Exam Feedback November 15, 2008

[[email protected]]

The day is over.
The day for which we waited with lot of anxiety,enthusiasm,skepticsm,optimism,stress,hardwork and lot more...
First of all best of luck to all for January 20, 2009 (result day).

Thanks a lot DAVID for all your support & guidance!

FRM 2008 was really good & tough test.
It turned out to be concentrating more on VaR concepts.
Morning session was little lengthy.
Afternoon session had many descriptive numericals.
Was surprised to see questions on Binomial/Poisson distribution!
Good coverage of Basel.
Many new readings were not tested at all.
Ques were mostly based on real world applications.

In the end I felt I lacked practise of solving questions. May be preparation material could have been more exam oriented!
Why doesnt GARP issue/provide recent exam papers / sample questions covering current syllabus??

David,
request you to put some light on-
1. Passing criteria/ Quartile system
2. Number weightage
3. Role of sectional cutoff??
4. Grading system
5. How GARP is dealing with timing difference of diff countries


enjoi everybody!
~anil

 

[David Harper CFA FRM]

Anil,

Thank you for giving feedback...

"Was surprised to see questions on Binomial/Poisson distribution!"
What? These were removed from syllabus in 2008 (replaced by sampling distributions), I know we had indirect exposure to Binomial. But are you saying the Poisson was mathematically tested??

"Good coverage of Basel."
Can you give exampes, how much detail. Did they ask question(s) like "What is charge for A-rated corporate exposure of $10 MM." I'd love to know what kind of Basel questions?

"It turned out to be concentrating on VaR concepts."
I'd love examples...

"Many new readings were not tested at all."
Can you elaborate, this seems very important. Another customer wrote me that the following were not tested. Is this true?

1. DB LDA @ Work: please tell me they tested this reading, this is the OpRisk reading with more AIMs than any other OpRisk reading...
2. Andrew Lo's hedge fund replication

Can you please cite other readings that were not tested?

"Why doesn't GARP issue/provide recent exam papers / sample questions covering current syllabus??"
I don't know...It is feedback I have been giving them. I will give them more of this feedback.

"David-request you to put some light on"
I don't have new information in addition to what Daniel @ GARP told me here. The grading is ex post (after the fact) "on a curve," not by section but in aggregate; therefore, it is impossible to anticipate your result on an absolute basis (results are somewhat relative, or largely relative, I've been told).

Also, was there a question about the qualifying SPE (QSPE)? Curious to know what question that could be since it was eliminated from FAS 140 over six months ago..

Thanks, David

 

[[email protected]]

In Basel Op risk ruled.
There were questions on all 3 methods & other points.
There was numerical asking for min charge for given data of business lines.
AMA approach was tested for its applicability.
Thr was que on calc Mkt risk chg.
There were ques on std/foundation app.
Mostly testing principal rule,requirements etc.

DB Lda wasnt tested in that depth except 1-2 ques on severity distbn.
No testing of Hedge fund replication. Grinold Performance analysis, Hull CDS, continuous distbn (chi sq, t/F test), R2. It seems they ignored credit derivatives.
Ong wasnt tested in depth.
Que like which is true ELa+ELb=ELp or ULa+ULb=ULp etc.
No adjustd exposure,UL,RC!

Thr ws ques on Quants 2007,Wilmott-var based on € parameter (good one).
Thr were numericals on binomial. I cudnt read poisson full ques.
FAS 140 related to securitztn was tested.
Ths was 2-3 que on subprime friction-whch dint cause sprime, what factor led to crisis.
Sm que on even Amarnath study.

PS : All info is as far as I can recall.

 

[David Harper CFA FRM]

Anil, Thank you that is very helpful. Can i just clarify the following, because they are hard to accept:

There was a question on Amaranth? It was dropped, it is not in the syllabus. Are you sure there was an Amaranth question, it is not on the reading list...

Nothing on Grinold?

Thanks, David

 

[peter333]

Hi David

I donot know about amarnath so cannot say. Grinold yes there was a question on active beta calculation......nothing to calculate about poisson....i am sure there was a question on SPV but may be also on QSPV (50% doubt)....VaR was everywhere....looked like they wanted to be 99.9% sure that you know it..... Actual calculation of capital charges even for market risk.....J period Var....one question whether hedge fund returns are correlated (i) among themselves (ii) with stocks ..(would you help on this one?)....DB LDA at least 3 question e.g. severity and frequency distribution....hypo test...r2.... Var questions included marginal, conditional, jperiod (4/5 times), cannot remeber which aim was not tested in perspective of var....one question seemed stupid (sorry, strange is more appropriate, perhaps) to me where you were supposed to tell them actual prob of a rare event from normal distribution table as well from pareto distribution (why?)....... BSM nice thing they give formulas....too much distractions...I mean tooooo much irrelevant info...duration was also much...convexity of call options on lower yield.... If it had been cfa they would definitely consider it a violation of code of conduct but I am not sure about FRM whether you can discuss actual exam questions or not. If it is violation please delete my response and if it is not then please ask me whatever anyone wants.

best ..peter

 

[vjoyram]

Indeed Amaranth was present. The difficult part is when you have questions that take 1 minute to read/digest/understand with some triple negativity. I also found some of the questions lengthy. I was actually caught by surprise in the first part of the exam and it seems like I was not the only one...

 

[john.ophof]

What was in the exam and what was not.

* The Qualifying SPE was in!!!
* Transation matrices were in The one were you must calculate the PD when jumping from state to state.
* A lot of credit model questions. Not about risktracker.
* A lot of VAR (especially scaling from 99% to 95 mixing VAR possible?)
* Number of questions about statistcal testing (confidence level increase what will happen with type 1 risk)
* Covariance calculate given E(XY) and E(x) and E(y) given
* Not a lot of cost of carry questions
* Barrings rogue trader
* Metallgeschelschaft question
* Put of call parity arbitrage replicating the arbitrage trivial question ( you could forget the calculation since the arbritrage scenrio was asked)
* EVT testing including given a confidence interval given a distribution of EVT/GBD compared to normal confidence interval
* TROR no CLNote question TROR calculating
* One cat bond question
* marginal cumultive default possibilities
* No questions about aligning SOx Basel
* No CDS spread
* No calculation of a swap
* No shifting of MR and MArginal cost curve
* Marginal VAR component VAR (I know the concept but failed on that one)
* A lot of Hedge funds (no lineair reprlicating questions)
* What happened to the quant (two questions)
* Firm wide risk (Silo approach Hub Spoke Level 1 which riks is best diversified)
* A lot of questions about figures on a hedge fund
* Couple of Bayse questions.
* A lot of Basel questions (Basic indicatior compared to Standard)
* No LDA at work questions
* Moral hazzard two questions (A life pension fund is buying a cat bond on excess death rate is it moral hazzard)
* Securitization (predatory lending question what was not a friction.....)
* Unexpected loss and EL compared no questions about UGD
* Hedging DV01, macauly duration
* Lease rate backwardation, matal gesell shaft was shifting to question, calculation the storage cost by compounding do this slightly different than Hull
* Standard error question (t-stat or z given a large sample)
* Regression question Y=a what is regressed on what and what is the error term
* Marginal default probability I watched you clip last night and know the concept but I couldn't do it in time
* Formula for Merton and formula for BSM was given in two questions and there was a critcal Z table

Reading and understanding the question was sometimes one of the problems. For hedge questions a lot of data was presented with analysis on that.

Indeed simulating the exam would be nice it is so long waiting to do it again if needed.

They mixed in some new topics like extending capital insurance to banks recently.

Hope that gives you some vision about the exam.

 

[john.ophof]

LDA was indirectly tested in some sub questions (severity and frequency questions)

John

 

[[email protected]]

JoP Covered it pretty well. The second half of the exam had very long questions (lots of reading before getting to question).

I thought the exam was somewhat difficult. It tested on very exact details on some topics and Basel II, some current events. The questions I list below are mostly the ones I didn't know or wasn't sure of.

Would love to hear others' thoughts on the exam.


There were a couple of questions on null hypothesis testing(had to figure out the t stat or z and determine whether to accept or reject. didn't have too much practice with this before exam. David, I know you covered this.).

One question you if you move from 5% (95%) to 1%(99%) confidence interval, what would happen. Type 1 error increase, type 2 error decrease and I don't remember the other choices.

-question:where you're giving the mean and sigma, upper bound and lower bound, you had to determine the alpha or confidence.

There was a bunch of VaR calculation questions. Some I couldn't quite get the answers.

-Several bond questions.
You're given Three bonds in three periods and their corresponding yield, calculate the Duration.
Had to calculate cheapest to deliver.

-Several questions on Stress testing (what makes good qualities for stress testing).

Basel II questions- asking you to distinguish difference between Market, credit, and Op VaR. Very specific distinctions.
-Give three scenarios, determine whether it was credit, market or ops risk , or legal risk
-One question asking which has the greatest diversification benefit (across risk in SAME business, across risk across different business lines, single risk.)
-which distribution most likely for op risk for the frequency and severity, respectively (think best answer is possion,lognormal).
-questions about the IRB approach (very specific)


-several questions on causes of subprime, and the frictions (question asks which was not a friction), Think you covered it well in the notes.

-calculating marginal var and component var

-a few questions on options (which strategy gives limited upside and downside risk. The answer was a call spread (buy call and sell higher strike) other answers included buying straight calls and puts with some long or short stock that is on the delta as the call/put).

-there was a question on some exotic call options where you are long it, YET short delta, how was this so. - choices included in and out, up and down call options

-question about the Fed's liquidity risk and what was affected least (choices included sovereign, counterparty, credit risk. I think sovereign risk may be the answer).

John
Forgot to say, Thank you David. You're instruction was excellent! I learned a great deal from your notes, screencasts, and the forums. Hopefully I passed and others did as well.

 

[skcd]

usually what are the max, min, average (like 80, 120 and 95 or something like that) in terms of cut off marks out of 140? Is this year just in the same range as last 5 years or better or worst?

 

[vkaul1]

I had not prepared that extensively like you guys and you can know how bad it was for me. I didn't expect it will be bad for people who have prepared for 300 hours also. Yes poisson was tested, VAR was tested in a way linking it with EDF and LGD (atleast 10 questions on VAR testing in a vague way). Amarnath was also there.

Do you guys think the passing score will be 50% or 70 out of 140 questions or else it will be too hard for me to understand if I can clear this exam ever?

 

[dennis_cmpe]

I was surprised to see the binomial and Poisson critical stat tests too, so I got caught off guard there.

I don't recall any questions concerning the Chi or F distributions.

 

[ploy_j]

Yeah, it was tough...especially the afternoon session when my brain was already not working properly. I got so tired after reading those long and lengthy questions...some of them I even have to read twice cause I lost track along the way. It's a very time-consuming exam and one really need to beware of the time and move on. I barely have time to come back to the questions I skipped and just had to try my best to guess...

Anyway, would like to thank DAVID for his effort and hardwork preparing all the notes and slides. I am very satisfied with what I learned from your course (FYI, I only studied accounting in uni and has only 1 yr experience in internal audit. So you are really a good teacher, I managed to understand most of the concepts!). When I began, I did not have any idea at all and now I am happy that I know a lot more than before (I think this is more important that the certificate ;-) )

All the best for everyone with the result. Don't get to caught up with what has been done!

 

[skcd]

Definitely great work David. I guess i have to come back here next year (not a pleasant feeling...preparing again) but i guess you got to do somethings to pass a test.

 

[[email protected]]

Now it all depends on "probability distribution", whether we fall in "fat tail" area or in "mean" location. This "hypothesis test" results will be out in january when GARP's "multiple regression equation" grading will be out & its difficult to forecast of "exponentially weighted moving averages".
To predict "forwards" of "spot" efforts is not easy because "risk free rate" is very low.
Value of result's call can't be calculated by even "BSM Model".
But we know "binomial distribution" of two outcomes.
"Value at risk" were around 13000 candidates allover.
So lets hope "probability of default" remains low & beyond boundaries of "soveriegn risk" we find "unexpected loss" minimum.
Let's do "performance analysis" & "risk budgeting" of our attempt as no "linear replication" is possible.
New reading would be "what happened to FRM in 2008".
No more "hedging strategies".
Lets "LDA at work (Lets Do Again)" ie get back to work.
In this "severity distribution" please remember there are no grading requirements disclosed same as "Basel II Oper risk AMA approach".
All the best!

~anil

 

[john.ophof]

Some other things I do remember:

* Regarding the Fed question mentioned above one possible answer was also system risk. I taught system risk was the risk of total failure effects of banking system as a whole.
* Indeed no F or Chi test restricting constraints questions on multi linear regression questions. Two (or 1) ANOVA like questions calculate R^2 e.g..
* A couple of up or down questions (binomial pricing).
* One question about ARAROC straight question.
* Dividing economic capital within the organization what is the best way to do this.
* First question was: calculate Jensen alpha straight question.
* Which one is more increasing rapidly EL or UL (M. Ong)?
* TED spread is rising, what will be the result of this?
* Daylight overdraft question was in
* Information Ratio T-statistic question when it is significant was asked.
* Calculating the marginal PD given recovery rate, PD and risk-free rate. You taught it was low testability but it was in (Saunders chapter 11)
* Use of economic capital.
* Basic indicator and standard approach for operational risk a lot of data presented and one of them was a retail bank (12% in stead of 15%) capital usage, what method would be more favorable for both banks. The other bank was a small bank.
* No questions about WACC, cost of financial distress, EMH.
* Hedge of hedge funds question.
* Barbell strategy question what will happen with the convexity
* Accrued interest question combined with cheapest to deliver (given the prices and CF)
* Garch (1,1) creditmetrics question, what is the long run average given an equation.
* FX calculate the forward rate one year from now (annual compounding)
* Calculate the Unexpected loss hiding the parameters in a long story
* no FRA or continuous forward rates questions
* Surplus at risk I am not sure of this of it was in. Probably in one of the long story questions.
* Bull call spread was presented in one of the answers.
* Delta question about options long and a negative delta call option what could be the case (binary option as reason?, or one of the up or in the like answers).
* One question of mixing in external data for operational loss (LDA at work like question).
* Type of model risk questions. Three errors were given and you have to classify in implementation, calibration.
* Another model risk error question what would not be a concern (like small things etc).

About the time you have for the exam I think most of the people were short of the time. You get final time warning signals at half hour before and 5 minutes before the end of the exam. This raise the stress level (stress testing questions were more than normally expected).
Especially in the last 5 minutes you don't calculate a question calm but easily switching to another which may result in missing most of the left questions. So if you are short in time you must deduct 10 minutes for the exam.

Your website is great and helped us a lot, thanks David!

John

 

[maitreyi mandal]

Thanks David for all the scree casts n a very useful site. It was very helpful for this tough exam.

Maitreyi

 

[Sunil Natarajan]

Hi David,
I have to say that FRM 2008 exam was one tough exam.But the experience was unparalleled.What surprises me the most is most of the the new readings in 2008 FRM's were not tested. Some questions outside AIMS were tested.
As most of them have said, the questions were very lengthy especially in the first half, the reason being that they were case study based.It took time to read the whole question understand what was given and what was asked. By the time you did this your 2 mins of time per question were over.Then you had to go to the choices and see which would be the right one.By this time you have already utilised 3 mins of time and if were not sure of the answer, you would leave it and then go to the next question. But when you came back to this question later(incase time was available) you read it again and and incase still don't know you mark some answer. But the point here is that you have wasted a good 4-5 mins time and still not been able to answer it.I would say there were some good 8-10 questions of this type in the Ist session.
You were attacked with VaR questions right,left and centre.
The new topics which were not covered were:
(a) All the three chapters of Micheal Ong.
(b) What happened to quants in Aug 2007?
(c) Replicating Linear Clones.
(d) John Hull's reading on Credit Derivatives.
(e) Chi-square & F test.
(f) Only one question on Subprime.

There were questions on Transition Matrices in Credit (eventhough there were no formal AIM on that).
Binomial Distribution etc.
The difference for this exam as compared to others is that the questions were application based. We are not aware how the testing is going to be. The samples questions in FRM 2008.2007 and 2006 are different as the testing in those exams were different as readings in those exams were different. I think the success for this exam is not how much of time you have allocated for reading the material, but how much time you have allocated for practicing innovative questions on new as well as important material.

David you done a fabulous job. I think new application based practice questions is the key.

I would say keep your fingers crossed for the exam results in Jan 2009.

All the best to one and all FRM takers.

Regards,
Sunil.

 

[vkaul1]

It is pretty good to see in the poll 8 people have said that their exam went well and 1 said it went very well. They must be brilliant and have taken maximum advantage of David's notes. When can you say that one did well in this exam expecting 70-75% of 140 or what? I wanted to ask people who did well about how much they expect?

 

[rocky420]

Can't GARP provide 1 pencil and a good eraser after charging $700 for the exam??