Exam Feedback May 2018 Part 1 Exam Feedback

BoobyMiles

New Member
3. Also one question was about MGRM failure, was it 'customers started to close their contracts' as the reason for its failure?
yes - i believe that was correct. they gave clients the right to cancel contracts, and when those clients took them up on the offer, they ran into cashflow problems given the maturity mismatch (funding long-term assets with ST funding/"rolling over of contracts")... could be wrong, but pretty sure that was the correct answer
 

BoobyMiles

New Member
Did the question give the odds of it being a AA, BB or C bond? I don't recall it. If it did give the odds, then I'm definitely wrong and wonder why I didn't go down that route. Unless I did and am not remembering it.

If the odds were given (say AA=50%, BB=30%, CC=20%)...is such the formula?

0.3*P(DefaultB) + 0.2*P(DefaultC) / [0.5*P(DefaultA) + 0.3*P(DefaultB) + 0.2*P(DefaultC)]

Now that I think about it I definitely used the proportions if it were provided.

if it asked the probability of a BB bond defaulting (or becoming a CC bond) within 1 year (which i believe was indeed on the exam at one point ), the trick is to remember to include the probabibility that it stays a BB bond for one year (which if i recall was ~80% and then defaults. I always forgot to add that one.

for the other one (the probability that a defaulting bond as BB or CC), if you draw out the table, it was pretty straightforward. for this question however, it asked the inverse, it gave you all the probability of not defaulting as a BB or CC, then the question asked the probability that a defaulting security was one or the other. if i recall the answer was, coincidently, also 80%.
 

BoobyMiles

New Member
Was it possible to solve this question without looking at the SMM and CPR formula? Cause I know there was a question relating to prepayment, and I got a matching answer using the NPV inputs on the calculator.

Or were there two calculation questions relating to prepayment?

the first calculator-generated answer was your "scheduled payment amount", you had to subtract that (around ~4500 i think) from the actual amount payed (around ~8000), might be off by a 0 on both those, but the answer for the prepayment amount was around ~3500 (or 35000).
 
There was a 2 tail t test. I selected, reject null hyphothesis as calculated value was lower than tvalue. Is that what anyone else selected

I believe that when you do the two-tail test you have to choose as the t-statistic the n-1 as it defines de DF. As it is a two tailed test you fail to reject when the value it´s between what i think was the case. ¿Any feedback?
 

krystynkatt

New Member
I believe that when you do the two-tail test you have to choose as the t-statistic the n-1 as it defines de DF. As it is a two tailed test you fail to reject when the value it´s between what i think was the case. ¿Any feedback?
I used the t statistic with n-1 df, and the value of the test statistic was above the critical value. i chose the answer to reject the null hypothesis (of inequality), as in schweser materials they always emphasised to choose as the null the hypothesis you want to prove false. I find this question really.confusing though, as the hypotheses were complementary, so in practice it does not matter which one you choose as null.
 

nikic

Active Member
It is quite clear that if the |TS| > |CV|, reject H0. H0 is always defined as the hypothesis containing the equal sign, be it for one or two tail tests. H1 is the hypothesis always without the equal sign.

As for the T distribution question, I honestly can't recall my answer. But if it was two tail, then the correct value would have been Df of n-1 and a significance level of alpha/2.
 
I used the t statistic with n-1 df, and the value of the test statistic was above the critical value. i chose the answer to reject the null hypothesis (of inequality), as in schweser materials they always emphasised to choose as the null the hypothesis you want to prove false. I find this question really.confusing though, as the hypotheses were complementary, so in practice it does not matter which one you choose as null.

To compute the t-statistic i aplied --- Mean - hip/ (s/sqrt(n) and for tht test (.... choose the 2,5 in each tail with n-1 DF.... as it was higher than the hip value i fail to reject .
 

nikic

Active Member
To compute the t-statistic i aplied --- Mean - hip/ (s/sqrt(n) and for tht test (.... choose the 2,5 in each tail with n-1 DF.... as it was higher than the hip value i fail to reject .

What does hip mean?

Also I seem to recall the tstat and cv were both around positive three or thereabouts.
 

ab88

Member
yes - i believe that was correct. they gave clients the right to cancel contracts, and when those clients took them up on the offer, they ran into cashflow problems given the maturity mismatch (funding long-term assets with ST funding/"rolling over of contracts")... could be wrong, but pretty sure that was the correct answer

I think it was prices moving from backwardation to contango caused margin calls in long positions and funding liquidity problems.
 
Last edited:

nikic

Active Member
There was a question on which is a source of operational risk.

The answer I selected was a rogue trader (or something to that effect). Correct?
 

ab88

Member
There was a question asking highest Expected Shortfall condition , wast it hi confidence alpha and higher sample size?

What the convexity answer for Barbell folio 14.xx or 16.xx and duration was 3.2(iam sure)??

Also, do clearing members post intital margin to default ffund? i remember this option, question was mitigating CCP risk?
 

nikic

Active Member
There was a question asking highest Expected Shortfall condition , wast it hi confidence alpha and higher sample size?

What the convexity answer for Barbell folio 14.xx or 16.xx and duration was 3.2(iam sure)??

Also, do clearing members post intital margin to default ffund? i remember this option, question was mitigating CCP risk?

I don't recall the ES question...but shouldn't it be high alpha and low sample size to achieve max ES?

Someone said it was 16.xx for the convexity. I had no idea but guessed that answer.

Initial margin definitely not posted to default fund. Correct me if I am wrong. I chose a different answer.
 

krystynkatt

New Member
There was a question on which is a source of operational risk.

The answer I selected was a rogue trader (or something to that effect). Correct?
I selected this answer as well. It is quite straightforward operational risk event example. Internal fraud (hence unauthorised trading) is one of basel op risk categories
As far as expected shortfall is cncerned, I guess it would be lower alpha and lower sample size, as you would take average of higher and higher losses; I chose other answer though due to some confusion
 
Last edited:

ab88

Member
I don't recall the ES question...but shouldn't it be high alpha and low sample size to achieve max ES?

Someone said it was 16.xx for the convexity. I had no idea but guessed that answer.

Initial margin definitely not posted to default fund. Correct me if I am wrong. I chose a different answer.

yes right lower aplha, but why smaller sample size?
for barbell i marked 14.xx
guess ccp answer is wrong for me

There was question asking effect of beta and volatility on ratios? remember that?
I think Gold margin answer was -22000at end of day. initial margin was 32000, maintenance was 24000 and first day loss 54000, so intital margin of 32000 falls below maintenance 24000 hence margin account value is -22000. Correct me if wrong, followed GARP book example. Variation margin here seems to be 46000?

One was on increasing/dereasing Monte Carlo Simulations, n-500,1000,2000,4000?
One was adjusted Rsq can be negative
What was the answer for lumber , two factor model was given and it was deterministic, what was the aswer for that 6.5 and 6.55% thats it?

i have listed around 70 questions..most of them discussed here.
 
Last edited:

krystynkatt

New Member
For monte carlo, i chose 8000, and for lumber i got 6.5 and 6.55 and that one model predicted better than the other or something similar
I remember i forgot fra valuation formula and finally took some guesses, anyone remembers their answer? I chose something around 57k
 

nikic

Active Member
For monte carlo, i chose 8000, and for lumber i got 6.5 and 6.55 and that one model predicted better than the other or something similar
I remember i forgot fra valuation formula and finally took some guesses, anyone remembers their answer? I chose something around 57k

Wasn't there only one answer with 6.5 and 6.55? I chose answer A.

FRA...can't remember answer but I got something that was a match.
 
Think that was positive..... I use the (forward in that moment - fix rate) * notional... don´t remember the answer but it was positive.
 
Top