-
|
Hey guys! I wanted to ask we can use a different distribution for the Bootstrap function, that being the gamma dist. instead of the ODP. Im trying to stochastically model Paid triangles and I think Gamma will be a better choice! Please let me know your thoughts and recommendations, thank you as always |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
This is supported by Theory (refer to 3.4. The GLM Bootstrap Model in Monograph 4. However, the implementation here is limited to ODP because the algorithm exploits the relationship between the ODP and the basic chainladder to get at a simple/fast implementation. A more comprehensive framework would be needed to generalize beyond ODP. I think any implementation should be treated as its own separate estimator. |
Beta Was this translation helpful? Give feedback.
-
|
I see! So a separate estimator must be constructed for the Gamma Bootstrap then right? Im going to try my luck with it to see if its doable |
Beta Was this translation helpful? Give feedback.
This is supported by Theory (refer to 3.4. The GLM Bootstrap Model in Monograph 4. However, the implementation here is limited to ODP because the algorithm exploits the relationship between the ODP and the basic chainladder to get at a simple/fast implementation. A more comprehensive framework would be needed to generalize beyond ODP. I think any implementation should be treated as its own separate estimator.