Classical and quantum computing methods for pricing for account-level uncertainty
AssigneeDeep Future Analytics, LLC
InventorJoseph L. Breeden
Life of this patent
Estimating the term…
Abstract
A Monte Carlo simulation is disclosed to propagate model estimate and macroeconomic uncertainties through the calculation of lifetime loss or yield and their uncertainties using a classical computer or a quantum for the purposes of adjusting loan pricing for the uncertainty in the yield.
Claims8
A method for estimating the uncertainty in a prediction of lifetime yield for an individual loan for given loan pricing parameters and credit risk data, the method comprising the steps of:
- feeding input data into a probability of default model PD (t) and a probability of attrition model PA (t) to create PD (t) and PA (t) for each time step to the end of the loan;
- prior to sampling, running test cases to determine the approximate shapes of the distribution of lifetime probability of default LPD and the distribution of yield to reduce the number of Monte Carlo simulations and thus reduce the processing time, next sampling the values of the distributions of PD (t) and PA (t) by way of Monte Carlo simulations to obtain distributions of the probabilities of unconditional default UPD (t) and unconditional attrition UPA (t) that fall within said approximate shapes in real time;
- and
- accumulating the sampling for the life of the loan to create the distribution of lifetime probability of default LPD and the distribution of yield based on the accumulated samples in real time.
A method for estimating the uncertainty in a prediction of lifetime yield for an individual loan for given loan pricing parameters and credit risk data, the method comprising the steps of:
- feeding input data into a probability of default model PD (t) and a probability of attrition model PA (t) to create PD (t) and PA (t) for each time step to the end of the loan;
- prior to sampling, running test cases to determine approximate distribution shapes of the distributions of LPD and the distribution of yield to reduce the number of Monte Carlo simulations and thus reduce the processing time, then, sampling the values of the distributions of PD (t) and PA (t) by way of a quantum Monte Carlo simulation algorithm that fall within the test distribution shapes to obtain distributions of the probabilities of unconditional default UPD (t) and unconditional attrition UPA (t);
- accumulating the sampling for the life of the loan to create the distribution of lifetime probability of default LPD and the distribution of yield based on the accumulated samples.
A method for estimating the uncertainty in a prediction of lifetime yield for an individual loan for given loan pricing parameters and credit risk data, the method comprising the steps of:
- feeding input data into a probability of default model PD (t) and a probability of attrition model PA (t) to create PD (t) and PA (t) for each time step to the end of the loan;
- prior to sampling, approximating the shapes of the distributions of LPD and yield, then sampling the values of the distributions of PD (t) and PA (t) by way of Monte Carlo simulations that fall within the approximated distribution shapes in order to reduce the number of Monte Carlo simulations and thus reduce the processing time for a given loan in real time;
- and,
- accumulating the sampling to create the distribution of lifetime probability of default LPD and the distribution of yield that fall within the approximated distribution shapes.
A method for estimating the uncertainty in a prediction of lifetime yield for an individual loan for given loan pricing parameters and credit risk data, the method comprising the steps of:
- feeding input data into a probability of default model PD (t) and a probability of attrition model PA (t) to create PD (t) and PA (t) for each time step to the end of the loan;
- prior to sampling, approximating the distribution shapes of the distributions of LPD and yield in order to reduce the number of Monte Carlo simulations and thus reduce the processing time for a given loan to enable real time results;
- sampling the values of the distributions of PD (t) and PA(t) by way of quantum Monte Carlo simulations to obtain distributions of the probabilities of conditional default UPD (t) and conditional attrition UPA (t) that fall within the distribution shapes;
- and
- accumulating the sampling to create the distribution of lifetime probability of default LPD and the distribution of yield that fall within the approximated distribution shapes.
Dependencies are inferred from the claim text.