Thirty360 problems

[Daniele De Francesco-2]

I just found out that the using the Thirty360 daycounter (US version) we get
an interesting result :

The yearFraction between March 30 and March 31 is 0.
I'm sure that can't be right. Shall we change the way the DayCounter works
by adding the following line:

if (mm1 == mm2 && yy1 == yy2) return (dd2 - dd1 ) ;

That should fix things right?
-Daniele

[Luigi Ballabio-2]

On 2004.07.23 20:53, Daniele De Francesco wrote:
> I just found out that the using the Thirty360 daycounter (US version)
> we get an interesting result :
>
> The yearFraction between March 30 and March 31 is 0.
> I'm sure that can't be right.

Yes, it looks strange, to say the least :)

> Shall we change the way the DayCounter works by adding the following  
> line:
>
> if (mm1 == mm2 && yy1 == yy2) return (dd2 - dd1 ) ;
>
> That should fix things right?

It will "fix things" in the sense that the result won't be zero.
I'm not sure that the result will be the correct one :)

Can anybody refer us to some paper detailing the convention?

Thanks,
        Luigi

[Ferdinando M. Ametrano-3]

Hi Luigi and Daniele

>>The yearFraction between March 30 and March 31 is 0.
>>I'm sure that can't be right.
>
>Yes, it looks strange, to say the least :)

the definitive answer is to be found in the ISDA documentation, anyway all
the implementations I've found of Thirty360 gives zero as result. Check
also Excel implementation. Zero happens for all months with 31 days. Unless
proven wrong I wouldn't consider this a bug.

Even stranger, if you start accruing on January 1st, March 1st account for
3 days of accrual: day360(1-Jan-2005, 28-Feb-2005)=57 while
day360(1-Jan-2005, 1-March-2005)=60

Market conventions: enjoy them!

ciao -- Nando

[Ferdinando M. Ametrano-3]

Hi all

I'm wondering if ISDA would consider giving a (paper or electronic) copy of
its documentation to the QuantLib team for free. Anyway working at ISDA?
Any suggestion for an ISDA contact? Anyone with a spare electronic copy ;-)

thank you

ciao -- Nando