FAQ Before Exam Market Risk Discussion forums

[Biju George]

Dear all,

I find Bionic turtle discussions mostly focused on the FRM exam .Can any one suggest few other good online discussion forums in the Market/Credit Risk BA space

Thanks in Advance

 

[bluesea]

Dear all,

I find Bionic turtle discussions mostly focused on the FRM exam .Can any one suggest few other good online discussion forums in the Market/Credit Risk BA space

Thanks in Advance

Yes, though it seems and pretty so. You will however agree with me that the FRM curriculum is practice oriented, engages a high number of globally renown practitioners and most likely covers contemporary issues in the industry. Notwithstanding, this forum I believe offers everyone the leeway to ask any question/clarification in the global financial risk domain. Regards.

 

[brian.field]

Absolutely - I would suggest asking your questions here! Perhaps we could all benefit from contemporary, industry-based concerns.

 

[Biju George]

Thanks and glad to see that we can discuss all general risk topics irrespective of FRM Exam.

Here is goes first question regarding CVA CS01 .

-- The CVA charges equation is easy to interpret when full re pricing is used
CVA Charges = EAD * LGD * PD
= LGD * ∑ exp (spread i-1 * ti-1) - exp (spread i * ti) * (EE * Di-1 + EE * Di) /2

-- But our firm calculates the CVA based on the CS01 sensitivity from CDS spreads
where
Regulatory CS01i= 0.0001 * ti * exp * (- spread *ti /Lgd Mkt) * (EE * Di-1 + EE * Di) /2
-- Trying hard to interpret the formula. I would expect Cs01 as a change in exposure value per basis point change in CDS spread
-- I would imagine Cs01 -- being calculated by bumping the Cds Spread by 1 BP and finding the change in value for your credit exposure
--If I split term by term
-- .0001 -- for converting BP spreads to absolute
-- exp * (spread *ti /Lgd Mkt) -- Survival probability
-- (EE * Di-1 + EE * Di) /2 - -average exposure
-- why multiply by t ??
-- can someone theoretically explain how this formula could be explained as a sensitivity to exposure .

Thanks in advance

 

[QuantMan2318]

Thanks and glad to see that we can discuss all general risk topics irrespective of FRM Exam.


-- But our firm calculates the CVA based on the CS01 sensitivity from CDS spreads
where
Regulatory CS01i= 0.0001 * ti * exp * (- spread *ti /Lgd Mkt) * (EE * Di-1 + EE * Di) /2
-- Trying hard to interpret the formula. I would expect Cs01 as a change in exposure value per basis point change in CDS spread
-- I would imagine Cs01 -- being calculated by bumping the Cds Spread by 1 BP and finding the change in value for your credit exposure
--If I split term by term
-- .0001 -- for converting BP spreads to absolute
-- exp * (spread *ti /Lgd Mkt) -- Survival probability
-- (EE * Di-1 + EE * Di) /2 - -average exposure
-- why multiply by t ??
-- can someone theoretically explain how this formula could be explained as a sensitivity to exposure .

Thanks in advance

@RiskGuy Superb question! I highly recommend you to ask industry specific questions as everyone gains in the process, there are a lot of practitioners here who would be more than willing to help. FYI, I am not a practitioner (yet), What I have garnered from FRM is purely out of passion and interest as I worked in a different industry. So, that aside, let me explain what I feel about this.

The Survival probability is exp*(-spread*ti/lgd mkt) however, I think your company rather than referencing it uses what is called instantaneous default probability which is ti*Survival probability which is equivalent to h*dt, therefore what your company uses is basically the same as the CVA formula with the Marginal Probability of default in the original CVA replaced by instantaneous default probability. Of course the CS01*LGD will give CVA

The link between CDS spreads and CVA has been the subject of discussion here as well, so, CVA is nothing but Spread*EPE, here is one link, hope this is useful https://forum.bionicturtle.com/threads/cva-increase-decrease-with-credit-spread.9378/

This is my opinion on this, Hope it helped and of course others may correct me if I am wrong
(EDIT: Forgot to put a minus in the Survival Probability)

 

[ami44]

CVA Charges = EAD * LGD * PD
= LGD * ∑ exp (spread i-1 * ti-1) - exp (spread i * ti) * (EE * Di-1 + EE * Di) /2

Just to be picky, the second equality is actually only an approximation, but I don't think that matters for the actual question.
But maybe more importantly the above formula seems to assume an LGD of 1. Otherwise exp(-spread(i) * ti) is not the survival probability until time ti, or you work with a definition of spread that is unexpected to me.
From what I know the CVA charge is calculated with a survival probabilities like
exp(-spread(i)/LGD * ti)

Regulatory CS01i= 0.0001 * ti * exp * (- spread *ti /Lgd Mkt) * (EE * Di-1 + EE * Di) /2
-- Trying hard to interpret the formula.

This seems to be the derivative by spread of the following CVA formula:
CVA = LGD * ∑ ( exp(-spread(i-1)/LGD * ti-1) - exp(-spread(i)/LGD * ti) ) * ( EE(i-1) * D(i-1) - EE(i) * D(i) ) /2

Deriving by the spread (sensitivity to 1bp spread change) gives you something like:
CS01 = 0.0001 * ∑ ti * exp(-spread(i)/LGD * ti) * ( EE(i-1) * D(i-1) - EE(i+1) * D(i+1) ) /2
which is pretty close to your formula. Is it possible, that you where inaccurate in transcribing the formula? Or there might be a last approximation step, that I don't see at the moment.

I hope that helped a little bit.
Feel free to ask, if not that clear.

 

[QuantMan2318]

Just to be picky, the second equality is actually only an approximation, but I don't think that matters for the actual question.
But maybe more importantly the above formula seems to assume an LGD of 1. Otherwise exp(-spread(i) * ti) is not the survival probability until time ti, or you work with a definition of spread that is unexpected to me.
From what I know the CVA charge is calculated with a survival probabilities like
exp(-spread(i)/LGD * ti)

I think he has approximated the Marginal Default Probability without the LGD

This seems to be the derivative by spread of the following CVA formula:
CVA = LGD * ∑ ( exp(-spread(i-1)/LGD * ti-1) - exp(-spread(i)/LGD * ti) ) * ( EE(i-1) * D(i-1) - EE(i) * D(i) ) /2

Deriving by the spread (sensitivity to 1bp spread change) gives you something like:
CS01 = 0.0001 * ∑ ti * exp(-spread(i)/LGD * ti) * ( EE(i-1) * D(i-1) - EE(i+1) * D(i+1) ) /2
which is pretty close to your formula. Is it possible, that you where inaccurate in transcribing the formula? Or there might be a last approximation step, that I don't see at the moment.

I hope that helped a little bit.
Feel free to ask, if not that clear.

You are right, Instantaneous Default Probability (Its theory, Gregory says so,I wonder if you guys apply it in the real world) is the derivative of the cumulative Default Probability wrt time. I think I made a mistake, My previous explain should have read something similar to instantaneous default probability, not instantaneous default probability itself. The CS01 formula that he mentioned is the derivative of w.r.t the hazard rate or spread not the time

 

[ami44]

Addendum to my post above:
I googled a bit and found the following resource
http://www.legislation.gov.hk/blis_...F2D8CC374186C8A848257B98002D7C27?OpenDocument

Which is very similiar to what I was saying above.
If you have problems getting from one formula to another, just let me know.

 

[QuantMan2318]

Nice article, have downloaded the same Thanks. Also check my earlier post, think we posted at around the same time