I am not sure I like the "hybrid" model, although the results are sort of OK. I don't like the "lumpiness" of the ISL. If you are at one end of a lump you get a windfall, the opposite end you are harmed.

I think I prefer filtering first with status and category, and then within each, filter again on a relative basis, perhaps weighted for longevity of the whole group in the status and category.

It is what it is.