I’m reporting a regression regarding the feature import functionality via clipboard and CSV. Since one of the recent updates, I’ve been unable to import features correctly.
The Issue: When I copy features from one AI Factor to another via the clipboard, I receive a normalization error “Line 1: Step 1 Normalization is not valid“, even though the normalization settings are identical to the source AI Factor. The same error occurs when attempting to import via a CSV file.
Why this matters for my workflow: While the "Import from AI Factor" option still works, it doesn't allow for the bulk editing that the clipboard/CSV method provides. My typical workflow involves pasting factors into Excel to quickly change normalization types across multiple rows (for example, switching from "Skip Date" to "Skip Date & Sector" for a large batch) without having to manually edit each line in the UI.
The Regression: Previously, this was a seamless process. Now, the import fails even if I haven't modified the data at all. This has significantly impacted the efficiency of managing and iterating on large feature sets.
Could you please look into this? I suspect other users who rely on Excel for bulk feature management may be experiencing the same issue.
There was an additional 'expanded formula' column added to the export. This column is useful for other purposes like importing into FactorMiner, but it will cause the import back into AIFactor to fail. If you delete that column from the Excel file, the import will work as it did before. Sorry for the confusion. We are working on the flow between to make things go smoother.
FYI - the 'expanded formula' column shows the formula with the normalizations applied.
For example, if:
formula = ROA%(CTR,TTM)
step 1 = 2Y Relative Hist (N=26)
step 2 = Date
then the expanded formula is FHistZScore("ROA%TTM", 26, 4, 2.5, NA)
Thanks, but it still doesn't work. I deleted the "expanded formula" column, and now it either complains about an invalid "Step 1 normalization" or about consecutive ",," commas in the "tag" field.
And if I replace all ",," with "tag" or something else in the tag column, then it starts saying that the command is now invalid for that field.
It's quite inconvenient that this no longer works correctly and requires so many manual changes. Maybe I'm doing something wrong?
Are you able to successfully re-import it by only removing the "expanded formula" column?
I can import from one AIFactor to another using the clipboard method as long as I have deleted the expanded formula column. It even handles the case with multiple commas in the tag field.
Your first comment mentioned "Skip Date". Just to be clear - those are 2 fields. I attached a table that shows many different normalization combinations that worked fine. Could you please share one row that is failing in your import?
My apologies — I tried doing it again today exactly as you described, but this time I used Excel instead of Google Sheets like before, and everything worked correctly after removing the "extended formula" column.
While I'm here, I'd also like to express my gratitude and respect. Your somewhat legendary factor list helped me quite a lot during one stage of development!
Thanks! The data from 'single-factor' testing was what enabled me to quickly find success with P123 since I had no finance background at the time. I am always glad to hear that it helped others. Now we have the FactorMiner tool which allows you to quickly test a big batch of individual factors.