If you run screen as of today, for GS, JPM and MS - you will get LatestFilingDate as 2019-05-06, 1970-01-01 and 1970-01-01 respectively
It looks incorrect to me - I have emailed P123
Thank you, Chipper and Quanto,
We may need more information about exactly what filings qualify, too, because corporations file with the SEC constantly.
Looking at the Edgar database, IBM did make a filing on 2/26/19 but, for example, also on…
2/13/19
2/5/19
2/4/19
2/1/19
1/31/19
1/30/19
1/28/19
1/25/19
1/24/19
1/23/19
1/22/19, etc.
I’m pretty sure the filing on 1/22/19 contains the kind of quarterly information which, generally, fuels Compustat and p123 and upon which we rely.
I think this absolutely is a PIT issue. Assuming the key release in terms of content really was January 22nd, 2019 that should be the LatestReleaseDate as of February 11th, 2019, and should remain so until the next release of meaningful quarterly information, which I believe took place on April 16, 2019.
We should never be running simulations based on information unavailable on the day in question. How could it be otherwise?
Hugh
Hugh,
I have probably read tens of thousands of SEC filings. I could get into a discussion of wonky details which may or may not be relevant to what you need to do. So, let’s take a step back. What are you using LatestReleaseDate for? Are you sure that you aren’t really looking for PeriodDateQ?
PeriodDateQ is mislabeled. If it was up to me, I’d call it “PeriodDateLatest” because it includes annual data too.
I was confused somewhat, so I created a screen report to list the various dates associated with tickers. Below shows the results for the three tickers mentioned. I’m still somewhat confused. On the face of it, I would assume Goldman Sachs has had no press releases since 1970, and JP Morgan Chase and Morgan Stanley haven’t filed anything with the SEC since 1970. ???
Try - using As-of-Date of 3/31/2019 and look at the LatestFilingDate
you will find all three (GS, MS, JPM) of them having 2019-02-26
And I think sometime next week the LatestFilingDate would change to a date in July
Hi Chipper,
I was hoping to use LatestFilingDate as an adjunct to WeeksIntoQ.
Hugh
“1970-01-01” means 0 seconds since the epoch. In other words, it may be a placeholder for N/A.
[quote]
I was hoping to use LatestFilingDate as an adjunct to WeeksIntoQ.
Hugh
[/quote]I see.
Have you looked into using PeriodDateQ and/or LatestActualPeriodDate? I wonder if they would be more helpful for your needs.
To obtain the number of calendar days since the most recent filing, I use the following formula:
IsNA (LatestActualDays, IsNA (DaysSince (Max (LatestFilingDate - 1, LatestNewsDate)), 7 * WeeksIntoQ))
It’s far from foolproof, but I find it works pretty well. Better than anything else I’ve tried . . .
LatestFilingDate is returning the SEC filing date of the most recent period exposed by the observation date. It works ok if the latest period is final (CompleteStmt=TRUE) but very confusing (and not PIT) if the latest period is preliminary (CompleteStmt=FALSE). In essence:
- In the past if will return a filing date in the future if latest period is preliminary
- In the present it returns 19700101 (the N/A value for dates) since nothing has been filed with SEC for latest period
The quick fix is to return NA in the past if the period is preliminary. Which is not very useful. Since I don’t think anyone is really depending on LatestFilingDate , let us know how you would like it to behave. Should it just return the filing date of either a) the current period if final or b) the previous period if current is prelim ? This is the only other logical behavior I think of something called ‘LatestFilingDate’, however it could be make the interpretation of ‘Latest’ confusing.
Thanks
either a) the current period if final or b) the previous period if current is prelim
why not the Latest date regardless whether it is final or prelim
Should it not be a date corresponding to the value of latest EPS I am seeing for MS (which is for Q2)?
If it were me, I’d start by offering the raw data and definitions, given these are clearly distinguished as such. Copy-pasting the source documentation would alleviate a lot of the strain on my end. Then I’d provide well-documented supplementary data points, where applicable. And finally, I would provide tools to manipulate these.
Yuval,
Your formula…
IsNA (LatestActualDays, IsNA (DaysSince (Max (LatestFilingDate - 1, LatestNewsDate)), 7 * WeeksIntoQ))
…is very clever but I hesitate to use it while LatestFilingDate could be returning future dates.
I am with Primus. Especially given the confusion, why not make the raw data and definitions available? Let the p123 community work with those and suggest additional tools, if necessary, as well as supplementary variables that might be useful.
I’m sure we all agree there are significant challenges with this topic but maybe we can do better.
Hugh
Strange. What are these dates supposed to present? How could PeriodDateQ be later than the latest news and filings?
EDITED:
71 stocks pass:
[quote]
!CompleteStmt
PeriodDateQ > Max(LatestNewsDate, LatestPeriodDate, LatestFilingDate)
[/quote]None of them are NA.
Some of these filings may come from analyst actuals. However, even after including analyst actuals, the rules:
[quote]
!CompleteStmt
PeriodDateQ > Max(LatestNewsDate, LatestPeriodDate, LatestFilingDate, LatestActualPeriodDate)
[/quote]brings up 38 stocks.
How are WeeksToQ and WeeksIntoQ determined / reset?
Most likely once earnings are reported WeeksToQ/WeeksIntoQ are reset (P123 is that correct?)
GS and JPM reported earnings on July 16 - so WeeksIntoQ is 1 (using 7/25/19 as-of). Not sure why WeeksToQ for JPM is 11 and GS is 12 though.
MS reported earnings on July 18 - so maybe by tomorrow - WeeksIntoQ would turn 1.
I am sure data dates for all earnings (prelim/ final) are being captured in P123. Question is can they be exposed to users?
Thanks
What do I use if I want to find out “time since last quarter release”?
Today using screener As of 7/28/2019 - all the three stocks (MS, JPM, GS) value of WeeksIntoQ is 1 and value of WeeksToQ is 11.
However values of WeeksIntoQ and WeeksToQ have changed for As of 7/25/2019, when I ran the screen on 7/25/2019 and when I ran the screen today.
That snapshot doesn’t really prove anything.
7/25/19 was on Thursday. Data is saved weekly only. So, the Thursday’s data is supposed to be overwritten. To verify PIT, you need to snapshot the data after the final weekend update–but before the Monday night/Tuesday morning updates. Then, come back to it in the future using the appropriate “as of” date.
The screenshot below was taken now during this window after the weekend update but before the daily updates. The screen is here. Let’s run it again next week to verify the PIT integrity of the dates.
[color=red]There is a PIT issue with LastFilingDateQ. [/color]
Last Monday, ABT LastFilingDateQ was NA (1970). Now it shows up retroactively as 6/30/19.
EDIT: I highlighted PeriodDateQ for ABT instead of LastFilingDateQ on the attached photo. I cannot edit it. Sorry. The issue still stands.
I keep stumbling on data points that raise concerns.
For example, on July 15, 2019 ticker “PERI” shows “LatestFilingDate” of 5/7/2019 and “WeeksintoQ” of 16.
Already the WeeksintoQ number is odd. But the company’s investor relations page indicates that its 1st quarter earnings-related press release was issued on 5/15/2019:
https://www.perion.com/press-releases/
It’s hard to make sense of this.
Hugh