Skip to content
- Use mechanism of “convolution”
- Use Bayesian theorem for the weight of default for each grade, i.e. the probability that a default occurred in a specific grade
- Use the binomial distribution to predict the probability that the number of defaults differs from the number of defaults in the grade
- Use then Poisson distribution to take into account the number of customers in each grade to generate a frequency distribution
- The use a convolution mode to combine the information and produce de matrix (how it is done is not clear)
- PD depend on number of customers in grades and changes over time. As such it is dynamic. How it is applied in practice is not clear

# Latest notes archived