Exam Feedback May 2019 Part 2 Exam Feedback

[nikic]

Folks, what about the question on Duration / Principal / Cash Flow mapping? I chose the first answer, which was VAR of Cash Flow mapping < VAR from Duration mapping. What's the actual answer to this?

 

[Jaskarn]

A

Didn't the answer say something like "it's sufficient to put into effect the contingency plan when the capital levels breach the minimum regulatory capital ratios"? In which case, it would be wrong since the contingency plan should be put into effect at some point before that?

Ans to this was that most of the products has been moved from BILATERAL OTC market to EXCHANGE TRADED.

 

[ziminli1228]

BCVA = CVA - DVA so it was 40K to me

I think mine was 4k too. 50*200-60*100=4k

 

[jkowalonek]

Anyone remember the one on the right hedge fund strategy? There was global macro, long short, and risk arbitrage .
I put risk arbitrage b/c others looked wrong, Global macro said it was passive and long short said it didn't include options.

 

[kylian.mbappe]

Do you happen to recall the question / answer in more detail? The distribution as I recall models the severity of losses, but the frequency of losses plays a role in setting the appropriate threshold. I'm getting all mixed up now as to what was tested.



For some reason I cannot seem to recall this question. Any additional details?



Yes, precisely this.

Yes I remember the question and the frequency of loss plays a role in the reliability of the chosen distribution. But the distribution itself mesures the size of losses.

 

[nikic]

Anyone remember the one on the right hedge fund strategy? There was global macro, long short, and risk arbitrage .
I put risk arbitrage b/c others looked wrong, Global macro said it was passive and long short said it didn't include options.

Risk arbitrage, the one which goes on to say that it represents the tail risk of equities or something of the sort? I picked it as well.

Ans to this was that most of the products has been moved from BILATERAL OTC market to EXCHANGE TRADED.

How did this tie into the question on capital planning? (PS: I might have selected this option, but just doesn't ring a bell yet...)

 

[Jaskarn]

Anyone remember the one on the right hedge fund strategy? There was global macro, long short, and risk arbitrage .
I put risk arbitrage b/c others looked wrong, Global macro said it was passive and long short said it didn't include options.

Risk arbitrage, the one which goes on to say that it represents the tail risk of equities or something of the sort? I picked it as well.



How did this tie into the question on capital planning? (PS: I might have selected this option, but just doesn't ring a bell yet...)

Oops i didn't realise, it is related to other question where it asked us that what are the changes that has been done after 2008 crisis something like this. For that question this is the ans.

 

[Jaskarn]

I think the correct thing to ask after part 2 exam is that Is there multiple options correct? Because i found that most of the questions I could filter out 2 options but had really tough time with remaining 2. And even not i am 99% sure that both option were correct.

 

[jkowalonek]

For Repo GC v. Special, I put A, which was Special is less than GC, and the price increased headed to auction, and decreased after.

 

[Jaskarn]

The reason below why I felt that the explanation that was provided for with regards to alpha and benchmark was better than for the peer group alpha. Because I do not believe they regressed the peer alpha for significance.

View attachment 2088

This explanation is absolutely correct but i feel context is different.
Here context is comparing alpha while choosing which fund to invest in as in we are outsider and we want to choose a fund, preferably hedge fund, where we can invest our money so we compare alpha of all the funds in that survivorship bias will come as hudge fund will not disclose alpha of those manager whose alpha is less.

situation of question was different.

In paper's situation, you are not investor anymore, you are fund manager under whom let's say 5 managers are working and you want to do there appraisal based on there performance~ Alpha each one had. SO you want to basically rank them in an order so that you can decide there rating hence you need a parameter that helps you in this comparasion. So you will regress excess retuen with peer.
Plus you are insider now so no surviourship baise any more as in you know what 5 managers alpha is.

 

[nikic]

This explanation is absolutely correct but i feel context is different.
Here context is comparing alpha while choosing which fund to invest in as in we are outsider and we want to choose a fund, preferably hedge fund, where we can invest our money so we compare alpha of all the funds in that survivorship bias will come as hudge fund will not disclose alpha of those manager whose alpha is less.

situation of question was different.

In paper's situation, you are not investor anymore, you are fund manager under whom let's say 5 managers are working and you want to do there appraisal based on there performance~ Alpha each one had. SO you want to basically rank them in an order so that you can decide there rating hence you need a parameter that helps you in this comparasion. So you will regress excess retuen with peer.
Plus you are insider now so no surviourship baise any more as in you know what 5 managers alpha is.

Yes, while comparing peer alpha should be superior, the methodology/basis described was flawed. Benchmark alpha outperformance was correctly described (i.e. based on statistical significance). But for peer alpha, they mentioned something else totally, when the basis should have tied back to statistical significance of the alpha. Hence I did not pick it.

 

[jkowalonek]

There was a real life probability v. risk neutral probability, I'm not sure what I put for the answer. Anyone remember? I think it was related to the difference between the two.

 

[nikic]

Oh well, there was the 98% ES for 252 trading days. 252 * 98% = 246.96....so do you take the 247th, 248th, 249th, 250th, 251st and 252nd observation (i.e. 6 worst observations)? Or do you round up the 246.96 to 247 and then only take the 5 worst losses?

I thought this was a silly question as there were answers for both using the average of the worst 5 or worst 6 trading days returns. By right for ES you consider the losses beyond the VAR threshold.

 

[nikic]

There was a real life probability v. risk neutral probability, I'm not sure what I put for the answer. Anyone remember? I think it was related to the difference between the two.

Mine was that the CVA using Risk Neutral Probability would be more than the CVA using Real Life Probability, since Risk Neutral > Real Life.

 

[Jaskarn]

There was a real life probability v. risk neutral probability, I'm not sure what I put for the answer. Anyone remember? I think it was related to the difference between the two.

KMV, Merton model , credit metric etc..

Merton was answer as option said value of EL will not change under merton model even if equity price of comoany changes which is correct because to calculate EL we need to input PD and in case of merton model PD is vlooked from a table but in case of KMV we use actually default data, which will change if equity price of company changes.

Oh well, there was the 98% ES for 252 trading days. 252 * 98% = 246.96....so do you take the 247th, 248th, 249th, 250th, 251st and 252nd observation (i.e. 6 worst observations)? Or do you round up the 246.96 to 247 and then only take the 5 worst losses?

I thought this was a silly question as there were answers for both using the average of the worst 5 or worst 6 trading days returns. By right for ES you consider the losses beyond the VAR threshold.

There was this one question were we can to calculate 95% ES and we were given WORST 10 returns BUT THOSE RETURNS WERE NOT ARRANGED IN ORDER they were reported as they happened date wise so were need to order them can calculate ES

 

[nikic]

Oh well, there was the 98% ES for 252 trading days. 252 * 98% = 246.96....so do you take the 247th, 248th, 249th, 250th, 251st and 252nd observation (i.e. 6 worst observations)? Or do you round up the 246.96 to 247 and then only take the 5 worst losses?

I thought this was a silly question as there were answers for both using the average of the worst 5 or worst 6 trading days returns.

KMV, Merton model , credit metric etc..

Merton was answer as option said value of EL will not change under merton model even if equity price of comoany changes which is correct because to calculate EL we need to input PD and in case of merton model PD is vlooked from a table but in case of KMV we use actually default data, which will change if equity price of company changes.



There was this one question were we can to calculate 95% ES and we were given WORST 10 returns BUT THOSE RETURNS WERE NOT ARRANGED IN ORDER they were reported as they happened date wise so were need to order them can calculate ES

Neither of the above are referring to the same question you quoted? KMV and Merton were a different question, not the risk neutral / real world probability of default question.

The ES question where the data was not arranged was for the 98% question. The 95% ES data was arranged nicely according to the percentile.

 

[chiru696]

there was a question on unsmoothing returns

 

[David Harper CFA FRM]

@nikic regarding ES, we do not "consider losses beyond the VaR threshold." Averaging losses is called a conditional VaR (or less often, tail conditional expectation, TCE) which has never been assigned and is ambiguous when the VaR is ambiguous. We use expected shortfall (ES) which is never ambiguous and is delimited by the probability not the quantile.

So, for example, if T = 252 trading days and let's say the worst 10 losses were conveniently sorted and given by {-10, -9, -8, -7, -6, -5, -4, ...}, then the 98.0% ES wants the conditional average of the 2.0% tail, which is neither the average of the worst five or six but rather it is: [1/252*(10 + 9 + 8 + 7 + 6) + ([2% - 5/252]*5)] / 2.0% = 7.976 ... just slightly below the 8.0 that is the 98.0% ES if T = 250.

@Jaskarn & @nikic are you sure it was 98.0% and T = 252? because that would require a more difficult calculation than simple average of worst 5 or 6? Much easier is 98.0% ES and T= 250 because that is "squarely" the simple average of the worst five ...

 

[christiano86]

I think the problem with quadratic programing answer was something in the answer saying that the variance is well estimated. But variance is an input is not estimated in the quadratic programing.

I did the same for this question. I hadn't read that content in too much detail, but something jumped up from the back of my head with the two peaks shape so went with that instinct.

There was a question on the equity volatility smile which I said that the ATM call would have a volatility greater than the out of the money call option, something like that.

I said that LTCM should have done some stress testing.

Mean reversion, long-run mean rate -> Answer = 0.26/0.72 = 0.36 - if thats right then I might have fluked it. I was just shoving numbers into my calculator to see if I could generate any of the options.

Portfolio construction techniques -> Answer = quadratic programming - I had no clue about this.

Specials spreads, i.e. is specials rate above/below GC rate, and the nature of the spread before/after auction - I think I said that the spread would increase on the run up to the auction and reduce thereafter? I figured the less time the person holding the special asset had, the more that they would need to accept a higher rate.

I'm amazed at the number of questions that have been recalled above. I am happy to see that I seem to agree with the answers which have already been given. I thought the exam was tricky, I had put a lot of work into the exam so was glad I had. I think I might have over half of them correct but with the nature of those qualitative questions you never know for sure if you have them right - sometimes its just on gut instinct .

K

 

[AK88]

I think mine was 4k too. 50*200-60*100=4k

Should not you take LGD into account as well, like LGD*EE*PD?