Not really a bug, but an unintuitive interaction:
It's possible to use context to define a column that is an attribute of the pick field, which can be tagged as a group_by or pick_sum. If used as a group_by, then any name_sum columns provided from the draft will group by that column (but not "pick") and then pivot to be tagged with the name associated to the column. This results in a Cartesian product of card names with pick attributes.
The paradigm that currently works is to identify such columns as "pick_sum" columns and divide by num_taken in the agg step. This could be improved.
Not really a bug, but an unintuitive interaction:
It's possible to use context to define a column that is an attribute of the pick field, which can be tagged as a group_by or pick_sum. If used as a group_by, then any name_sum columns provided from the draft will group by that column (but not "pick") and then pivot to be tagged with the name associated to the column. This results in a Cartesian product of card names with pick attributes.
The paradigm that currently works is to identify such columns as "pick_sum" columns and divide by num_taken in the agg step. This could be improved.