Exam Feedback FRM Part 1 (November 2014) Exam Feedback

[David Harper CFA FRM]

@jonkong thanks, i didn't mean to suggest you didn't know where you were . It's very interesting (I am still processing this thread). Of course @raspberry.cookie gives the justification: it's a binomial problem so technically it's fair, you can argue. But, on the other hand, it's easier if you've practiced VaR backtest questions. Because if it's the first time you've seen it, I could see how "backtest" actually throws you off. I think a lot depends on the wording, the backtest question can be worded in different ways to make it harder/easier to reveal as essentially a binomial question. But i think your comment is instructive: they "borrowed" from a P2 idea (backtest) to query to the P1 topic (binomial).

 

[hamu4ok]

@jonkong thanks, i didn't mean to suggest you didn't know where you were . It's very interesting (I am still processing this thread). Of course @raspberry.cookie gives the justification: it's a binomial problem so technically it's fair, you can argue. But, on the other hand, it's easier if you've practiced VaR backtest questions. Because if it's the first time you've seen it, I could see how "backtest" actually throws you off. I think a lot depends on the wording, the backtest question can be worded in different ways to make it harder/easier to reveal as essentially a binomial question. But i think your comment is instructive: they "borrowed" from a P2 idea (backtest) to query to the P1 topic (binomial).

Var backtesting was mentioned in Alen's book. I also used binominal probability distribution.

 

[David Harper CFA FRM]

Var backtesting was mentioned in Alen's book. I also used binominal probability distribution.

@hamu4ok Not in a way that's much relevant to this question. The mention is "Determining which methodology is appropriate requires backtesting (see Appendix 2.1)" (page 59, Ch 2), and then Appendix 2.1 skips over binomial--which is not even mentioned in the appendix--in favor of MAE (a P2 topic) and a more sophisticated view of backtest. Jorions VaR backtest (as a binomial) is directly on topic (to this question) but I think it's a stretch to say Allen speaks to this. I'm not saying it's not a valid query of binomial ...

 

[JDGutzmann]

I agree with raspberry.cookie. However, since HP-12c doesn't do 90C2 (at least to my knowledge) you had to use the full binom. formula and replace a little:
90! / ( (90-2)! 2! ) had to become ( 89 * 90 ) / ( 1! * 2! ) and follow with the rest of Binomial dist.
I got 16,54 % as well.

 

[Goldfish3]

I was just going through the L2 posts and saw post #107 which is mentioning the exact same var es question as we got. Is this even possible or is the poster in the wrong thread?

 

[Filip Nikoloski]

thought it was really easy. Think there was an error with one question

Yes, I mean the same. It was a question 77 I think. It was asked which of the measures did not use benchmark or market index for their calculations? And in the answer were: Information, Treynor, Sharpe and Sortino. Because only Information ratio use return od the benchmark, I think something was wrong here.

 

[raspberry.cookie]

Yes, I mean the same. It was a question 77 I think. It was asked which of the measures did not use benchmark or market index for their calculations? And in the answer were: Information, Treynor, Sharpe and Sortino. Because only Information ratio use return od the benchmark, I think something was wrong here.

I don't think it is wrong. I think the answer should be Sharpe. Treynor uses Beta, which is relative to the market, and both Sortino and Information ratio requires some sort of threshold/benchmark
The question that seems to confuse most is the one about buying 8,000,000 at 0.9, and for some reason 1,000,000 appeared in all the answer choices

 

[hamu4ok]

There was a qestion to derive forward rate from spot rates and convention of actual/365 was mentioned at the end. Also the qestion to make convexity adjustment of futures to make it close to forward contract. The latter did not produce any difficulty but the former took some time.

 

[jonkong]

There was a qestion to derive forward rate from spot rates and convention of actual/365 was mentioned at the end. Also the qestion to make convexity adjustment of futures to make it close to forward contract. The latter did not produce any difficulty but the former took some time.

@hamu4ok do you mind sharing your working for former? I couldn't figure this one out concretely and guessed '1y and 1.5y'

 

[LucasC]

There was a qestion to derive forward rate from spot rates and convention of actual/365 was mentioned at the end. Also the qestion to make convexity adjustment of futures to make it close to forward contract. The latter did not produce any difficulty but the former took some time.

the first question gave 3-month rate and asked for annual rate, What I did was still just multiply the quarterly forward rate by 4, even though I noticed actual/365. What should be the real "actual" number of days?

 

[LucasC]

@hamu4ok do you mind sharing your working for former? I couldn't figure this one out concretely and guessed '1y and 1.5y'

I also had no idea for this question but so far no one in the forum can really explain their answer yet....

 

[Goldfish3]

the first question gave 3-month rate and asked for annual rate, What I did was still just multiply the quarterly forward rate by 4, even though I noticed actual/365. What should be the real "actual" number of days?

If i remember correctly wasnt this the one where the 3 mo and 6 mo spots were given and the 3 mo, 3 months from now forward was to be calculated? I used 182/365 for 6 mo and tried both 92/365 and 91/365 for 3 mo to calc the forward.. I remember both answers were fitting into one choice.

 

[hamu4ok]

If i remember correctly wasnt this the one where the 3 mo and 6 mo spots were given and the 3 mo, 3 months from now forward was to be calculated? I used 182/365 for 6 mo and tried both 92/365 and 91/365 for 3 mo to calc the forward.. I remember both answers were fitting into one choice.

I used 180/365 and 90/365, not actual quite, but tried to get an approximation to answer options.

 

[Tural Hasanov]

In the question where it was asking about the standard deviation of the portfolio and didn't give any covariance or correlation in the question but the scattered plot. Because the scatters were mainly in 2nd and 4th quadrants we could guess that the correlation is negative. In the answers there were 2 answers for positive correlation and one answer for the correlation less than minus 1 (which is not possible). Therefore, I eliminated 3 answers and chose the one with negative correlation.

 

[LucasC]

In the question where it was asking about the standard deviation of the portfolio and didn't give any covariance or correlation in the question but the scattered plot. Because the scatters were mainly in 2nd and 4th quadrants we could guess that the correlation is negative. In the answers there were 2 answers for positive correlation and one answer for the correlation less than minus 1 (which is not possible). Therefore, I eliminated 3 answers and chose the one with negative correlation.

OMG I never thought of that. Working out the correlation instead of trying to guess the negative correlation... I chose A anyway. 1.10% if I'm not wrong. hope this is your choice as well

 

[Tural Hasanov]

OMG I never thought of that. Working out the correlation instead of trying to guess the negative correlation... I chose A anyway. 1.10% if I'm not wrong. hope this is your choice as well

I am not sure about the number (as i didn't calculate the answer just chose it by eliminating others), I think it was B in my question. Calculating stdev with just 0 and -1 correlations was enough to get the corridor for possible answers and solve the question.

 

[jonkong]

I am not sure about the number (as i didn't calculate the answer just chose it by eliminating others), I think it was B in my question. Calculating stdev with just 0 and -1 correlations was enough to get the corridor for possible answers and solve the question.

I had a similar idea as yours and whittled down to A or B, of which I eventually answered B after some number crunching.

 

[raspberry.cookie]

anyone remembers the question about what a company should do: Monte Carlo ; use Poisson distribution to calculate ; use information of other companies...?

 

[Babak083]

I think they asked us about operational risk loss data estimation, my choice was where it states that company can use external information and scale it:
(Estimated Loss for Bank A= External Loss Bank B * (revenue Bank A/revenue Bank B)^0.23

 

[Babak083]

I had a similar idea as yours and whittled down to A or B, of which I eventually answered B after some number crunching.

Do you remember given Std??