Re: [Quantlib-dev] QuantLib-dev Digest, Vol 3, Issue 11

[Ferdinando A. <na...@am...>]

Hi Theo

> Has the to do list for marketmodels project been sorted out so that one can
> work whats completed and outstanding on the project.
yes, please see ql/MarketModels/TODO.txt, also attached below

> One can then look out
> the outstanding and then
> contribute to it after agreement with you guys.
As far as I'm concerned I would keep the agreement part as light as
possible: the project is going on and as soon as resources in my dev
team are available they work on whatever is high priority or suit them
best.
My suggestion is take a look at the todo list and just scratch your
itch, in the best open source tradition

ciao -- Nando

=======================

priority in first parenthesis, extimated complexity in the second one

Path generation
- (1) Sobol with Brownian bridging (multiple methods of variate allocation)
  (MEDIUM)
- (4) antithetic sampling (EASY)
- (4) moment matching (EASY)

Drift
- (1) Computation with rapid algorithm for reduced factor models (MEDIUM)
- (4) Glasserman Zhao spot measure approximation (MEDIUM)
- (4) capc, cani, and PPR terminal measure approximations (MEDIUM TO HARD)

Statistics
- (2) convergence tables (EASY)

Termination criteria
- (3) ability to terminate on time passed and select termination
criterion (EASY)

Pseudo-roots\calibration
- (1) make rankReducedMatrix work
- (1) a,b,c,d form (EASY)
- (2) calibrator to produce abcd form from market data, with displacements
  (MEDIUM to HARD)
- (2) swaption calibration using Z matrix with displacements (MEDIUM)
- (2) different correlation structures (VARIABLE)
- (2) alignment with covariance structures in legacy code (?)

Curve state
- (4) Set on coterminal swaps.  (MEDIUM)

Products
- (1) composite instrument
- (3) floorlets with one or many steps.  (EASY)
- (1) (coterminal) swaps
- (1) coterminal European swaptions (EASY)
- (1) Some real products! (MEDIUM)

Quantlib
- (1) Get LMM code in the same use format as other code via instrument classes
  creating objects from new hierarchy as necessary (MEDIUM)
- (3) check why Duff's device is ineffective
- (?) relevantRates

TEST
- (1) payment date different from reset dates

============================

Early Callability
- Lower bounds for early exercise
- Longstaff-schwartz\Amin (time consuming but not hard)
- Anderson method for Bermudan swaptions (time consuming but not hard)
- Jackel method for Bermudan swaptions (time consuming but not hard)
- Rogers\anderson-Broadie\Joshi methods for upper bounds  (time consuming )
- Jamshidian method for upper bounds (MEDIUM )
- Upper bounds for early exercise

Evolvers
- Coterminal Swap-rate market model (MEDIUM TO HARD)
- Coinitial Swap-rate market model (MEDIUM TO HARD)
- CMS swap rate market model (MEDIUM TO HARD)
- Normal Forward-Rate Market Models (?)

Scripting
- Ability to specify new pay-offs and price them without code recompilation.
  (HARD)

Variance reduction
- Control variates, importance sampling (MEDIUM)
- Randomized QMC (HARD)

Greeks
- More sophisticated Greek methods (HARD)

Calibration
- Simultaneous calibration to caplets and swaptions (HARD)