Exam Feedback FRM Part 1 (May 2014) Exam Feedback

Roshan Ramdas

Active Member
.....

I strongly felt that none of the options were correct, as the return on 40 M portfolio was 8 % and the same on 60 M portfolio was 9 % . Which gives average return of 8.6. Hence at-least one of the options had to have mean return off 100*8.6. i.e. the mid point of the +/- Sigma range. I checked for it but did not get it. And then decided to guess "A" and move ahead.

I could be wrong here but I think I saw 2 options
1 with a mean of 100 +/- (2 * Sigma)
and the second with a mean of 100*1.086 +/- (2* Sigma)
Gosh,....I am now getting a feeling that my memory may no longer be reliable :)
 

Wolf

New Member

Plulutes, Babak083,
I too took Nov 13 Part I and although I found it tricky, I nevertheless found May 14 trickier especially in the Quants area.
Regarding your comments I couldn’t agree more. i.) no shame in re-sitting ii.) I have benefited more studying 2nd time round, especially with BT material (compared to Schweser first time round).

I too skipped a number of questions (10-12) hoping to have time to review them at the end, which I didn’t. I too wish I had more time to review in greater detail. What I found most frustrating, on personal level, was that I skipped questions which was bread and butter to me during the mocks but I just couldn't answer on the day i.e. my calculations were incorrect. Referring to percentages, I think I answered 50% solidly (let's say I got 40 correct), yet 30% reasonably guessed (let's say i got 15 of these correct) with 20% as a flat out guesses (let's say i got 5 of these correct).
In terms of knowledge and answering questions, I did do better than in Nov 13 but as mentioned I just didn't have time to review the skipped questions which I think will cost me.
I have benefitted immensely from the BT material, but feel that based on the outcome of Nov 13, I still haven’t done enough. :(
Regards
Wolf
 

plulutes

New Member
Plulutes, Babak083,
I too took Nov 13 Part I and although I found it tricky, I nevertheless found May 14 trickier especially in the Quants area.
Regarding your comments I couldn’t agree more. i.) no shame in re-sitting ii.) I have benefited more studying 2nd time round, especially with BT material (compared to Schweser first time round).

I too skipped a number of questions (10-12) hoping to have time to review them at the end, which I didn’t. I too wish I had more time to review in greater detail. What I found most frustrating, on personal level, was that I skipped questions which was bread and butter to me during the mocks but I just couldn't answer on the day i.e. my calculations were incorrect. Referring to percentages, I think I answered 50% solidly (let's say I got 40 correct), yet 30% reasonably guessed (let's say i got 15 of these correct) with 20% as a flat out guesses (let's say i got 5 of these correct).
In terms of knowledge and answering questions, I did do better than in Nov 13 but as mentioned I just didn't have time to review the skipped questions which I think will cost me.
I have benefitted immensely from the BT material, but feel that based on the outcome of Nov 13, I still haven’t done enough. :(
Regards
Wolf

I had a few that I skipped that I thought I could do but passed on them at the time in order to make more time for easier questions, but I never got back to them. The probability questions I thought were messing with my head a bit. Don't know why, but they weren't clicking in my noggin that day. Usually, I am pretty fair with them, especially Bayes, but oh well. Based upon your estimation of the exam (i.e. 50% answered correctly so you're figuring 40--safe! I like it!, etc.) which I think is a good proxy, then I think you're on the right side of a squeeker. Our practice scores were very close to each other too. I used 40%(80% of those correct) with 30% 50-50, and 30% outright guess for the first exam and it got me 3,3,2,2 and this time I'm guessing I correctly answered 50 (80% of those correct), and 35 were 50-50, and about 15 were flat out guesses (and a couple of those were in areas that I knew, but could wrap my head around how to do it with the way they asked it). So, Im thinking I improved about 8%, approx. But the thing is is that this was a brand new test and was nothing like the first one, SO WHO KNOWS! All I can say for sure is that I felt a little better than the first test, and I can't see scoring worse than previously; and I'm sure you didn't either. Anyway, keep your chin up. If the test from Nov. is any indicator, you did better than you think.
 
Last edited:

Adelaide

Member
I think the correct answer was 1.600. First time I saw question like this. I did it in some logic.

The difference to the mean in the new confidence interval was 4 times less than the previous one. That's why I think also that no. of scenarios should be 200 * 8 = 1600 to reduce 4x. But I picked up 800.....
 

Jo_

Member
Subscriber
The difference to the mean in the new confidence interval was 4 times less than the previous one. That's why I think also that no. of scenarios should be 200 * 8 = 1600 to reduce 4x. But I picked up 800.....

I don't remember the exact question, but indeed the difference was 4x, so if i'm remembering correctly, you need 4² = 16 times more scenarios then in the original case (don't remember how many there were in the first place, 200?). I believe it was the highest number from the possible answers.
 
Last edited:

amresh

Member
Subscriber
I could be wrong here but I think I saw 2 options
1 with a mean of 100 +/- (2 * Sigma)
and the second with a mean of 100*1.086 +/- (2* Sigma)
Gosh,....I am now getting a feeling that my memory may no longer be reliable :)
I might have not checked well.. :D
 

an12

New Member
for the number of scenarios the correct answer as per me should be 3200 as quadrupling scenarios will give half CI.. in question CI was 1/4 times the original.. therefore 800 will give 1/2 CI. to reduce it further one should further mutlipy it 4 times. again sqaure root of 3200 is 4 times square root of 200
 

an12

New Member
on the dollar duration question my piece is.. i considered it aas DV01 and got the answer it was thr in the answers .. so i guess it shld b correct.. the formula was (Delta BV/ delta Y) * (1/10000)
 

an12

New Member
There is one question when they gave the computed B1 and B2 t-stats value (1.64 and 2.33) and asks to compute F stats and determine if it is significant at 95% (reject and accept the Null Hypothesis that B1=0 and B2=0)..I dont know how to compute F stats from B1 and B2 so I make a guess that if B1 and B2 are both significant at 95% then F stats is also significant at 95%...Is that correct ?
for this question i took correlation as litmus test. as sqr of correlation = R sqr. therefore the R sqr came to .09.. which means the statistical significance of independent variables is too low. accordingly the f test is not statistically significant
 

an12

New Member
on the std
If i am correct, there actually was an exercise on this in David's set, which is were i based myself on. Basically, if i remember correctly, you need to chose the one for which SE times sqrt(t) is lowest. Stdev, number of scenarios and total time to run scenarios was given. So i calculated SE for each based on Stdev/sqrt(#scenarios run), then multiplies with sqrt(#time needed to run all scenarios) and took the minimum of that. However, i'm not 100% sure if i needed to use sqr(#time all scenarios) or sqrt(#time one scenario).

And I thought of you for the crack spread question indeed :) However, the one you explained was 1 output from 2 inputs, here it was 1 input for 2 outputs. Despite spotting that difference, i didn't find the correct answer either. I came up with 5, whereas the options were -10 and 10.

On the barbell one, i didn't get that one either. I'm getting the feeling there were some questons which many many candidates won't get :)



Never really thought of it that way, but that's absolutely correct! Scares me a little bit, but i remain hopeful :)
on the efficient estimator the formula as per book is one with lowest std/sqr root of time.. i m confused whtr it shld b total time or time for 1 scenario.. i finally selected option c that was one with lowest std dev per scenario
 

Adelaide

Member
on the std

on the efficient estimator the formula as per book is one with lowest std/sqr root of time.. i m confused whtr it shld b total time or time for 1 scenario.. i finally selected option c that was one with lowest std dev per scenario
was that the the on with 25% standard deviation or with 30%...? if you remember...
 

an12

New Member
wat a blunder i did i completely ignored its lambda ^T= 1/2 .. i took lamda as 0.5.. such a shame.. 4 hours r really too long to concentrate :(
Had to skip 7/8 questions while doing the exam because i had no idea how to begin with them. Got 40 minutes left in the end of the exam to have a second look, so time mangement wise, i'm happy i could finish the entire exam without having to skip questions. Wonder how other candidates did time-wise? I felt most people in my testing site were able to complete the exam, contrary to expectation.

Content-wise, i guess many of the questions were to be expected. Maybe somewhat more qualitative questions then expected (there were 2 on the new country risk chapter, btw). For future candidates, i do think it's important to stress the GARP practice exams. There were at least 3 questions which were literally copy/paste. I do think that's a nice gesture of GARP, to "reward" candidates with thorough prepration.



There is a formula to calculate F-stat from the individual t-stats and correlation. I knew it exists but I personally skipped it when i was studying because i thought it was an overexagerated formula to expect candidates to know. If GARP truely wants to be "practice-oriented", i really think these kind of questions are ridiculous. In practice, you'll get the p-value. In that sense, i liked the more qualitative questions from QA more (like the interpretation of the scatterplot and R² implications). Anyway, applying the formula on 1.64, 2.33 and 0.3 correlation gives an F-stat of 7-something, so the answer should be significant if i'm correct.



I found that a bit strange too indeed. I did: lamda^23 = 1/2 --> lamda = 0.97.. Then 0.03 times 0.97^4 was one of the given answers.

On the forward rates: i believe the currency was quoted in reverse (in David's exercises the EUR/USD was always given as e.g. 1.2. Garp now gave it as 1/1.2. That's probably why the order of the rates was inverse too. I converted the rate to the "norrmal" notation, did the "normal" calculation and then converted back. Maybe a bit of a detour, but i felt most safe like that and the answer was in the list.
 

an12

New Member
can anyone confirm the answer for following 2 ques
1. prob ques.. the one at beg.. saying joint prob of a and b is 5 percent.. and conditional prob of b on A being 0.5. prob of neither happening ..
i selected 85 % .. as A independent prob wld be 10% i.e. 0.5/ 0.05 and neither a and b= 1- 0.05-.10 = .85 ..
2. the total number of payments on tht ISDA confirmation stuff.. with qrtly floating payment and fixed payment.. options were 5, 14, 21 .. i selected 14 .. dunno if its right or wrng
can anyone pls confirm
 

hoang.vu90

New Member
can anyone confirm the answer for following 2 ques
1. prob ques.. the one at beg.. saying joint prob of a and b is 5 percent.. and conditional prob of b on A being 0.5. prob of neither happening ..
i selected 85 % .. as A independent prob wld be 10% i.e. 0.5/ 0.05 and neither a and b= 1- 0.05-.10 = .85 ..
2. the total number of payments on tht ISDA confirmation stuff.. with qrtly floating payment and fixed payment.. options were 5, 14, 21 .. i selected 14 .. dunno if its right or wrng
can anyone pls confirm
1. Yes I got 0.85% for that one as well. I used different method to derive it..tree method as I learn it in my uni.
2. I am not sure about this question as well. neva seen it in the reading..the length of the swap is 2 years with quarterly payment. so I choose 7 since the start date of the fixed leg payment seems to be off a little bit from the floating leg payment but its just a guess...
 
Top