hi Yuval,
After switching to FactSet data, those industry based factor should be based on Factset industry. I tested this change of industry based factors will produce big difference between legacy GICS based simulation and RBICS based for example(FactSet’s result is is much worse) i.e. Pr4W%ChgInd, is it because still some mapping work not completed or some issue here? Thank you.
Hi P123 / Yuval,
It looks like SecCount in buy / sell rules might still be counting GICS and not Factset sectors (e.g. in a buy rule setting a limit using GICS codes rather than RBICS).
Or did I get this wrong?
Thank you
Jerome
Everything seems to be working as it should. I don’t know what the matter is with your simulation. I made a screen with one rule: FRank(“Pr52w%chgInd”) > 95. I ran a rolling backtest with a six-month holding period and a one-week frequency over the last ten years. FactSet significantly outperformed Compustat with this rule. It underperformed Compustat by a little using Pr4W%ChgInd. But different performance should be expected with differently composed industries.
I hate to say it, but SecCount works as it should. Run a simulation using All Fundamentals as your universe and 10 to 20 positions. Use two buy rules: GICS(50) and SecCount < 3. On FactSet you’ll get 8 stocks from 4 different sectors; on Compustat only 2.
Unfortunately the Sector column in the Holdings tab still gives GICS. We’ll fix that.
Hi Yuval,
the sector pie chart in the holdings section of a simulation is indeed where i picked up an issue. i had a buy rule seccount limiting a sector to 33% under RBICS but the pie chart was showing 50% in tech sector which i understand now is the gics tech sector
how do we know when this is implemented? trello?
thank you
jerome
You’ll know right away because you’ll start seeing sectors with names like “Business Services,” “Consumer Services,” “Consumer Cyclicals,” “Consumer Non-Cyclicals,” and “Telecommunications” (rather than “Communication Services”).