Re: [Quantlib-users] Is anyone able to match the cds spread to price conversion with isda?

[Luigi B. <lui...@gm...>]

Hello,
    just thought I'd check.  Did either of you manage to make progress on
this?

Luigi


On Mon, Jan 22, 2018 at 10:22 PM yuzhao88 <yuz...@gm...> wrote:

> Thanks for checking horel. You are right, in this case, the dollar amount
> matches. To give you an idea why I am looking into survive probabilities,
> on Bloomberg, try using curve date as 12/29/2017 and use 12/19/2018 as the
> valuation date and leaving all other fields unchanged, the cash amount
> obtained from Bloomberg is 886319. Quantlib is giving 890,543.
>
>
>
>
>
> On Jan 12, 2018 4:51 PM, "Guillaume Horel" <gui...@gm...>
> wrote:
>
>> Where do you get the survival probabilities in bloomberg? I'm not sure
>> why they don't match exactly. I can try to investigate more, but what
>> matters is the dollar amount, survival probabilities are not tradeable.
>>
>> On Fri, Jan 12, 2018 at 4:40 PM, yuzhao88 <yuz...@gm...> wrote:
>>
>>> Thanks for coming up with the example. I just tried your code. The
>>> survival probability to 3/20/2018 is 0.990768 whereas Bbg is showing
>>> 0.9909. I actually got something very similar to what you got in survival
>>> probability and it is off by 0.0001 across the board. Any idea why this is
>>> off by about 0.0001?
>>>
>>> On Jan 12, 2018 12:51 PM, "Guillaume Horel" <gui...@gm...>
>>> wrote:
>>>
>>>> Hello,
>>>>
>>>> see the code that I used here:
>>>> https://gist.github.com/2123c9e524e2b8c50f0f75ebfffa3ae2
>>>> I get an upfront payment of 638,870 which matches bloomberg for me. One
>>>> thing to remember is that the upfront is payed T+3 so it's slightly bigger
>>>> than the NPV of the cds.
>>>>
>>>> Cheers,
>>>> Guillaume
>>>>
>>>> On Tue, Jan 9, 2018 at 8:59 AM, yuzhao88 <yuz...@gm...> wrote:
>>>>
>>>>> As an example, the survival probability I got for 3/20/2018 is around
>>>>> 0.99076538 whereas Bloomberg is showing 0.9909.  In general default
>>>>> probabilities I got are slightly lower than Bloomberg across the board.
>>>>>
>>>>> (I am using Spread 248, evaluation date 12/29/2017)
>>>>>
>>>>>
>>>>> On Tue, Jan 9, 2018, 8:29 AM yuzhao88 <yuz...@gm...> wrote:
>>>>>
>>>>>> Sorry forgot to click Reply all. Thanks for your reply.
>>>>>>
>>>>>> Yes. There discount factors match exactly with Bloomberg, but the
>>>>>> default probability does not match.
>>>>>>
>>>>>> I am using the following code to construct the credit curve.  The
>>>>>> spread I am plugging in is 248. / 10000.. The evaluation date is 12/29/2017
>>>>>> and the default probability I got on 3/20 2018, 6/20/2018 etc is lower than
>>>>>> bbg.
>>>>>>
>>>>>>
>>>>>> Calendar weekendsOnly = WeekendsOnly();
>>>>>>
>>>>>> CreditDefaultSwap::PricingModel model = CreditDefaultSwap::ISDA;
>>>>>> Real recovery_rate = 0.4;
>>>>>>
>>>>>> std::vector<boost::shared_ptr<DefaultProbabilityHelper>> instruments{
>>>>>> boost::shared_ptr<DefaultProbabilityHelper>(
>>>>>> new SpreadCdsHelper(cds_spread, 6 * Months, 1, WeekendsOnly(),
>>>>>> Quarterly, Following,
>>>>>> DateGeneration::CDS2015, Actual360(), 0.4, tsCurve, true, true,
>>>>>> Date(),
>>>>>> Actual360(true), true, model)),
>>>>>>
>>>>>> boost::shared_ptr<DefaultProbabilityHelper>(
>>>>>> new SpreadCdsHelper(cds_spread, 5 * Years, 1, WeekendsOnly(),
>>>>>> Quarterly, Following,
>>>>>> DateGeneration::CDS2015, Actual360(), 0.4, tsCurve, true, true,
>>>>>> Date(),
>>>>>> Actual360(true), true, model)),
>>>>>> };
>>>>>>
>>>>>> // build credit curve
>>>>>> Handle<DefaultProbabilityTermStructure> probability =
>>>>>> Handle<DefaultProbabilityTermStructure>(
>>>>>> boost::make_shared<PiecewiseDefaultCurve<SurvivalProbability,
>>>>>> LogLinear> >
>>>>>> (0, WeekendsOnly(), instruments, Actual365Fixed()));
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Tue, Jan 9, 2018, 4:22 AM Luigi Ballabio <lui...@gm...>
>>>>>> wrote:
>>>>>>
>>>>>>> Do you have any intermediate results you can compare?  (Schedules,
>>>>>>> annuities, coupons, probabilities...)
>>>>>>>
>>>>>>>
>>>>>>> On Mon, Jan 8, 2018 at 2:27 PM yuzhao88 <yuz...@gm...> wrote:
>>>>>>>
>>>>>>>> I did try that, but did not resolve the discrepancy.
>>>>>>>>
>>>>>>>>
>>>>>>>> On Jan 8, 2018 5:18 AM, "Luigi Ballabio" <lui...@gm...>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>> Just to rule it out, may you try increasing the accuracy of the
>>>>>>>> bootstrap? (It can be passed to the constructor of PiecewiseDefaultCurve;
>>>>>>>> the default is 1e-12, but you might try requesting a smaller error.)
>>>>>>>>
>>>>>>>> Luigi
>>>>>>>>
>>>>>>>>
>>>>>>>> On Sun, Jan 7, 2018 at 2:38 AM zcg6433 <yuz...@gm...> wrote:
>>>>>>>>
>>>>>>>>> I am wondering if anyone can match ISDA calculation with this
>>>>>>>>> change. I could
>>>>>>>>> match the pv of the coupon/premium leg exactly, but the protection
>>>>>>>>> leg is
>>>>>>>>> off by about $200 for the $10mm notional cds I tested. Anyone else
>>>>>>>>> had any
>>>>>>>>> luck or has idea what is going on?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> Sent from:
>>>>>>>>> http://quantlib.10058.n7.nabble.com/quantlib-users-f3.html
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> ------------------------------------------------------------------------------
>>>>>>>>> Check out the vibrant tech community on one of the world's most
>>>>>>>>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>>>>>>>>> _______________________________________________
>>>>>>>>> QuantLib-users mailing list
>>>>>>>>> Qua...@li...
>>>>>>>>> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Check out the vibrant tech community on one of the world's most
>>>>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>>>>> _______________________________________________
>>>>> QuantLib-users mailing list
>>>>> Qua...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>>>>>
>>>>>
>>>>
>>