Are these topics still relevant for 2012 part 1?

[sleepybird]

Hello David,
I've noticed that some topics in the past FRM are no longer in the 2012 curriculum, or at least not part of the AIMs. Can you comment the relevance/importance of these topics and if it's safe to forget about it? Specifically,
1. The Bayes' theorem formula
2. Chebyshev's inequality
3. Multi-year Restructuring Agreement (MYRA)
3. Jarque-Bera statistic formula
4. APT: statistical vs. structural models
5. Type of data: cross sectional vs. time series (this is still in your notes but not in the video)

Thank you very much for your help.
Sleepybird

 

[Aleksander Hansen]

Random thoughts:
Takes 15 min to learn so what's the difference?

I don't know for a fact but you might get away with not spending a ton of time on 3).

Bayesian statistics is EXTREMELY important in risk and asset allocation

Jacque- Bera is very common

I don't see eye to eye with GARP on the CAPM/APT topic but you have to know what a statistical model is, what a structural model is, what an idiosyncratic (aka fantasy model) and the more logically coherent (and mathematically correct) dominant-residual, if not for the FRMthen for your own sake.

Chebyshev: what's thereto know? Time commitment equivalent to of making a cup of coffee.

Cross-sectional/time series/panel: what good is it to know about regressions, H-testing and so forth if one cannot distinguish these types of data from one another, and how to approach them methodologically?

 

[David Harper CFA FRM]

Hi sleepybird,

I agree with the substance of Aleksander's response. Of course, time is short and I understand how important it is to know what you can forget (maybe this is the key skill, knowing what can be forgotten!). I like your list, and this is not helpful, because it illustrates the fluid nature of the FRM. From my perspective, your list makes a point about the FRM; e.g., it constantly churns and leaves things in its wake; and the AIMs have always been imperfect predictors

(1) and (2): since you are aware of them, you might as well know them. The Bayes theorem is a useful application of conditional probability concepts anyway.
(3): I don't know. I frankly view GARP's relative omission of sovereign risk as a mistake (http://forum.bionicturtle.com/threads/first-impressions-2012-frm-part-1-l1-reading-list.5012/), all statements are just my opinion. Given the top-down exam methodology (ie., sourced by practitioners then vetted), and the topicality of sovereign risk, I don't view this as wasted time
(3): J-B: they won't test it, it's out of Stock Watson, they should know that, as it is sort of a specialty, you don't need to worry about it
(4 & 5) APT & data: still useful, still definitely worth knowing

I hope that helps

 

[sleepybird]

Hi David,
While below topics are in the 2012, can you advise the testability?
1. Test of differences between 2 means. You had a slide in the video but you quickly skipped it saying you'll go over it in an example later, but I never came across the example. Did I miss?
2. Using GARCH or EWMA to update COVARIANCE and also the formula for using GARCH to predict variance in the future (n+k).
3. OpRisk Distribution as in this question http://bionicturtle.com/how-to/question/oprisk-distributions-ops. I find the question very challenging and time consuming.
4. Transition Matrix as in this question http://bionicturtle.com/how-to/question/transition-matrix-credit. I could solve for 2 year period but got significantly slowed down for 3 year calculation.
5. Marginal, Incremental, Component, Liquidity VARs
6. Calculating the probability that a EUR call option will be exercised as in this question http://www.bionicturtle.com/wiki/Hull.13.08/

Thank you very much for your help.
Sleepybird.

 

[David Harper CFA FRM]

Hi sleepybird,

Okay but mere opinions, right? GARP's AIMs are all technically testable.

1. i probably meant my 2nd semester application videos (where i will focus on topics & questions rather than AIMs). Not sure, no time to check
2. yes, they have tested. although less testable than mere estimation of current volatility with GARCH/EWMA
3. yes, they like to test
4. don't worry about 3-year. Too tedious for actual exam. But transition matrix is a favorite
5. yes, they like to test
6. low testable (but I really don't know)

Thanks,

 

[sleepybird]

Thanks.

 

[Aleksander Hansen]

Just to clarify David,

Are you here referring to Part 1 and Part 2?
I would think that for
Part 1:

1. yes, but not necessarily testable
2. yes, but as you mention less likely than estimation of current vols
3. somewhat surprised to see that they like to test this for Part 1? I would expect few questions on this...?
4. same as you
5. Only for Part 2 from what I can infer?
6. yes, but low testability

 

[David Harper CFA FRM]

Aleks, sorry, I admit i did not go to the top of the post and so I was not mentally distinguishing between P1 and P2. I do basically agree with you:

1. P1 only. mean difference testable but i'd be surprised (low is my guess).
2. agree
3. OpRisk distribution is Part 2 only (yes, Hull has an awkward chapter in P1, but this is P2)
4. EITHER P1 or P2 is my experience with the transition/migration matrix (it's "in bounds" P1)
5. Yes, analytical portfolio VaR is squarely P2
6. As BSM is P1, yes but low testability

Thanks,