Exam Feedback May 2016 Part 2 Exam Feedback

1) Half life of Vasicek - independent of the original distance
2) JP Morgan - ignored & raised risk limits
3) Market maker earns illiq premium even after all the costs
4) Fundamental review of trading book - can shift assets from trading to banking (vice versa) only under strict circumstances
5) NSFR (currently at 1.05) decreases when wholesale loan matched dollar for dollar with mortgage loan
6) RAROC - 8.8%
7) Liq-adjusted Var: difference between exogenous vs constant spread
8) Duration mapping: increase in interest rate reduces VaR
9) Stress testing: -2% vs -2.7%. I chose the latter. Not sure
10) Illiquid returns underestimates beta
11) How to account for lliquidity when running a model (aggregating data, lags on dependent variables, etc.)
Other options were wrong. (unfiltering was described as opposite). Hesitated between aggregating data and lag on dependent V.
12) Market risk charge using stress VAR
13) Change in CVA given the 3 years of pd, EAD, LGD values. In the second year there was a collateral of 12%
Reduced LGD by proportion covered by collateral when calculating new CVA starting from 2nd yr
14) CVA where discount factor LIBOR was regressed against another factor
15) Cyber security, what should be first step of CEO? Classify info value
16) Q-Q plot: Fat tails. I got this wrong
17) Calculate DD using KMV model (long formula) and map to respective probability to default bucket. 2.2 or something (use long formula)
18) Optimal allocation among managers using information ratio and TE and assign the left over to the index.
6% or something.
Use the ideal allocation formula: wi * TEi = (IRi / IRp) * TEp

19) Expected shortfall using 96%-99% VAR value
20) One question on netting factor calculation
21) One question on netting and other without netting.
22) Bond valuation using binomial tree
23) Wrong way risk for both options
24) Operational risk capital charge using basic indicator approach. (average the positive years and multiply by 15%?)
25) One where Frechet distribution was the answer (Extreme Value Theory)
26) One where I selected negative binomial distribution for frequency, lognormal for body and pareto distribution for tail.
I didn't see poisson, log, pareto as an option.
27) Central Clearing party where I selected 50 million as left over contribution share after one party defaulted.
28) One where an option was increase in default affects senior more
I chose the same answer but not sure about this. Anybody confirm?
29) Definition of step up
I still don't know step-up. loll anybody know the answer?
30) What leads to decrease in credit of junior tranche - either margin step-up for all tranches or increase in notional of senior: increase in notional of senior because of less OC. Margin step-up only enhances credit?
31) LIBOR 1 year forward calculation question
Spent too much time on this Q.. C and D were very close to each other, so I guessed 'D'
32) 33) 34) 35) MVAR CVAR Incremental VAR, etc.
There was a Q about selling a specified amount of portfolio and which factor needs to be considered. I chose incremental because I remember the amount was pretty big. Not sure.
36) Mean Reversion of 10 yr correlations
difference between Forecast following the end month and mean is less in period 1 or 2
37) Long Corn WWR (other options included short in a bank stock and long a call on oil with a plane company)
38) Exception Problem (fail to reject? 95% VaR at 99% confidence)
Fail to reject
39) Last question - liquidity decrease will increase model risk
40) Prepayment Mortgages (PSA, etc)
SMM
41) Initial Cost of CDS A is higher than B (three different curves of probability of default were given and there were several options relating them)
Is this the answer? Why is higher PD after 1 year wrong?
42) Ho-Lee drift - 2 period interest. Calculate what is the minimum volatility that will equal to rate 0% in the lowest node
r0 + (c1 + c2)dt - 2*sigma*sqrt(dt) = 0 solve for sigma
43) Huge Case #1 (one of the options was pension fund had RWR wi(th the company)
No idea
44) Var Position 100000 when EUR were removed
709k?
45) Var two stocks one with with sigma 8%, other with sigma 6% and correlations equal to 65% and 70%, respectively. Which to choose, X or Y?
46) Stress scenario developed by regulators or he can use his own parameters? (house price index exposure and domestic gdp)
47) Probability of default (AB, AC, AD) correlation matrix 0, -0.5, 0.5 (this was the first question)
Highest joint PD - highest correlation
48) WCDR (10%, 15%, 25%?)
10%. One of you mentioned mat adjustment but that formula is so long with many constants so that it's not likely.
49) Put & call delta what happen when the stock price goes up
Both inc
50) Central Bank lending or regulation (which one would best suit the situation)?
One where banks refrain from lending to one another and stock cash
 
Last edited:
51) property of ES (under normal distribution ES = 97.5% VaR?)
The first option of 95% VaR = 97.5% ES was wrong. I don't remember the answer.
52) calculate VaR with collateral
53) question about CCAR...would equity market value aggressive or conservative capital buffer? ( I put aggressive buffer that takes advantage of the resubmission program)
Use mulligan and pass
54) how does an increase in a company's asset volatility impact its senior debt, subordinate debt and equity. (I put increase in equity and subordinated debt but decrease in senior debt)
increase in equity and subordinated debt but decrease in senior debt
55) one question about best practices after the crisis..can't recall what exactly the question and answer was
Transparency of OTC markets? I don't even remember.
56) how to tackle historical simulation with not enough data (I put bootstrapping)
57) duplicate
58) calculate CS of a five year corporate bond
59) question about liquidity dimensions: immediacy, resilience, depth etc
60) question about hazard rate, recovery rate and CS (I put the impact on hazard rate due to increase in CS will be partly offsetted by the decrease in RR)
61) delta, gamma, theta, vega of call & put - chose delta of call increased more than put decreased. Other options didn't make sense because they were same distance apart from ATM
62) 99.5% ES from 1000 trading days and 20 worst returns. Since 995th worst return is 99.5% VaR, average 996, 7, 8, 9, 1000th worst return
63) The long-ass case study question. Effect of increase of 200bp in interest rates. I chose decrease in MBS reduces counterparty risk of the hedge fund.
64) T-bill rates appropriate for long-term derivatives

Feel free to add more or disagree with above answers.
 
Last edited:
The 4 options were
Smm 0.5
Smm 0.75
Psa 90
Psa 115

Smm 0,75 : (1-(1-0.75)^12) = 8.64% > 115%*6

Gosh for this one I mistakenly was solving for "(1- SMM)^(1/12) " instead of "(1- SMM)^12 "...hence was getting very small values for first two options :( :( :( /...Silly silly mistake..
 
The exam was tough...
I left the exam site within 2 and a half hours, since I did not expect
reviewing further will enhance the result. It was simply too difficult.
20% no idea of correct answer, 30% answered correctly and
the rest is a guesswork with intuition. I think my score is around 55-65%.
 
The exam was tough...
I left the exam site within 2 and a half hours, since I did not expect
reviewing further will enhance the result. It was simply too difficult.
20% no idea of correct answer, 30% answered correctly and
the rest is a guesswork with intuition. I think my score is around 55-65%.
Still impressive that you were able to finish the exam in 2.5 hours.
 
Thanks, but this is the second trial for me. First time was 3 years ago, and I failed.
3 years ago I left the room after 2.5 hours, but this time, with slightly
more confidence.
 
having done a preliminary analysis of questions attempted...I attempted all 80 with 1-2 guesses..expecting 55-60 questions right with a WCS of 50 right answers....hope to get thru...what shall be the cut-off..worried..
 
This was a lot harder than Level 1 and much more difficult than GARP practice exams. More on par with BT practice questions. Some topics tested that didn't feature large in the material at all. The qualitative questions were also difficult, often testing several concepts within a single question, so if you didn't know exactly, you had to guess. Math was hard and calculations required long and not straight-forward. A tough exam! No idea if I have a chance to pass. Took it in London and didn't notice too many no-shows or people leaving early. They combined it with ERP exam takers, too, which was a bit odd.
 
Even the top most scorer could have only attempted 80,and with so many close calls,would easily get 15 wrong so 65-60 would be a top 5% score and not any lower.Now we have to consider the fact that many people would have have genuinely attempted 65-70 odd to begin with,and also that there would be absentees/people living real early and also people who appeared for both parts on a single day and flunk one or both(P1 flunk==P2 invalid).Add to that the 50-55% pass rate.So 40 seems a reasonable cutoff(could be a notch lower).Views most welcome.
 
Even the top most scorer could have only attempted 80,and with so many close calls,would easily get 15 wrong so 65-60 would be a top 5% score and not any lower.Now we have to consider the fact that many people would have have genuinely attempted 65-70 odd to begin with,and also that there would be absentees/people living real early and also people who appeared for both parts on a single day and flunk one or both(P1 flunk==P2 invalid).Add to that the 50-55% pass rate.So 40 seems a reasonable cutoff(could be a notch lower).Views most welcome.
absentees and people not passing part one, are not part of the statistics. IMHO, people not passing part one do not seem that likely to pass on the part two exam and are not part of the population to be graded against on the curve. That's why I believe the cutoff is probably between 50 and 57. Even a participant with limited preparation (only exam bootcamp cram) will have 35 of the 80 questions correct.
 
Na, 40 seems too low. Would assume the cut off to be around 60%, give or take 5% depending on how well/bad the results were for the whole population. This exam was probably one of the harder ones (in my opinion only), so a lower cut off would seem plausible.

Anyone found the exam relatively trouble free? Or as straight forward one can expect from these type of exams? Last year, quite a few said it was not too bad, but i've not read such a comment for this year's paper. Would be good to know
 
One aspect is the number of questions outside of all study packages(5-6 atleast),where even the most prepared folks would need to half-guess.Plus not may would have genuinely attempted 77-80.And even if we take the absentees out(not sure if that's valid),still more than half would be passing.So yes,maybe 40 is a bit low but still when we say cut-off,or minimum passing marks,doesn't seem that'll jump 45.Just an opinion.
 
Do we need to score section wise or combined score ? As in, if we score well in some sections and not well in others will we still stand a good chance ?
 
Top