Generic Calibration of one delta and several basis curves


#1

Hi All,

My first post on the forum, looking at calibration of multiple basis curves aimed at understanding curve capabilities in Strata.

Imagine a very simple Dollar ‘curve cluster’ comprising:
⦁ a strip of delta drivers (say, USD Fixing, ED Strip and AM3 straight IRS 3y-20y)
⦁ LIBOR/FF basis swaps wavg on ois leg (say only 5 … 3m, 6m, 1y, 5y and 10y)
(For now, one delta and one basis - a 3m/ois basis - in our dollar curve cluster. Will add further complexity as we build on this thread).

Questions:

  1. What is the calibration outcome internally? Am I right to assume it is
    ⦁ A term structure (strip of fwd rates) for the 3M Libor rates
    ⦁ A term structure for the 3M-OIS basis - a “basis curve” with daily spread rates
  2. Do you allow for supplying a interp/exterp spec for the outright and spread term structures separately?
  3. Would the solver be able to resolve separate delta drivers for separate segments of the curve? (say I want to use OIS swaps for first 6 months in some combination and want the 3m/OIS spread supplied in spread table to represent a “spread to the underlying OIS” … and beyond 6 months supply EDs and IRS and the 3m/OIS spread represents “a spread under the 3M”)

Cheers
HM


#2

Please see a similar curve calibration example in strata-example. The ED strip can be added by modifying the config file and adding relevant market quotes.

  1. The resulting curves in this configuration are FED-FUND curve and 3M-LIBOR curve.
  2. Yes. Please see the curve setting file.
  3. Yes, as far as the curves for underlying indices of the curve node instruments (FED-FUND and 3M-LIBOR in the example above) are defined in the curve group.

#3

Hi, thank you for your response! this is very helpful.

My second set of questions is around speed, risk localisation vs smoothness of bootstrap:

  1. With the suite of instruments provided in my earlier post, what would be the typical order of calibration time? Is ‘futures tick’ sync practical in real-time.
  2. This might be a favour I ask (I have yet to set up Strata environment) - what would the risk localisation look like on a PCHIP interp? Let’s say we’re doing a (i) 5y5y IRS AM3, (ii) 1w-fwd 5y IRS AM3
  3. Lastly, let’s say we do a 5Y outright OIS swap. Am I right to assume that we will see risks on (1) 5Y IRS for delta, (2) 5Y LIBOR/OIS Basis for spread? More generally, risks will be implied on individual instruments supplied (whether basis or outright) and not the underlying fwd-curves ?

I understand (2) is a massive favour I ask but it’ll really help to understand capabilities of the framework,

Regards
HM