Re: COUP_ bases.

[Neil Booth <neil daikokuya demon co uk>]

Andreas J. Guelzow wrote:-

> Okay, perhaps then you could _explain_ the following answers (from 
> Windows Excel):
>
> COUPDAYS ("25-Oct-2019","28-Feb-2023",1,1) == 365
>
> COUPDAYS ("22-Oct-1996","28-Feb-2002",4,1) == 92
>
> COUPDAYS ("29-May-2015","31-Oct-2016",2,1) == 183
>
> COUPDAYS ("29-May-2015","31-Oct-2016",4,1) == 91
>
>
> COUPDAYS ("27-Apr-2013","30-Oct-2015",2,1) == 181
>
> COUPDAYS ("27-Apr-2013","30-Oct-2015",4,1) == 89

They're all explained by Excel's inability to work out a coupon date.
If you can't get a coupon date right, you're kind of stuffed trying to
get anything right for bonds.

Excel's "Analysis Pak" is more broken than I originally thought; I had
the impression it was just minor nits.  Seeing as they don't get
coupon dates right, they won't get anything right.  (I have strong
reason to believe that Excel's basis 0 is actually 30/360 implemented
by someone who was confused - Excel implements the mirror image of it.
This feeling is further reinforced by the DAYS360 function, which is
also b0rken.)

Neil.