Dear David,
After I watched your video on "CreditMetrics: first building block – 10 min briefcast" (http://www.bionicturtle.com/learn/article/creditmetrics_mapping_transition_probabilities_to_random_normal_variables_1/), I have understood the essence of the creditmetrics approach, however, I can't help wondering if we will be tested on the actual calculation for CreditMetrics just like you illustrated in the video. Or for CreditMetrics, FRM candidates will only be tested on the concepts level, including advantages and drawbacks? The few drawbacks I can think of are that:1) reliance on external credit rating 2)historic backward-looking 3) assuming static zero term structure; whereas advantages are: 1)incorporate spread change and thus market value change.
If indeed we need to be able to do the calculation, can you kindly answer my following question concerning the video?
1) Why the probability of state in the final stage of calculation needs to be transposed, instead of simply extracting the probability of migration from the rating migration table?
2) I think that variance is usually calculated by considering the difference between observed values and the arithmetic average/expected value which is equally weighted. ie. mean of a sample of 1 to 10 is 5.5. But in your example, the average is probability weighted. I understand that the mean should be probability weighted but just would like to know if this is an exception for calculating the variance that is to be applied in this type of scenario.
Thanks!
Cheers
Liming
3/10/09
After I watched your video on "CreditMetrics: first building block – 10 min briefcast" (http://www.bionicturtle.com/learn/article/creditmetrics_mapping_transition_probabilities_to_random_normal_variables_1/), I have understood the essence of the creditmetrics approach, however, I can't help wondering if we will be tested on the actual calculation for CreditMetrics just like you illustrated in the video. Or for CreditMetrics, FRM candidates will only be tested on the concepts level, including advantages and drawbacks? The few drawbacks I can think of are that:1) reliance on external credit rating 2)historic backward-looking 3) assuming static zero term structure; whereas advantages are: 1)incorporate spread change and thus market value change.
If indeed we need to be able to do the calculation, can you kindly answer my following question concerning the video?
1) Why the probability of state in the final stage of calculation needs to be transposed, instead of simply extracting the probability of migration from the rating migration table?
2) I think that variance is usually calculated by considering the difference between observed values and the arithmetic average/expected value which is equally weighted. ie. mean of a sample of 1 to 10 is 5.5. But in your example, the average is probability weighted. I understand that the mean should be probability weighted but just would like to know if this is an exception for calculating the variance that is to be applied in this type of scenario.
Thanks!
Cheers
Liming
3/10/09
