Hello All,
I get different results in my ranking system depending if I neutralize the subsector bias using the rank against using Frank which is not the intended behaviour.
For example:
Ranking using GMgn%TTM against SubSector yields
Ranking with Frank("GMgn%TTM",#Subsector,#desc,#inclna) against Universe
The results should be the same shouldn’t they?
Okay I think I understood,
By using FRank inside a ranking node, you prematurely convert an NA (a flag for missing data) into a hard numeric percentile (e.g., 40) before the main ranking engine processes it.
The classical ranking engine then sees the number "40" instead of recognizing missing data, which completely bypasses your global "Percentile NAs Negative" system setting. The engine treats your missing data as a legitimate tie between healthy companies, altering the final tie-breaking math and slightly shifting your bottom quantile boundaries.
@tr Using #ExclNA instead of #InclNA should produce similar results. It's best to use ranking system nodes if at all possible.
