Loading...
Home / Collateral Spread for COLVA

Home Forums Features Collateral Spread for COLVA

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #6184
    rkapl
    Participant

    Dear All!

    I’ve seen that the spread used to calculate COLVA is different from the spread in the netting Agreement:
    Parameter “CollateralCompoundingSpreadReceive/CollateralCompoundingSpreadPay” from netting.xml: used in OREAnalytics/collateralaccount.cpp

    // apply “effective” accrual rate (i.e. adjust for spread specified in netting set definition)
    Real accrualRate = (accountBalances_.back() >= 0.0) ? (annualisedZeroRate – csaDef_->collatSpreadRcv())
    : (annualisedZeroRate – csaDef_->collatSpreadPay());
    // bring collateral account up to margin payment date (note accrual rate is assumed to be compounded
    // daily)
    accountBalances_.push_back(accountBalances_.back() * std::pow(1.0 + accrualRate / 365.0, accrualDays) +
    marginCalls_[i].marginAmount());

    Parameter “collateralSpread” from ore.xml: used in OREAnalytics/postprocess.cpp

    Real colvaDelta = -balance * collateralSpread_ * dcf / samples;
    Real floorDelta = -balance * std::max(-indexValue, 0.0) * dcf / samples;
    colvaInc[j + 1] += colvaDelta;

    It would be great to have the same Parameter(s) CollateralCompoundingSpreadReceive/CollateralCompoundingSpreadPay being used for COLVA calculation as well, now we have to set the collateralSpread for each counterparty separately.

    -regards,
    Roland

    #6185
    rkapl
    Participant

    Something like the following could be added right before the usage of the collateralSpread_:

    collateralSpread_ = (balance >= 0.0 ? netting->collatSpreadRcv : netting->collatSpreadPay);
    

    Real colvaDelta = -balance * collateralSpread_ * dcf / samples;

    -regards,
    Roland

    #6227
    rkapl
    Participant

    Thanks a lot for integrating this suggestion in the new release!

    Regards,
    Roland

Viewing 3 posts - 1 through 3 (of 3 total)
  • You must be logged in to reply to this topic.