Posted by Ferdinando M. Ametrano-3 on URL: http://quantlib.414.s1.nabble.com/yield-curves-fail-on-holiday-tp14181p14186.html
On Wed, Apr 3, 2013 at 2:35 PM, Ballabio Gerardo <[hidden email]> wrote:
even if the market is closed, I should be able to calculate theoretical market values on my rate curve. Is that wrong?
yes, that's plain wrong in my book.
You should and could calculate NPVs using the last good business day (lgbd) as evaluation date. Or (more tricky but feasible) the next good business day.
In both cases you would be using the lgbd's market data, so I would suggest to go for the lgbd.
It does make very little sense to 'mark-to-market' a portfolio to a non-existent market, such as a Good Friday market.
Anyway if hard-pressed to come up with sensible values the proper way in my opinion would be to bootstrap the curves as per the lgbd, then use the qlImpliedTermStructure class/function in order to obtain a curve which has discount factor 1.0 at Good Friday. This way you would be using the forward implied curve for Good Friday as seen from the lgbd.
Last but not least on Good Friday you should be on holiday, but don't get me started on why oh why a Christian Holiday is not a public holiday in a supposedly Catholic country ;-)