Financial Modelling in Python

P1: JYS

c10 JWBK378-Fletcher May 12, 2009 19:2 Printer: Yet to come

Pricing Financial Structures in Hull–White 157

Before finishing this section one final point needs to be made. Some path-dependent trades

may involve path variables dependent on the pay discount factor and, for seasoned trades, the

path variables still need to be calculated even though they are in the past. For this reason,

the instances of the controller classes in the Monte-Carlo pricing framework set the

historical

df to unit.

10.3 CONCLUDING REMARKS

We have successfully applied the pricing frameworks developed in Chapter 9 to the pricing

of both Bermudan swaptions and target redemption notes. In any real business application

the main body of code would be in C++, which would then call out to Python to perform

the evaluation of the payoff. For efficiency reasons, you would not wish to do this for every

Monte-Carlo path because the cost of crossing the C++/Python boundary is too punitive –

indeed you would have to cross the boundary for every simulation (or, equivalently, path). If

instead you pass the current state of the world at a particular time, then you only cross the

boundary as many times as there are flows (for each leg). Moreover, by passing the state of the

world through gives the writer of the Python payoff the opportunity to employ parallelisation

techniques when performing arithmetic operations within the payoff class. ...