Exam Feedback November 2018 Part 2 Exam Feedback

[Linghan]

Clustering identifies outliers that do not have strong connections with the rest of the data. You can't predict PD with clustering only.

For example , you can use clustering to divide the applicants into 10 classes. But without any PD information about some applicants in each class. You will not know the PD of each class.

But the question mentioned data sources are unstructured - metadata and various other sources from app or other channels, so it should be unsupervised learning

 

[fxlprasetyo]

But the question mentioned data sources are unstructured - metadata and various other sources from app or other channels, so it should be unsupervised learning

Unstructured data has nothing to do with being able to use Supervised vs Unsupervised learning. Emails for example is an unstructured data, so is an image, etc. and one can definitely apply supervised learning for these data types.

 

[FreemanMI]

I'm not sure what the answer was but I just did DV01 times the position / dv01 of the hedge... but that seemed to easy, but my answer was there.

Yes indeed, nothing difficult there, just divide one dv01 to the second, then multiply by value we had, just not sure which one is the divisor..but i just followed the logic, as he had some bond with around 0.12 dv01, and the treasury one had something like 0.06, so it makes sense that to hedge one bp move had to be twice the hedging treasury bond

 

[mshahawy]

Yes indeed, nothing difficult there, just divide one dv01 to the second, then multiply by value we had, just not sure which one is the divisor..but i just followed the logic, as he had some bond with around 0.12 dv01, and the treasury one had something like 0.06, so it makes sense that to hedge one bp move had to be twice the hedging treasury bond

You still need to multiply with the beta

 

[voichi]

The answer wouldn't be linear programming since, the option is "Linear Programming where transaction cost can't be incorporated explicitly", which is not true, unfortunately, which is my answer as well. Many answers required a very detail knowledge of the topic, as this was a tricky exam indeed.

Indeed.

 

[runal]

Does any one remember the question of volatility smiles with 4 graphs in the options? Also what was the correct choice for it ?

 

[runal]

It was 1 year spot rate at 3% initially, and then 1 year forward rates (1 year from now) are given as 3.8% (up node) and 2.8% (down node). The compounding is to be done semi annually. Equal real world probabilities are to be assumed to get present value of 2 year ZCB.

I got 93940 as the answer. Can someone verify the same

 

[FreemanMI]

Does any one remember the question of volatility smiles with 4 graphs in the options? Also what was the correct choice for it ?

The correct one was the volatility frown where the ATM option had higher volatility due to expected huge decrease or increase of stock price after the deal if it fails or not

 

[fxlprasetyo]

Looks like most people have found the exam quite challenging and can't really say for sure if they'll pass.
What do you think would be a cutoff score to pass ?
Inputs from previous takers would be much appreciated

In the past, David mentioned about the supposedly method GARP used to calculate the passing cutoff. See this link => https://forum.bionicturtle.com/threads/percentage-of-correct-answers-to-pass.7028/

 

[jq0714]

The answer can't be clustering. The question is to predict PD, which is a y label. You should choose a regression model.

Same, don't think should be clustering since dependent variable is given thus should be supervised learning.

 

[Linghan]

Unstructured data has nothing to do with being able to use Supervised vs Unsupervised learning. Emails for example is an unstructured data, so is an image, etc. and one can definitely apply supervised learning for these data types.

Classificaition trees are applied to situation where data is divided into groups rather than investigating a numerical response and its relationship to a set of descriptor variables.....

 

[FreemanMI]

The problem is...that we are brainstorming here on the forums...but the biggest problem is, that we will never find out which were the correct answers..hehe

 

[Linghan]

The problem is...that we are brainstorming here on the forums...but the biggest problem is, that we will never find out which were the correct answers..hehe

Maybe just giving us the free point for that question will work.....

 

[Linghan]

Same, don't think should be clustering since dependent variable is given thus should be supervised learning.

But for dependent variable which is a factorial number (non-continuous) that works too for trees?

 

[fxlprasetyo]

But for dependent variable which is a factorial number (non-continuous) that works too for trees?

The choice is not only Classification. The option A) Penalized Regression and Classification. I'm assuming, this is to be used in tandem, ie. one can use classification to decide if there's a default or not, than use the penalized regression to find the PD, which is what the question asked.

 

[Navneet02]

yes i also marked option B greater than 9 million.... option C was to reject at 95% confidence and option D was to accept at 99% confidence again if i remember it correctly...

I have also marked option B with VaR greater than 9m

 

[Navneet02]

I'm not sure what the answer was but I just did DV01 times the position / dv01 of the hedge... but that seemed to easy, but my answer was there.

40m was the answer

 

[Linghan]

The choice is not only Classification. The option A) Penalized Regression and Classification. I'm assuming, this is to be used in tandem, ie. one can use classification to decide if there's a default or not, than use the penalized regression to find the PD, which is what the question asked.

Sounds like you are the correct choice!!

 

[serg1993]

The correct one was the volatility frown where the ATM option had higher volatility due to expected huge decrease or increase of stock price after the deal if it fails or not

I think the correct answer was volatility jump picture, because of the extremly uncertain situation

 

[Flashback]

I thought that volatility jump and volatility frown is the same thing.