|
From: Philippe H. <phi...@ex...> - 2021-05-05 15:26:25
|
I posted this on StackOverflow. Not a final answer, but some ideas to help debug. I would set the prepay penalty (call strike) very high so that it is always uneconomic to call, then observe/confirm that your OAS is zero. That would at least validate some of your overall setup. if it passes that test then I would incrementally make one of them economic, and try pricing the European option separately (you could do closed-form with Jamshidian Engine on top of your HW process which is affine) then see if the decompounded value of the option on the dv01 of the bond is close enough to your OAS (assuming the latter is positive). Although if you have a negative OAS with an American set of call dates, it's unlikely that it will become positive with an European call schedule. But these tests may provide some insights. Philippe Hatstadt On Wed, May 5, 2021 at 11:06 AM Jan Muller <jan...@gm...> wrote: > Hey, > > I am attempting to price a callable bond and determine the option-adjusted > spread of a callable bond under the Hull-White model. I have an example of > code attached enclosed: > > https://stackoverflow.com/questions/67370737/oas-quantlib-of-callable-bond > > Unfortunately, the results I am obtaining for OAS are always negative, > though the obtained bond yields from the same object seem reasonable. Does > anyone have any ideas of what is going wrong? > > The bond should be callable on either a) the call dates (European) at the > prices provided or b) on the coupon dates (Bermudan/American). > > Best Regards, > Jan Muller > _______________________________________________ > QuantLib-users mailing list > Qua...@li... > https://lists.sourceforge.net/lists/listinfo/quantlib-users > -- Broker-Dealer services offered through Exos Securities LLC, member of SIPC <http://www.sipc.org/> / FINRA <http://www.finra.org/> / BrokerCheck <https://brokercheck.finra.org/>/ 2021 Exos, inc. For important disclosures, click here <https://www.exosfinancial.com/general-disclosures>. |