along the maturity.
Can someone shed some light on why option that says " bond prices can not take log normal" is wrong. I marked this(first option) option not reading other options.
I think the correct answer to this question was the one in which Geometric Brownian Motion was mentioned. I am forgetting the exact statement though.

#### Alberto G

##### New Member
Subscriber
I dont think CVA can be perfectly hedged but I forget other choices

I don't remember the exact wording but I doubt there was the word "perfectly" - the only perfect hedge is in a Japanese garden, right? ;-)
Though, if it was "fully" then it make sense to me (I chose that answer btw), as the CVA charge is the cost associated with hedging the position and it is relevant for Basel III. And it is also important to hedge it in advance.

As per Canabarro: "The lesson to be learned from the latest playing of this story is that banks should hedge their CVAs outright, otherwise they are likely to start hedging it too late".

Agree?

#### Alberto G

##### New Member
Subscriber
Also, I understand that bond prices cant follow black scholes options pricing because they dont have constant valatility
along the maturity.
Can someone shed some light on why option that says " bond prices can not take log normal" is wrong. I marked this(first option) option not reading other options.

rates can follow lognormal distribution (black-karasinski), but prices are pull-to-parity as they approach maturity (a non-defaulting bond pays par in every scenario once it expires - not lognormal at all).
As a result, price volatility decreases near the maturity date (right answer).

#### saran11

##### New Member
rates can follow lognormal distribution (black-karasinski), but prices are pull-to-parity as they approach maturity (a non-defaulting bond pays par in every scenario once it expires - not lognormal at all).
As a result, price volatility decreases near the maturity date (right answer).

I remember it was mentioned prices " prices follow log normal". I might be wrong.

#### Alberto G

##### New Member
Subscriber
I remember it was mentioned prices " prices follow log normal". I might be wrong.

Oh yes, I meant it (i did not see the "not" in your previous post) ;-)
The sentence "bond prices follow lognormal distrib" is wrong (they are in fact pulled-to-parity, with decreasing vola towards maturity).

#### vsism

##### New Member
I have a small request to everyone. If you guys can please tell how much you are expecting to score on an average? It will give a rough indication on where everyone of us are standing, atleast here on forum. I am really anxious about the results and want to know my relative performance with respect to you all. Please if you all can give your rough estimate of scores. I think mine will be somewhere between 40-50 (on the conservative side). Request you all to please post yours.

#### Alberto G

##### New Member
Subscriber
I have a small request to everyone. If you guys can please tell how much you are expecting to score on an average? It will give a rough indication on where everyone of us are standing, atleast here on forum. I am really anxious about the results and want to know my relative performance with respect to you all. Please if you all can give your rough estimate of scores. I think mine will be somewhere between 40-50 (on the conservative side). Request you all to please post yours.

Normally this kind of exams have a cutoff point calibrated as the 70% (or 75%) of the 95th quantile (or 95-100 q average) of the results distribution. 70% of 80 is 56... with anything above 56 (but probably 50 is enough) you should be a supersafe pass...

Consider then that BT forumers have a higher pass-rate average than the rest of the world (they advertise!), so probably we are outperforming (on aggregate) the folks out there (hope so, at least) this time as well ;-)

Last edited:

#### saran11

##### New Member
Normally this kind of exams have a cutoff point calibrated as the 70% (or 75%) of the 95th quantile (or 95-100 q average) of the results distribution. 70% of 80 is 56... with anything above 56 (but probably 50 is enough) you should be a supersafe pass...

Consider then that BT forumers have a higher pass-rate average than the rest of the world (they advertise!), so probably we are outperforming (on aggregate) the folks out there (hope so, at least) this time as well ;-)
I have a small request to everyone. If you guys can please tell how much you are expecting to score on an average? It will give a rough indication on where everyone of us are standing, atleast here on forum. I am really anxious about the results and want to know my relative performance with respect to you all. Please if you all can give your rough estimate of scores. I think mine will be somewhere between 40-50 (on the conservative side). Request you all to please post yours.

I might get around 45-50.. :/

#### vsism

##### New Member
Thanks a lot
Normally this kind of exams have a cutoff point calibrated as the 70% (or 75%) of the 95th quantile (or 95-100 q average) of the results distribution. 70% of 80 is 56... with anything above 56 (but probably 50 is enough) you should be a supersafe pass...

Consider then that BT forumers have a higher pass-rate average than the rest of the world (they advertise!), so probably we are outperforming (on aggregate) the folks out there (hope so, at least) this time as well ;-)
Thanks a lot for the info I think I have to give it again this November. Just hoping to cross the boundary line though :|

Replies
328
Views
59K
Replies
57
Views
11K
Replies
158
Views
30K
Replies
59
Views
12K
Replies
424
Views
69K