#include <qle/models/basket.hpp>

Inheritance diagram for Basket:

Public Member Functions
	Basket (const Date &refDate, const std::vector< std::string > &names, const std::vector< Real > &notionals, const boost::shared_ptr< Pool > pool, Real attachmentRatio=0.0, Real detachmentRatio=1.0, const boost::shared_ptr< Claim > &claim=boost::shared_ptr< Claim >(new FaceValueClaim()))

void	computeBasket () const

Size	size () const
	Basket inception number of counterparties.

const std::vector< std::string > &	names () const
	Basket counterparties names at inception.

const std::vector< Real > &	notionals () const
	Basket counterparties notionals at inception.

Real	exposure (const std::string &name, const Date &=Date()) const
	Returns the total expected exposures for that name.

const boost::shared_ptr< Pool > &	pool () const
	Underlying pool.

std::vector< DefaultProbKey >	defaultKeys () const
	The keys each counterparty enters the basket with (sensitive to)

const Date &	refDate () const
	Basket inception date. More...

Real	attachmentRatio () const

Real	detachmentRatio () const
	Detachment point expressed as a fraction of the total pool notional.

Real	basketNotional () const
	Original basket notional ignoring any losses.

Real	trancheNotional () const
	Original tranche notional ignoring any realized losses.

Real	attachmentAmount () const
	Attachment amount = attachmentRatio() * basketNotional()

Real	detachmentAmount () const
	Detachment amount = detachmentRatio() * basketNotional()

boost::shared_ptr< Claim >	claim () const
	default claim, same for all positions and counterparties

std::vector< Probability >	probabilities (const Date &d) const

Real	settledLoss () const

Real	settledLoss (const Date &) const

Real	cumulatedLoss () const

Real	cumulatedLoss (const Date &) const

Real	remainingNotional () const

Real	remainingNotional (const Date &) const

const std::vector< Real > &	remainingNotionals () const

std::vector< Real >	remainingNotionals (const Date &) const

const std::vector< std::string > &	remainingNames () const

std::vector< std::string >	remainingNames (const Date &) const

const std::vector< DefaultProbKey > &	remainingDefaultKeys () const

std::vector< DefaultProbKey >	remainingDefaultKeys (const Date &) const

Size	remainingSize () const
	Number of counterparties alive on the requested date.

Size	remainingSize (const Date &) const

std::vector< Probability >	remainingProbabilities (const Date &d) const

Real	remainingAttachmentAmount () const

Real	remainingAttachmentAmount (const Date &endDate) const

Real	remainingDetachmentAmount () const

Real	remainingDetachmentAmount (const Date &endDate) const

Real	remainingTrancheNotional () const
	Remaining basket tranched notional on calculation date.

Real	remainingTrancheNotional (const Date &endDate) const

const std::vector< Size > &	liveList () const
	Indexes of remaining names. Notice these are names and not positions.

std::vector< Size >	liveList (const Date &) const

void	setLossModel (const boost::shared_ptr< QuantExt::DefaultLossModel > &lossModel)
	Assigns the default loss model to this basket. Resets calculations.

Basket Loss Statistics
Methods providing statistical metrics on the loss or value distribution of the basket. Most calculations rely on the pressence of a model assigned to the basket.
Real	expectedTrancheLoss (const Date &d, Real recoveryRate=Null< Real >()) const

Probability	probOverLoss (const Date &d, Real lossFraction) const

Real	percentile (const Date &d, Probability prob) const

Real	expectedShortfall (const Date &d, Probability prob) const

std::vector< Real >	splitVaRLevel (const Date &date, Real loss) const

std::map< Real, Probability >	lossDistribution (const Date &) const

Real	densityTrancheLoss (const Date &d, Real lossFraction) const

Real	defaultCorrelation (const Date &d, Size iName, Size jName) const

std::vector< Probability >	probsBeingNthEvent (Size n, const Date &d) const

Probability	probAtLeastNEvents (Size n, const Date &d) const

Real	recoveryRate (const Date &d, Size iName) const

QuantLib::Real	correlation () const

Detailed Description

Credit Basket.

: A basket is a collection of credit names, represented by a unique identifier (a text string), associated notional amounts, a pool and tranche information. The pool is a map of "names" to issuers. The Basket structure is motivated by CDO squared instruments containing various underlying inner CDOs which can be represented by respective baskets including their tranche structure. The role of the Pool is providing a unique list of relevant issuers while names may appear multiple times across different baskets (overlap).

Constructor & Destructor Documentation

◆ Basket()

Basket	(	const Date &	refDate,
		const std::vector< std::string > &	names,
		const std::vector< Real > &	notionals,
		const boost::shared_ptr< Pool >	pool,
		Real	attachmentRatio = `0.0`,
		Real	detachmentRatio = `1.0`,
		const boost::shared_ptr< Claim > &	claim = `boost::shared_ptr< Claim >(new FaceValueClaim())`
	)

Constructs a basket of simple collection of constant notional positions subject to default risk only.

The refDate parameter is the basket inception date, that is, the date at which defaultable events are relevant. (There are no constraints on forward baskets but models assigned should be consistent.)

Member Function Documentation

◆ refDate()

const Date& refDate ( ) const

Basket inception date.

Loss Given Default for all issuers/notionals based on expected recovery rates for the respective issuers.

◆ attachmentRatio()

Real attachmentRatio ( ) const

Attachment point expressed as a fraction of the total inception notional.

◆ probabilities()

std::vector<Probability> probabilities ( const Date & d ) const

Vector of cumulative default probability to date d for all issuers in the basket.

◆ settledLoss()

Real settledLoss ( ) const

Realized basket losses between the reference date and the calculation date, taking the actual recovery rates of loss events into account. Only default events that have settled (have a realized RR) are accounted for. For contingent losses after a default you need to compute the losses through a DefaultLossModel

Optionally one can pass a date in the future and that will collect events stored in the issuers list. This shows the effect of 'programmed' (after today's) events on top of past ones. The intention is to be used in risk analysis (jump to default, etc).

◆ cumulatedLoss()

Real cumulatedLoss ( ) const

Actual basket losses between the reference date and the calculation date, taking the actual recovery rates of loss events into account. If the event has not settled yet a model driven recovery is used.

Returns the realized losses in this portfolio since the portfolio default reference date. This method relies on an implementation of the loss given default since the events have not necessarily settled.

◆ remainingNotional()

Real remainingNotional ( ) const

Remaining full basket (untranched) notional after settled losses between the reference date and the given date. The full notional for defaulted names is subracted, recovery ignored.

◆ remainingNotionals()

const std::vector< Real > & remainingNotionals ( ) const

Vector of surviving notionals after settled losses between the reference date and the given date, recovery ignored.

◆ remainingNames()

const std::vector< std::string > & remainingNames ( ) const

Vector of surviving issuers after defaults between the reference basket date and the given (or evaluation) date.

◆ remainingDefaultKeys()

const std::vector< DefaultProbKey > & remainingDefaultKeys ( ) const

Default keys of non defaulted counterparties

◆ remainingProbabilities()

std::vector<Probability> remainingProbabilities ( const Date & d ) const

Vector of cumulative default probability to date d for all issuers still (at the evaluation date) alive in the basket.

◆ remainingAttachmentAmount()

Real remainingAttachmentAmount ( ) const

Attachment amount of the equivalent (after defaults) remaining basket The remaining attachment amount is RAA = max (0, attachmentAmount - cumulatedLoss())

The remaining attachment ratio is then RAR = RAA / remainingNotional()

◆ remainingDetachmentAmount()

Real remainingDetachmentAmount ( ) const

Detachment amount of the equivalent remaining basket. The remaining detachment amount is RDA = max (0, detachmentAmount - cumulatedLoss())

The remaining detachment ratio is then RDR = RDA / remainingNotional()

◆ remainingTrancheNotional()

Real remainingTrancheNotional ( const Date & endDate ) const

Expected basket tranched notional on the requested date according to the basket model. Model should have been assigned.

◆ probOverLoss()

Probability probOverLoss	(	const Date &	d,
		Real	lossFraction
	)		const

The lossFraction is the fraction of losses expressed in inception (no losses) tranche units (e.g. 'attach level'=0%, 'detach level'=100%)

◆ expectedShortfall()

Real expectedShortfall	(	const Date &	d,
		Probability	prob
	)		const

ESF

◆ lossDistribution()

std::map<Real, Probability> lossDistribution ( const Date & ) const

Full loss distribution

◆ probsBeingNthEvent()

std::vector<Probability> probsBeingNthEvent	(	Size	n,
		const Date &	d
	)		const

Probability vector that each of the remaining live names (at eval date) is the n-th default by date d.

The n parameter is the internal index to the name; it should be alive at the evaluation date.

------—TO DO: Implement with a string passed-------------------— ------—TO DO: Perform check the name is alive------------------—

◆ probAtLeastNEvents()

Probability probAtLeastNEvents	(	Size	n,
		const Date &	d
	)		const

Returns the probaility of having a given or larger number of defaults in the basket portfolio at a given time.

◆ recoveryRate()

Real recoveryRate	(	const Date &	d,
		Size	iName
	)		const

Expected recovery rate of the underlying position as a fraction of its exposure value at date d given it has defaulted on that date. NOTICE THE ARG IS THE CTPTY....SHOULDNT IT BE THE POSITION/INSTRUMENT?????<<<<<<<<<<<<<<<<<<<<<<<

◆ correlation()

QuantLib::Real correlation ( ) const

Return single correlation number from the underlying loss model if it is a one factor model. If the underlying loss model is null, this returns a Null<Real>().

Public Member Functions

Basket Loss Statistics