Index | Recent Threads | Who's Online | Search

New Thread
This topic has been viewed 147 times and has 7 replies
jmh
bizarre bug on screener for past date when 2 or more aggregate functions are used

Hi P123,

When using 2 or more aggregate functions in a screen and for a date in the past more than ~1 or 2 weeks ago -> display NA for some of the original factors.

This is not only a display bug as it seems to also create issues for my calculations in a more complex screen.

I can share the more complex screen later but if you can first have a look at this, it might fix all the problems in one go.

See attached screenshot with link to public screen.
https://www.portfolio123.com/app/screen/summary/247753?st=1&mt=1

You can play with it by
1) changing the date (it works with today's date but not when the date is more than 1 or 2 weeks old)
2) turning on/off 1 or 2 aggregate()

Thank you,

Jerome

Attachment Bug past date with 2 aggregates in screener.pdf (220703 bytes) (Download count: 8)


Oct 13, 2020 11:32:53 AM       
Edit 2 times, last edit by jmh at Oct 13, 2020 1:09:44 PM
marco
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

That's the data limitations with historical screens imposed by data provider. Columns become NA after a certain number

Portfolio123 Staff.

Oct 13, 2020 3:06:07 PM       
jmh
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

Hi Marco,

No - this is not it in this case. There are only 40 stocks displayed on the screen courtesy of the last row "RBICS(leisure)"

-> And it would be the same problem whether I have 2 rows of aggregate calculations or one
-> and it would not work either when I use today's date (shift the date to today and everything works fine)

can you please have another look at it?

J

Oct 13, 2020 3:08:28 PM       
Edit 1 times, last edit by jmh at Oct 13, 2020 3:10:28 PM
yuvaltaylor
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

I'm not able to reproduce this. I ran it a number of times on different dates and didn't get any N/As.

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

Oct 13, 2020 4:27:41 PM       
jmh
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

Uh? I just reran the screen per the link above and still have the same problem.
Right hand column, screenshot attached -> N/A only for BetaInd.

Did you
1) Make sure that both Aggregate rows were active?
2) Choose the date 10/01/2020 (01 Oct 2020) for example

Thank you,

Jerome

Attachment Screenshot 2020-10-13 224225.jpg (599001 bytes) (Download count: 36)


Oct 13, 2020 4:41:58 PM       
danparquette
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

Hi Jerome. We were able to replicate the issue and we are looking into it.

Dan

Oct 13, 2020 6:54:14 PM       
marco
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

Jerome, it's a different restriction. Historically there's a limit with rows & # of columns. After a certain number of columns all values are NA. Try re-ordering the rules and you will see a different column become NA.

Yuval cannot reproduce bc he has an admin account

Portfolio123 Staff.

Oct 13, 2020 8:58:54 PM       
jmh
Re: bizarre bug on screener for past date when 2 or more aggregate functions are used

Thank you Marco.

1) To be safe, can you confirm that this is just a display thing i.e. the values that I store in @variable are correctly retrieved, stored and used?

In other words, when I use Showvar(@x, some function using P123 data)...
-> the column of P123 data might display N/A
-> but the variable @x would have stored and used the correct value.

For example, see this public screen where I was playing around with calculating the beta5Y of stocks based on {industry ; yield yes/no ; Debt to Equity ratio} -> https://www.portfolio123.com/app/screen/summary/247744?st=0&mt=1

2) I had a lot of N/A on my calculated Beta "@MyStockBeta" in this screen which is what attracted my attention.

However, it looks like the N/A are due to TxRate%() which seems to be missing quite a few historical data points.
Thoughts?

Thank you,

Jerome

Oct 14, 2020 12:27:55 PM