Just in the process of programming a calculator for the payment schedule for interest rate swaps. What I'm not sure about is the typical method of determining the payment dates, is it by counting back from the end date or counting forward from the start date. Fincad seems to do it by counting back but Quantlib defaults to counting forward. Just wondering what convention was the norm.tkx

The norm is counting forward unless you have cock dates in which case you count backward...

QuoteOriginally posted by: judoJust in the process of programming a calculator for the payment schedule for interest rate swaps. What I'm not sure about is the typical method of determining the payment dates, is it by counting back from the end date or counting forward from the start date. Fincad seems to do it by counting back but Quantlib defaults to counting forward. Just wondering what convention was the norm.tkxThere is a concept of a "roll date" which definitively defines an accrual period break. You then count backwards and forwards from this date to generate the periods. In the absence of such a field use the maturity date and count backwards. It's more difficult than you realise to get it absolutely right. You need to know whether stubs are left short or combined with their adjacent accrual periods, plus the complications of non-standard accrual dates (perhaps predating the start date to match a bond), plus holiday calendars and day roll conventions. You may also be given the data with adjusted (for holidays) dates. In this case it is quite difficult to work backwards to the unadjusted regular dates. It's quite easy for code to think a completely regular swap has fictitious stubs if it's presented with the adjusted end date.You may well be only looking at market quoted swaps in which case they'll all be "regular". What fields are you actually given to specify the swap schedules?