Curves on bank-holidays

[Simon Ibbotson - Straumur]

Hi,

 

Does anyone have any hints for me for pricing instruments when the valuation date is a non-working day?

 

For instance, we had an EUR bank holiday yesterday, but I needed to value certain EUR denominated instruments.

 

The main issue is the EUR curve – which fails to build (requires fixings for swaps).

 

I “could” set the valuation date to be different for those instruments – but was hoping someone had a neater solution (i.e. have a means whereby the curve doesn’t fail to build).

 

Cheers,

 

Simon

 

Simon Ibbotson

Quantitative Analytics

Capital Markets

Straumur

 

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Ferdinando M. Ametrano-3]

Hi Simon

> Does anyone have any hints for me for pricing instruments when the valuation
> date is a non-working day?
> [...]
> The main issue is the EUR curve – which fails to build (requires fixings for swaps).

my favorite approach is to set the evaluation date equal to the last
good business day.

In my opinion this is the only solution that really makes sense from
the financial point of view: the market quotes you are probably using
do refer to the last good business day anyway; to apply them to a
different business day might result into relevant errors in some cases
because of calendar and market conventions (end-of-month,
turn-of-year, etc.).

Ex-post if you really care you could adjust your NPVs for the one-day
discount factor

Last but not least: build failures because of missing fixings is
revealing that you are not taking into account the fixings of the last
good business day. If you roll back the evaluation date to the last
good business day they are not required (as the system assumes they
might be not available yet) but you should take care and include them.

ciao -- Nando

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Simon Ibbotson - Straumur]

Hi Nando,

Thanks for the quick reply.

I agree that changing the evaluation date to be the last good business
day is the only solution that makes financial sense for the curve:
however, from a systems point of view this is tricky (as you have to
discriminate between different instruments, changing the evaluation date
for each).

Also, for cross-currency instruments (e.g. a simple basis swap), I have
a real problem when one currency has a bank-holiday and the other
doesn't. Ideally, I'd use the previous day's EOD curve for the "invalid"
currency and the live curve for the "valid" currency, without changing
the valuation date.

Regarding your point about fixings: the fixings in any system will be
the Ibor fixings, whereas I would prefer to use the EOD curve which I'm
passing to the curve.
I also don't usually store (as fixings) the O/N, T/N and 1W quotes,
which are also required by the QuantLib library when building the curve.


Is it possible to build a curve (i.e. derive the discount factors) "as
of" 1 day (e.g. the 30th April) and use it with a different valuation
date (e.g. 1st May): i.e. use it for forward valuation purposes?

Cheers,

Simon





-----Original Message-----
From: [hidden email]
[mailto:[hidden email]] On Behalf Of Ferdinando Ametrano
Sent: 02 May 2008 10:15
To: Simon Ibbotson - Straumur
Cc: [hidden email]
Subject: Re: [Quantlib-users] Curves on bank-holidays


Hi Simon

> Does anyone have any hints for me for pricing instruments when the
valuation
> date is a non-working day?
> [...]
> The main issue is the EUR curve - which fails to build (requires
fixings for swaps).

my favorite approach is to set the evaluation date equal to the last
good business day.

In my opinion this is the only solution that really makes sense from
the financial point of view: the market quotes you are probably using
do refer to the last good business day anyway; to apply them to a
different business day might result into relevant errors in some cases
because of calendar and market conventions (end-of-month,
turn-of-year, etc.).

Ex-post if you really care you could adjust your NPVs for the one-day
discount factor

Last but not least: build failures because of missing fixings is
revealing that you are not taking into account the fixings of the last
good business day. If you roll back the evaluation date to the last
good business day they are not required (as the system assumes they
might be not available yet) but you should take care and include them.

ciao -- Nando

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users