FpML Issues Tracker

Summary

Zero coupon swaps with payment frequencies of 1T fire ird-2 even though the examples are correct from a business perspective.

See answer from Harry:

Marc, Lucio,

I think the use of 1T for payment frequency in this case is justified - it is consistent with use of the idiom in other contexts, and provides a clear signal that we are dealing with a zero-coupon leg.

If ird-2 prevents us from using this form, I propose that we revise the rule, either by re-expressing it such that:

paymentDates/paymentFrequency must be an integer multiple (could be 1) of calculationPeriodDates/calculationPeriodFrequency, or 1T

- or by revising the frequency equivalence definitions, as suggested by Lucio, so that 1T is considered to be a positive integer multiple (>= 1) of any frequency.

We (BNPP) encountered this issue in the SystemWire implementation of ird-2 some time ago, and (as I recall) resolved it by changing the implementation of the rule to permit paymentFrequency = 1T in conjunction with a regular calculationPeriodFrequency.

Best regards, Harry