What's new

Exam FeedbackNovember 2019 Part 2 Exam Feedback

Anibal86

New Member
I think the slope would change throughout the QQ plot since both tails of the T-dist are fatter than the Normal? There's some helpful diagrams here: https://stats.stackexchange.com/questions/101274/how-to-interpret-a-qq-plot

Also had no idea on the scaling liquidity cost with VaR/notional.

The GEV question was worded very oddly--they appeared to include the u parameter (from GEV, which must be estimated) and the threshold level (from PoT, which is chosen) which I thought were mutually exclusive?
Correct!

Please find the link to this website, where you have the chance to change the degrees of freedom to see how the qq plot is drawn:
https://rpsychologist.com/d3/tdist/

Detective

Active Member
Subscriber
On the QQ Plot, I put the choice mentioned in p.1 of this thread which read something like: "the tails and central mass of the distribution exhibit different slopes".

Slope less than 1 (or different than 1) does not necessarily imply normal distribution is not valid. All a slope different than 1 implies that the standard deviation might be different than 1 (standard normal).

y=x -> standard normal, mean = 0, st dev. = 1
y=ax+b -> implies normal with mean=b and st dev. = a, but still normal if slope is not changing

Pciruna

New Member
Subscriber
On the QQ Plot, I put the choice mentioned in p.1 of this thread which read something like: "the tails and central mass of the distribution exhibit different slopes".

Slope less than 1 (or different than 1) does not necessarily imply normal distribution is not valid. All a slope different than 1 implies that the standard deviation might be different than 1 (standard normal).

y=x -> standard normal, mean = 0, st dev. = 1
y=ax+b -> implies normal with mean=b and st dev. = a, but still normal if slope is not changing
On the QQ Plot, I put the choice mentioned in p.1 of this thread which read something like: "the tails and central mass of the distribution exhibit different slopes".

Slope less than 1 (or different than 1) does not necessarily imply normal distribution is not valid. All a slope different than 1 implies that the standard deviation might be different than 1 (standard normal).

y=x -> standard normal, mean = 0, st dev. = 1
y=ax+b -> implies normal with mean=b and st dev. = a, but still normal if slope is not changing
I think it was unnecessarily tricky. Because, Indeed the tails have a diff slope from the central mass, but holistically the overall slope of the QQ plot would be less than a straight line of 1. Thank you for your response.

Anibal86

New Member
I think it was unnecessarily tricky. Because, Indeed the tails have a diff slope from the central mass, but holistically the overall slope of the QQ plot would be less than a straight line of 1. Thank you for your response.
If you refer to this website: https://rpsychologist.com/d3/tdist/
You will see if the degrees of freedom are higher, the t-Student QQ plot converges to the Normal QQ plot, but in any case, the slope for the QQ plot won't be less than one.

Last edited:

Passer-by

New Member
For the QQ plot question, it didn't specify which axis is normal and which is the empirical, but if the slope is different from 1 then it must be non-normal distribution.

Passer-by

New Member
I think this exam was hard because of a lot of reading, very distracting and hard to manage time carefully. It was so easy to lose track of time once you became stuck on one question where you were almost there, eliminated 2 choices already, but struggling between the other two, and for sure you don't want to give up and skip! It was really a pain to figure out the relationships from the wordy stories, such as A had a transaction with B, B had some transaction with C, and A had some other transaction with D, and then D and C... although by looking at the choices, I am familiar with the concepts being tested, just can't figure them out in such limited time.

Tinavit

New Member
Subscriber
There was a question about IR=0.45, BR=225 what is the correlation? I wasn't sure as the word 'correlation' threw me off, but I just calculated IC=IR/SQRT(BR)=0.003. Any thoughts on this question?
Did the same thing, not sure if it is correct...

mar

New Member
Subscriber
while we wait...if they ask you to review the exam..what are the chances that you passed or do they ask everyone?

Detective

Active Member
Subscriber
while we wait...if they ask you to review the exam..what are the chances that you passed or do they ask everyone?
What do you mean? They emailed you something?

amit.m.sharma

Member
while we wait...if they ask you to review the exam..what are the chances that you passed or do they ask everyone?
the candidate survey is for everyone...i wouldn't be surprised if HK candidates were sent the survey as well.

mar

New Member
Subscriber
What do you mean? They emailed you something?
for my Part 1 i got an email with a survey to review the exam (the difficulty, coverage of topics etc)...and i passed..so am just asking if the survey is extended to everyone who took the exam or they randomly select..or only to those who passed

Detective

Active Member
Subscriber
for my Part 1 i got an email with a survey to review the exam (the difficulty, coverage of topics etc)...and i passed..so am just asking if the survey is extended to everyone who took the exam or they randomly select..or only to those who passed
OK, should be everyone since I doubt they even graded the exams yet or decided on cutoff. I got email too.

Last edited:

Peterfcm

New Member
Subscriber
for my Part 1 i got an email with a survey to review the exam (the difficulty, coverage of topics etc)...and i passed..so am just asking if the survey is extended to everyone who took the exam or they randomly select..or only to those who passed
i'd be surprised if they even have the papers in their hands yet!

antogiar

Member
Subscriber
for my Part 1 i got an email with a survey to review the exam (the difficulty, coverage of topics etc)...and i passed..so am just asking if the survey is extended to everyone who took the exam or they randomly select..or only to those who passed
In the survey they even asked if you seat for the exam or not. It isn`t an early hint for sure

aksl

New Member
Did anyone manage to solve CPR and SMM question?
I also got confused about GEV and Pareto distribution question (mu, sigma,chi, chnages).

mar

New Member
Subscriber
Did anyone manage to solve CPR and SMM question?
I also got confused about GEV and Pareto distribution question (mu, sigma,chi, chnages).
can anyone plz recall the exact question..my memory is the worst but i think i saw something with initial amount given,then the monthly premium, then an amount for the deposit made during the month..the answers 28,*.%...29,**..% etc

frmtrial2019

New Member
Hello guys on an average how did you find the exams tough fair or easy and what is your guess-estimate of correct answers
Very Very Difficult!

krisn

New Member
can anyone plz recall the exact question..my memory is the worst but i think i saw something with initial amount given,then the monthly premium, then an amount for the deposit made during the month..the answers 28,*.%...29,**..% etc
There was a question in the prep somewhere, not sure exactly where.
But formulas I recall using were from the books, and are the following:

1. Solve for SMM: SMM = Prepayment / (balance at beginning - scheduled principal repayment)
2. Solve for CPR: CPR = 1 - (1-SMM)^12

Don't remember my numerical answer but the result I obtained with the formulas above was in the answers.