Documentation request - GBM and TOWT model hyper-parameter tuning.

[chrisbalbach]

When conducting (GBM or TOWT) model hyper-parameter analysis against more than one baseline model (i.e. batch or portfolio analysis), are the resulting hyper-parameters static across multiple models or are they uniquely determined for each model?

The existing textual model description is not clear to me on this issue.

[samirtouzani]

For the GBM model the optimal hyper-parameters are estimated separately for each model. Regarding the TOWT there is no cross validation so the same hyper-parameter (timescale for weighting function) will be used for all the models of the batch.