I would like to make sure I understand the delay between an earnings release and the stock’s associated change in rank. For example, CJES announced unexpectedly good quarterly earnings after market close on Feb 11, but when I run my screening/ranking with various “As of” dates I don’t see CJES in the results until February 14, by which time its price has increased by 13%. Is there anything I can do to reduce this delay, or is it inherent?
Eric, that is probably because at the moment the ranking is calculated only once a week, so the data you see in the sims and screens are only as of the Saturday.
Eric,
Iavanti is correct about there being a difference between sims and ports regarding when the ranks are updated. But it looks like the database did not have the earnings for the most recent quarter for CJES until the fourteenth. Using ShowVar in the screen EPSExclXorQ went from .42 on the 13th to .40 on the 14th for CJES.
Surprise%Q1 was not updated until the 14th either.
I see that the 8-K was not filed until the next day (time of day unknown). If the database is updated from the SEC filing rather than from the press release that would account for one day of delay. Still, one would hope to see the new rank on the morning of the 13th. I’m just running a screen & ranking system, by the way.
The delay can be up to one month, depending on when Compustat gets around to processing it. I believe that LargeCaps are processed fast, smallCaps slow.
Steve
Thanks Steve,
So that brings up an important question. Does CompuStat go back and revise the data? It still could be PIT, technically, if PIT just means not before the data was made public.
My impression is that they probably do not revise fundamentals but probably revise estimates. This is based on my comparison of the holdings for sims versus ports.
I think the way different data is handled (e.g., large-cap vs small-cap, fundamentals vs estimates, data on short interest, etc) explains much of why some sims fail out-of-sample. That and market timing.
My holdings are mostly small caps, so I guess I’ve been very lucky never to have seen a delay that long. Anyway it’s good to have a better understanding of the issue.
Compustat flags every financial statement item with an “effective date”, or point-in-time date. The effective date is when Compustat added it to their database. It’s independent of when the company released that data and is not changed after the fact ( a new item can supersede it with a new effective date). This data can come from press releases (preliminary data) or SEC filings. We do not show any data whose effective date is after your point in time date thereby preventing look ahead biases.
Yes, smaller stocks take longer than larger stocks to process during busy periods. Compustat prioritizes processing new data by market cap.
Does that mean Compustat is “late” vs the market ? That depends…
Is it late compared to someone downloading the raw SEC filing and coming up with their own conclusions? YES
Is it late compared to high frequency traders monitoring in real time press releases, maybe with some artificial intelligence parsing the actual english sentences? YES
But the main two data providers (Compustat & Reuters) do it the same way . The filings are hand-reviewed by analysts to standardize the data which is why it takes longer for smaller companies (big funds , the main data clients, don’t care about those).
Standardization of the data is a very important step for our type of analysis, and most big managers. So Compustat is not late in providing hand-cleaned, standardized data compared to most of the market.
For more info on Compsustat data process please see Marc Gerstein document “Why Compustat?” here: https://www.portfolio123.com/doc/doc_data.jsp
Thanks Marco! Works for me: not just the best data available but they are doing it right. Thanks for the clarification.