<t>I read some papers on callibrating the CIR++, or any other model with deterministics shift, for default probability, and all of them are not too specific what data to use. If one uses just CDS, then the deretministic shift will do the work. On the other hand, using options on CDS is not suitable ...