Calculating market value of a vanilla swap at a later date in QuantLib

Question

I am following the cookbook example for pricing a Vanilla Swap in QuantLib Python, given here.

Now let's assume that a week passed, and we are trying to calculate the mark to market value of the same swap. After a bit of looking around, I have found this thread, which leads me to believe that in order to re-price a swap at a later date, one needs to change the settlement date to get the updated market value of a swap. This does not seem right, since changing the settlement_date for the swap would result in shifting the cash flow schedules as well.

What I am looking for is to keep the swap definition (cash flow dates, maturity etc), load in a new yield curve at a later date, and re-price my swap using the new yield curve.

Can anybody show me how to do this?

Thanks!

Answer 1

You can keep the Swap instance you created, which contains the swap definition, and change the evaluation date and the curves you're using.

The evaluation date is simple: in Python,

Settings::instance().evaluationDate = new_date

will do the trick.

As for the curves, you'll have to change the declaration of discount_curve and libor_curve so that they use YieldTermStructureRelinkableHandle instead of YieldTermStructureHandle. This will give you the possibility to relink them to a different curve. Your modified script will be something like:

# as in the original notebook: create the swap,
# the curves for pricing on the first evaluation date, etc.

print swap.NPV()   # as of the first evaluation date

Settings::instance().evaluationDate = new_date

libor_curve.linkTo(
    # the curve for the new evaluation date
)
discount_curve.linkTo(
    # the curve for the new evaluation date
)

print swap.NPV()  # as of the new date, using the new curves.

In short: keep the swap, change everything else as required. The swap will recalculate accordingly.

One thing you might need to be aware of: if any of the non-expired floating coupon fixed before the evaluation date, you'll have to load the fixing of the underlying index. The interface to do that is, e.g.,

libor3M_index.addFixing(date, value)