strange bond behavior??

[laotze00]

Hi, I noticed some strange behavior with the bond FixedCouponBond
class. When outputing the cashflows, the first coupon is wrong. Here
is my simple python code that demonstrate the problem. Thanks a lot
for your attention!

from QuantLib import *

calendar = UnitedStates()
settlementDays = 1
issue = Date(15,11,2005)
d1 = Date(1,12,2005)
mat = Date(15,11, 2010)
coupon = [1, 0.1]
bond = FixedCouponBond(issue, d1, mat, settlementDays, coupon, Annual,
Thirty360(), calendar)
flows = bond.cashflows()
for item in flows:
    print item.date(), item.amount()

output:

November 15th, 2006 95.5555555556
November 15th, 2007 10.0
November 17th, 2008 10.0555555556
November 16th, 2009 9.97222222222
November 15th, 2010 9.97222222222

[Luigi Ballabio]

On 1/12/06, [hidden email] <[hidden email]> wrote:

> Hi, I noticed some strange behavior with the bond FixedCouponBond
> class. When outputing the cashflows, the first coupon is wrong.
> Here is my simple python code that demonstrate the problem. Thanks a lot
> for your attention!
>
> from QuantLib import *
>
> calendar = UnitedStates()
> settlementDays = 1
> issue = Date(15,11,2005)
> d1 = Date(1,12,2005)
> mat = Date(15,11, 2010)
> coupon = [1, 0.1]
> bond = FixedCouponBond(issue, d1, mat, settlementDays, coupon, Annual,
> Thirty360(), calendar)
> flows = bond.cashflows()
> for item in flows:
>     print item.date(), item.amount()
>
> output:
>
> November 15th, 2006 95.5555555556
> November 15th, 2007 10.0
> November 17th, 2008 10.0555555556
> November 16th, 2009 9.97222222222
> November 15th, 2010 9.97222222222

In what sense is wrong? If you're referring to the large amount, it's
because you defined the coupons as:

> coupon = [1, 0.1]

which means that the first coupon is 100% and the following are 10%.
If it's referring to dates or day counts, it might be some convention
to set. what did you expect?

Later,
    Luigi

[sercan atalik-2]

Hi all,
Are there any known bugs with MSVC6 when implementing PiecewiseYieldCurve? I
can not compile my basic sample code, can you provide an example? I think a
have a trouble with function template arguments.  

Thanks a lot.




template <class T, class I>
void testYieldCurve (const T&, const I& interpolator) {
...
..
.

boost::shared_ptr<YieldTermStructure> termStructure;
termStructure = boost::shared_ptr<YieldTermStructure>(
new PiecewiseYieldCurve<T,I>(Settlement,bondHelpers,
Actual360(), 1.0e-12, LogLinear()));

Handle<YieldTermStructure> ts;
ts.linkTo(termStructure);

.
..
...

}


int main(int, char* [])
{
 testYieldCurve(ZeroYield(), LogLinear());
 return 0;
}


C:\works\SXL\SXLL\QL_TEST\src\main.cpp(142) : error C2061: syntax error :
identifier 'PiecewiseYieldCurve'
        C:\works\SXL\SXLL\QL_TEST\src\main.cpp(166) : see reference to
function template instantiation 'void __cdecl testYieldCurve (const struct
QuantLib::ZeroYield &,const class QuantLib::LogLinear &)' being compiled

[Luigi Ballabio]

On 1/13/06, sercan atalik <[hidden email]> wrote:
> Are there any known bugs with MSVC6 when implementing PiecewiseYieldCurve?

Yes---it doesn't compile. VC6 cannot handle it as it is, and I didn't
have time to try and find an alternate implementation. The whole thing
is #defined out when using VC6.

Sorry,
    Luigi

[laotze00]

Thank for pointing that out. I might misunderstood the convention
here. I was just following the example in
/QuantLib/test-suite/bonds.cpp:

              FixedCouponBond bond(issue, dated, maturity, settlementDays,
                                   std::vector<Rate>(1, coupons[k]),
                                   frequencies[l], bondDayCount,
                                   calendar, convention, redemption);

How would I set the convention if I want it to refer to the "1" to
refer to dates? Thanks a lot!


On 1/13/06, Luigi Ballabio <[hidden email]> wrote:

> On 1/12/06, [hidden email] <[hidden email]> wrote:
> > Hi, I noticed some strange behavior with the bond FixedCouponBond
> > class. When outputing the cashflows, the first coupon is wrong.
> > Here is my simple python code that demonstrate the problem. Thanks a lot
> > for your attention!
> >
> > from QuantLib import *
> >
> > calendar = UnitedStates()
> > settlementDays = 1
> > issue = Date(15,11,2005)
> > d1 = Date(1,12,2005)
> > mat = Date(15,11, 2010)
> > coupon = [1, 0.1]
> > bond = FixedCouponBond(issue, d1, mat, settlementDays, coupon, Annual,
> > Thirty360(), calendar)
> > flows = bond.cashflows()
> > for item in flows:
> >     print item.date(), item.amount()
> >
> > output:
> >
> > November 15th, 2006 95.5555555556
> > November 15th, 2007 10.0
> > November 17th, 2008 10.0555555556
> > November 16th, 2009 9.97222222222
> > November 15th, 2010 9.97222222222
>
> In what sense is wrong? If you're referring to the large amount, it's
> because you defined the coupons as:
>
> > coupon = [1, 0.1]
>
> which means that the first coupon is 100% and the following are 10%.
> If it's referring to dates or day counts, it might be some convention
> to set. what did you expect?
>
> Later,
>     Luigi
>

[Luigi Ballabio]

On 1/13/06, [hidden email] <[hidden email]> wrote:
> Thank for pointing that out. I might misunderstood the convention
> here. I was just following the example in
> /QuantLib/test-suite/bonds.cpp:
>
>               FixedCouponBond bond(issue, dated, maturity, settlementDays,
>                                    std::vector<Rate>(1, coupons[k]),
>                                    frequencies[l], bondDayCount,
>                                    calendar, convention, redemption);

Oh, I see. In the std::vector constructor that would be the number of
elements, but in Python it is not required---to build a list, one just
enumerates its elements. For a bond paying 10%, you can just write

coupon = [0.1]

or if you want coupons with different rates,

coupon = [0.1, 0.08, 0.04]

Luigi

[sercan atalik-2]

Its bad, I tried with VC80 Express edition, its fine. The problem is with
MSV6. Have you got any suggestion about that, I want to solve that,
ill-fated banks are usually using VS6.





-----Original Message-----
From: Luigi Ballabio [mailto:[hidden email]]
Sent: 13 January 2006 10:24
To: [hidden email]
Cc: [hidden email]
Subject: Re: [Quantlib-users] implementing PiecewiseYieldCurve via MSVC6

On 1/13/06, sercan atalik <[hidden email]> wrote:
> Are there any known bugs with MSVC6 when implementing PiecewiseYieldCurve?

Yes---it doesn't compile. VC6 cannot handle it as it is, and I didn't
have time to try and find an alternate implementation. The whole thing
is #defined out when using VC6.

Sorry,
    Luigi