hi there, i’ve been trying out the LoopStreak function and have noticed a few places where the results aren’t quite as expected.
here is a simple example -
LoopStreak("Sales(CTR,QTR)", 20, 0, 1, #Positive, 1)
for ALTS:USA
why does the streak of Positives break in the circled area even though the SalesQ number is positive?
my guess is - it has something to do with a later than usual 10-k/q filing - but even if so, i’m not sure if breaking the streak is the right thing to do here.
thanks.
Which NA treatment did you choose-just the default? Have you tried adding a break setting or simply adding an OR salesq=NA to the formula if you don’t mind NA?. Could help test if that is the reason. Curious to hear more about this since I use a couple streak functions.
I think it's related to the long delay in filing Q4. There's a huge gap from Q3 to Q4. Also in Q4 but SalesQ remains the same. Something fishy. We'll investigate. Thanks
Looks like we'll need to tweak our logic. About 7.5 months passed before q4 was filed. That is currently disconnected to filing data and starting over. But since it is q4 I think we can be a bit more forgiving.
You can see what "disconnecting" means buy running this screen on 7/7/2018
Ticker("alts")
@0:sales(0,qtr) or @1:sales(1,qtr) or @2:sales(2,qtr) or @3:sales(3,qtr)
You will get this. In other words the data start from scratch bc previous filing exceeded the time we wait.
If I have a “salesq>0 or salesq=na” in my formula would that sidestep the issue or not really?
This is the interim data we have. Q4 for 2017-12-31 is missing because company did not file 10K on time. This will break the streak, and break continuity of the data which I think is a good thing. The 10-K was filed around June 12. This should have also filled in Q4 but it did not. So right now it looks like FactSet messed up a bit. Either way, nothing good going on with the company.
| 10-Q Period | Q# | Src date |
|---|---|---|
| 2018-06-30 | 2 | 2018-08-14 |
| 2018-03-31 | 1 | 2018-07-02 |
| 2017-12-31 | ||
| 2017-09-30 | 3 | 2017-11-14 |
| 2017-06-30 | 2 | 2017-08-21 |
| 2017-03-31 | 1 | 2017-05-16 |
| 10-K Period | Src Date |
|---|---|
| 2018-12-31 | 2019-03-29 |
| 2017-12-31 | 2018-06-12 |
| 2016-12-31 | 2017-03-31 |
SEC Filings in EDGAR
Never mind! They just reported annual figures in their 2017-12-31 10K. Q4 of 2017 basically does not exist. I suppose you could derive it from the 10-K and Q1,2,3 but FactSet did not do that, and we certainly do not.
At this point what we did, starting over for the interims, is probably the only right answer. We can't ignore the hole.
https://www.sec.gov/Archives/edgar/data/862861/000168316818001663/arca_10k-123017.htm
thanks for the detailed explanation @marco . it’s helpful to know the impact of delayed and/or combined 10k/q filings on this (and possibly other) function(s).
i tried this - LoopStreak("IsNA(Sales(CTR,QTR),1)", 20, 0, 1, #Positive, 1)
it does seem to work well for this example (though i haven’t looked at it more broadly).
thank you.
Thanks for sharing
