RE: option pricing in quantlib [Mail virus free]

[Gilbert Peffer]

>
>Hey Folks,
>         the first idea for BSMOption and its derived classes was
>indeed that
>of implementing the Black-Scholes-Merton operators. However, the
>evolution of
>the many pricers introduced has shown that to be an unnecessary
>restriction.
>Threfore for many of the derived classes the prefix "BSM" was dropped.
>My idea is now to change the whole hierarchy to the name Pricer.

That sounds good

>The base class
>Pricer should start from the present BSMOption and not from the
>present Option,

Do you plan to have an Instrument Pricer and derive from that for instance
the Option Pricer? Or should the pricer be a pure interface from which
Instruments can inherit?

>which I would transform in OptionType and include only as a data member of
>Pricer.
>The idea is that Pricer will be a collection of classes that will help in
>giving
>a price to the instrument option, not much more than a solver of
>differential
>equations.
>
>There are some open problems:
>+ How do we include the Monte Carlo pricers into this hierarchy?

Good question. I think there is a related question about the relationship
between Assets, Pricing, and Models. Pricers use Models to value Assets. The
pricing of one Asset might require collecting quantities that one doesn't
need for other Assets. This should be the task of the pricer. It can do this
using different Models it has access to. I am not sure how this is
implemented at this stage since I just started looking at the code.

>+ Should we have a parallel hierarchy for pricers depending on
>   more than one asset?
>
>         have a nice day,
>                         Marco
>
>At 03:34 PM 7/4/01 +0200, you wrote:
>>Hi everybody,
>>Is it correct that the BSMOption class implements the Black
>Scholes model (r
>>and vol constant)?
>>Gilbert
>>
>>
>>_______________________________________________
>>Quantlib-users mailing list
>>[hidden email]
>>http://lists.sourceforge.net/lists/listinfo/quantlib-users
>

[Gilbert Peffer]

Follow up on Pricing:

At the moment I am experimenting around with the finite difference and
pricing stuff. One concept to think about is that of the Payoff. Perhaps it
makes sense to explicitly have an class that implements a payoff, which can
then be used by a pricer (which implements a model) to do the calculation.
The concept of Payoff would be the same for finite differences, Monte Carlo,
and trees, though the implementation differs for each methodology. What do
you think about this?
Ciao, Gilbert

>
>Good question. I think there is a related question about the relationship
>between Assets, Pricing, and Models. Pricers use Models to value
>Assets. The
>pricing of one Asset might require collecting quantities that one doesn't
>need for other Assets. This should be the task of the pricer. It
>can do this
>using different Models it has access to. I am not sure how this is
>implemented at this stage since I just started looking at the code.
>
>>+ Should we have a parallel hierarchy for pricers depending on
>>   more than one asset?
>>
>>         have a nice day,
>>                         Marco
>>
>>At 03:34 PM 7/4/01 +0200, you wrote:
>>>Hi everybody,
>>>Is it correct that the BSMOption class implements the Black
>>Scholes model (r
>>>and vol constant)?
>>>Gilbert
>>>
>>>
>>>_______________________________________________
>>>Quantlib-users mailing list
>>>[hidden email]
>>>http://lists.sourceforge.net/lists/listinfo/quantlib-users
>>
>
>
>_______________________________________________
>Quantlib-users mailing list
>[hidden email]
>http://lists.sourceforge.net/lists/listinfo/quantlib-users
>