Feedback on the FRM Exam Part I May 2011

[Jiew Kwang]

@alan: Thanks for your encouragement and by looking at your commitment level, i'm pretty sure you will do well in Nov.. Wow that's a lot of details your friend remembered.

On margining in OTC market, i chose netting.
On the delta hedge which your friend used simultaneous, i don't remember doing so.. But my answer was (A). Reason was that there was positive gamma and to hedge, sell options and buy shares. (can't remember the details)
On DV01 hedges, i remember doing bond value/100*100,000 scaling..
On the lowest mean reversion in GARCH model, i picked the one with the smallest gamma.
On op risk, i think it was decrease in R^2 leading to increase in estimate. (not very sure)

 

[[email protected]]

@JK, I hv posted the questions so that I can refer to the distribution/type of questions when I'll write the exam. Kinda helps. And this will help other BT members also.

Alan

 

[Prashant Raje]

Its amazing how you people remember all these questions. During the exam I deliberately try for one or more questions to remember but by the time I was outside the exam room I almost forgot, but thanks to you guys for remembering all these things.
Before coming to this discussion I was expecting myself on the fence but after reading all this stuff I am thinking of writing Nov 2011 exam.

 

[Jiew Kwang]

@alan: I'm sure people will be grateful for that.

@prashant: don't be discouraged. i think its just a small sample of people taking the FRM participating in this forums. i'm sure many feel the same. i myself am feeling it too after reading some of what alan said.. let's cross our fingers for 1 more month. =)

 

[[email protected]]

Ugh..

Just found out i got Q1 wrong

"Port has 7 yrs durations and is appropriatiley hedged with US treasury bonds (was this 3 month treasury??). what instantaneous change in the interest rates would cause an aggregate gain?" (note this isn't the q verbatim)

a) longterm rates rise short term fall (Steepen)
b) short term rates rise long term fall (invert)
C) parallel upward shift
d) none of these

I just now noticed that the yield curve pivots around the 7 yr (for some reason I picked B) i think the answer is D

Q:
What causes changes in currency exchange rate markets?
A) Money Supply
..
..
D) Inflation


I chose A), but i later realized after the test D was a better choice

 

[[email protected]]

@alan: in regards to the following question... would you answer change knowing the VaRs were estimated via linear approx ie the delta-normal approach?

"Qn Oh, I instantly loved this question when i heard it from my friend.
Given a plot of VaRs vs. Confidence level (92%, 94%, 96% and 98%)
Four possible curves were given:
1) Increasing, concave function
2) Increasing, mix of convex and concave
3) increasing, linear
4) increasing convex. (I wish I could post the plot)
Ans. answer is 4). For uniform loss distro (which is never the case), the ans. will be 3). For loss distro with more weight in the tails than the mid region, a concave plot might result (which is even more impossible)."

 

[jauharim]

Going to echo most responses. Did not look at the computer whole weekend after the exam. 40% confident. Like everybody else, wasted a good 10 min on the UL question.

Favorite tested topics: CAPM, Cost of carry model, Operational risk; cant remember others. There were atleast 1 question each on MGRM and Barings. Some other topics that were tested: hedge variance ratio, Basis risk, Sharpe, Treynor, Information Ratio, EWMA, GARCH, CTD bond, VAR, delta VAR.

Complex topics like Swap, FRA valuation, trading strategies like straddles, box spread etc. were not tested. z table was provided but I dont think it was required; unless I missed something.

BT was very very helpful.

Manish

 

[toofreak]

Hi, I prepared this long email for a friend telling him a little bit of what was asked in the PART 1 exam. I also took part 2, so if anyone is interested I can writtte what was asked there too. I think this kind of sharing is very important as the exam is very wide so you have to concentrate in the topics that really matters.

They play a lot with the lack of time, so I recommend to read the questions in reverse: first look what they are asking and then find the data on the body of the question. They give you a lot of unuseful and complicated information to distract you and make you waste time. This is important as you are tired and nervous and the questiosn are very bad written

1) FRA - easy and standard one, no tricks here
2) PUT -CALL Parity with dividends (not an usual case, they caught me here)
3) IR SWAP valuation, they ask you to valuate only 5 legs, but this excercise takes time. I didnt do it because I was out of time already. Be prepared to know how to valuate an IR Swap and a cross currency Swap.
4) Transition matrices - 2 exercises kinda tricky
5) Commodity forward - case with convenience yield and dividend yield. Very standard
6) 1 tricky question regarding GARPS code of conduct, the question was easy, but then they ask you for the “punishment” that the code indicates for cases like these (it was a person who received a gift) and I didnt know which punishment should apply.

7) Key rate duration, was there all the time. But mostly whenever they wanted to say DV01 or PVBP or just “sensitivity” they just give you the number in the form of a Key Rate duration.
8) Cheapest to deliver bond - This is a long exercise if you follow the formulas on Schweiser, but on GARPS exams they have a quicker formula that allows you to do these exercises in 5 seconds. One more time, do past years exams !
9) Delta-nortmal Var with the squared root of the variance
10) One excercice of Expected Loss (“EL”) and Unexpected Loss (“UL”). No tricks here, apply the formula right away.
11) Questions regarding Operational risk focused on very simple topics: 2 questions regarding using external data, and 2 questions regarding which distribution should you use for Frequency of losses (Poisson) and which for modeling Amount of losses (Weibull or something like that).
12) Hipothesis test - you dont have to perform any test , just know that you use the “F” distribution for multiple variable testing and the “T” distribution for single variable”
13) Hipothesis tests - Know the interpretation of R squared, how to compute it (the different formulas) and its relationship with the correlation coefficient
14) Sortino ratio, Jensens alpha and Sharpe are classics in finance and they will be in every exam you take in your life, even if it’s about cooking.
15) Probability distributions - In every exam there is a question on how to calculate probabilities using a Normal table - They want to make sure you know all the tricks on how to us the normal table. They give you a table with negative values only, so you have to know how to convert the probs for positive values of the domain (the symmetric property, very easy). And there was a question for the tired ones, to go and confuse that the Prob of (0,5) = 0,5
16) Binomial distribution - Know how to identify when they are meaning that you have to calculate a binomial prob. The most difficult part of this is that they make you think that it’s an excersise about hedge funds complex strategy and it’s actually a question about binomial distrib . Excersises are long as you have to calculate 4 probabilities, (takes about 5 or 7 minutes when the average time for a quiestion is 2,5 to perform well). On GARPs solutions for previous years exams there is a short cut on how to calculate these probs quick. READ IT .

17) Of course, a question regarding the Barings Bank and the Metaegash (i dont know the name) . The two popular cases of risk management failure. If you done previous exams this is a piece of cake.

18) LOTS of questions about the Enterprise risk management. And the Value addedd from risk Management. You have to know this reading very well, not only know it by heart, but also had performed some tricky questions before, to know what they are aiming for. The good thing it’s this is a very easy and short reading, but yes, they love it.

19) How to calculate Black sholes prices and those Brownian Motion formulas ( dont worry they give you N(d1) and N(d2)
20) EWMA and GARCH are always there
21) Foreign exchange easy problems (calculate the forward price given the domestic and forward rate and the spot price)
22) Couple of questions regarding CAPM (one theoretical question testing of you know what happens to the Possibility frontier when you change the correlations) easy exercises on the Security Market line and easy exercise on the ATP model. Nothing regarding the weird cases of CAPM at the end of the chapter, Rolls critique or zero beta CAPM.

23) Backtesting -diffucilt one, you had to be very well prepared or know that: Number of observations * alpha = number of exceptions. No book tells you this, I learned it by working in the field and it’s correct and helps me perform excersises very quick.

24) Theorical questions regarding Normal distribution issues - kurtosis (fat tails) and simmetry - very similar to previous years exams
25) Multicolinearity and Heterosckedacity . If you had read the Schwesser book at a glance you hardly get these questions. Wikipedia a little bit on these. Or read the assigned book or something. They are easy if you are an economist or stats background, if not, shweser dont help in the key point of this topic.
26) Binomial trees, two exercises very short and similar to previous years exams. If you do your homework and performed last years exams you should do these right and quickly.
27) Futures hedge ratio - a classic question with no surprises to get easy points fast
28) Basis risk, kinda tricky question - another classic topic that repeats itself each year.

29) American options, effects on dividends and early excersies - Classic question and predictable.This is a question mostly to memorize.
30) Question regarding corporate bonds - I dont remember this one, but I think it was easy.

I cant remember more questions now. But i wish someone had posted this for me before so I didnt spend time memorizing the formula of d1 or the formula for the variance hipothesis test, or the Bonds future with accrued interests ! I feel I wasted time with the wrong topics.

Good luck, and dont go without doing all previous years exams !

 

[[email protected]]

@alan: in regards to the following question... would you answer change knowing the VaRs were estimated via linear approx ie the delta-normal approach?"​

Hi asharma,

Delta normal VaR is used only when you want to reasonably estimate the VaR of one instrument whose price depends on another (which has an easily calculable VaR)..For example options VaR=delta*StockVaR.

However, the possible VaR curves in the plot were given for a certain security.

Let take a few examples:

1) what if the return distribution has a shape like normal (conventional return distro)?
That is tails have lower weights than the central region... Let the 92% VaR is 100 and 94% VaR is 120. To gather the 2% weight you need to move from 100 to 120. Now as the tails become thinner and thinner, you need to travel more to earn that 2%, say you earn the next 2% travelling 30 points. 96% VaR is at 150. The tail has become much thnner now and to earn 2% more you will have to travel 50 more points. Tht's why this curve will have a shape like A below.

2) What if the return distro has a shape like inverted normal (Never seen a loss distro like this)?
It implies that the central region has less weights and the tails have more weights. Suppose you start at the same point 92% VaR at 100. Now the tails are becoming thicker. So you need to travel lesser and lesser to earn the successive 2% weights. That's why initially you travel more (say 50 points) to earn 2% so that the 94% VaR is 150. But the tails are getting thicker. So to earn the next 2% weight, you need to travel less, say 30 points so that the 96% VaR is 190. To earn the next 2% you will have to travel even less, say 20 points. Curve C.

3) Lastly the return distro is uniform (this is also unlikely)?
In this case to earn every 2% weight you need to move an equal distance. Only then the straight line results. Curve B.

So, I guess it's not about the method of calculating the VaR. It deals with the underlying loss/return distro.

Alan

 

[[email protected]]

@alan - that definitely makes sense, However, this also makes me wonder if we all got the same exact test.... I'm 100% sure my test read it VaRs plotted were via delta-normal approach. which is why i was wondering how that changes the answer.

 

[[email protected]]

I dun know... I got this ques from someone else....

But even if the approach used is delta normal, the VaR of the underlying security will follow the shape "A"... Delta and the volatilities are only constants.

So, the VaRs will be constant*Z(alpha)*volatility and the Z(alphas) will follow curve A, since it is impllied from the normal curve.

Alan

 

[de]

Going by the comments here I'm pretty certain now that at least the questions came in different order for different people; my Q2 was the 7yr bond vs 3m Tsy Futures curve shift, and Q100 was variance of log-normal stock price.

But I haven't heard anyone describe a question I didn't get?

 

[[email protected]]

my Q100 was the question with table where we had to calculate probabilities ... I don't remember the question I was lacking of time (and tired!) but it was the one with 'California' ... So as you suggested the questions came in different order from one candidate to another.

 

[[email protected]]

Alan, that's true

thanks for the color!

 

[[email protected]]

@Prashant, Don't worry.... As someone rightly said, relative grading exms are unpredictable.... Anything is possible

@asharma, Glad to help. On the problem with 7yr duration bond, I guess the problem said appropriate hedges are in place, so that should take care of parallel/non parallel shifts and twist in the yield curve... I did not hear it said duration hedge...

@Toofreak, back-testing is a level-2 topic... no idea why they tested it in L1.

@Everyone, if the question set is just a permutation for different people, I think our US friends can benefit a lot if someone from the East posts the questions after the exam... Really would like to know what was tested in the US centers..

Alan

 

[toofreak]

alankarghosh I took the test in NY. Maybe Im confused with the Backtesting question as I also took part 2
PS I also work for HSBC !!! I will send you a message right now !

 

[Jiew Kwang]

Guys, anyway GARP has responded on facebook regarding potential errors, just thought i'd share:

It is GARP’s intention to offer a high quality examination that accurately and effectively assesses a candidate’s risk management knowledge. After every exam, GARP undertakes a thorough review of exam administration and performance. Among the many factors GARP looks at is the performance on the overall exam as well as performance on individual questions.

Any question that GARP determines to be problematic is removed from consideration to eliminate any adverse effects on candidate scores. As part of its review, GARP surveys all exam takers shortly after the exam date in an effort to better understand their exam experience. Exam takers are encouraged to participate in this survey.

Best regards,
GARP FRM

regards,
jiew kwang

 

[[email protected]]

GARP should keep these things in mind.... These are make or break when a few number of questions can turn things around...

Suppose A and B taking the exm... A knows UL, spent 10 mins trying to reach the ans... and B does not know UL, just skips the problem and marks 3 more questions than A...

Alan

 

[Jiew Kwang]

My thoughts exactly, Alan. It would be a pity to void questions that you know and like you said, spend enormous effort and time to crack. The additional drag on other questions is not justifiable.. You can never ignore the moral damper when hitting a roadblock on QN 3in an exam that has 100 QNs.

 

[[email protected]]

Totally agree JK !!

guess it was pure bad luck that the UL question showed up early in the paper for centers across India and Singapore.

@Toofreak, could you please post some L2 questions also, since I might sit for both L1 n L2. thanks.

Alan