PIT

Manage
21

Hi David,

I’ve been looking at PeriodDateQ and it seems to represent the fiscal quarter end date and not the date the results were reported. For Apple, the latest QTR ended 03/31/18 and results for that QTR were reported on 05/01/18. P123 reports the EPS data for that period now - 05/07/2018. And WeeksIntoQ is currently zero for Apple.

Do you agree?

Walter

EDIT: Also looked at IBM. Its PeriodDateQ is also 03/31/18 and it reported on 04/17/18. Its current WeeksIntoQ is two. So WeeksIntoQ has some relationship to the report date. I just can’t tell if it incorporates CapitalIQ lag, too.

22

Jim,

According to the document, you are onto something. I just got the impression that the distinction between the two has increasingly diminished since the 2004/2005 acquisition.

//

Walter,

Yes. You are right of course. My apologies for the confusion.

Since “AnnounceDaysQ” is defined as (announce date - period end), can we use that to estimate the reporting lag?

Anyhow, I’ll go fix my (incorrect) post as soon as we can collectively come up with a working solution.

//dpa

23

I think the PIT doc is from 2008. It’s probably most useful for concepts.
Walter

24

How about this for lag;

ShowVar(@DaysLag,DaysSince(PeriodDateQ)-AnnounceDaysQ-WeeksIntoQ*7)

Walter

25

That’s a good thought.

However, when I try it out, I only get values which 8 or less for the entire fundamentals universe, which indicates to me that it is based on the announcement date as Paul suggested, not the update date.

I would very much like clarity on this issue.

I will revisit the topic in a few days after finals.

//dpa

26

I think I agree with that.

Best of luck on your finals! Or is it ‘break a leg’?

Walter

27

Before Marco went to SnapShots it was exactly as the document describes with Capital IQ. You could pick a stock in a port and when you later ran the sim a stock or two that was in the port would not be in the sim. Instead, the sim—looking in retrospect–had better information than you did when you rebalanced the port and the sim picked a better stock. This should not happen with Compustat data–except as an error–according to the document.

The topic comes up because Marco has stopped using SnapShots or will stop doing it shortly.

Marco went to SnapShot for a reason, I think. This is the reason he did it, I believe. Not just because of a concept as there is extra work in SnapShots.

But I am not saying that SnapShots solved everything (e.g., how far back to SnapShots go anyway). And I am not being critical: CaptialIQ is a better data source than Zacks for example. Furthermore, Marco has been doing SnapShots for at least a couple of years. Maybe CapitalIQ is doing better.

I wish that once I picked a stock with P123 I had some way to confirm that there was not a more recent analyst change that was not in the data base yet. There is a Capital IQ service that updates you during the day in close to real time (they claim). I would pay extra for that service if during the day it would stop me from continuing to buy a stock that had a downward revision that was not in the database when I rebalanced in the morning.

I WOULD PAY MORE FOR THIS SERVICE NOW WITHOUT KNOWING HOW EFFECTIVE OR USEFUL IT IS AHEAD OF TIME.

I would be willing to pay more than twice what I pay now. Again, I am not sure SnapShots is a perfect solution anyway and I am happy with whatever P123 does on this—especially if the community is aware of this and keeps an eye on the effect. But as far a positive solutions: I would definitely be willing to check out the real-time service.

If we had realtime prices as well as realtime data from CapitalIQ enabling realtime ranking (at least for a watchlist) that would be something. A watchlist might just be 10 stocks from a morning rebalance with relative ranking changing as realtime prices and realtime Capital IQ data come in. Maybe for that port you would buy the 5 highest ranked stocks—even if those 5 stocks changed during the day. With IB you would just stop the algo (VWAP best-effort or whatever) from further buying during the day and move to a different stock. You would change stocks on any news or essentially when your stock hit a price target (more accurately, when a stock with a lower price and more value moved up in relative rank).

And in fact you would no longer have to worry about slippage. If you are having a market impact and that impact causes the stock to hit the target price in realtime (i.e., the stock no longer has as much value) you just move to a different (higher ranked now) stock.

Now I would pay for that!!! The potential benefit of not continuing to buy even a single stock that is no longer any good (during the year) could far outweigh the risk of paying a little extra money to P123 with no benefit. There can be no doubt that one’s risk would be reduced by keeping money in cash when the upside potential is reduced for a stock or diversifying if one chooses to move to a different stock.

BTW, no one is going to try to claim that there are no traders already doing this profitably, I hope. No traders with price targets and changing their buying as news comes out? Really? I think P123 could do this exceedingly well and possibly at a reasonable price. Heck, if I had enough time and energy I would follow the rank changes during the day based on the just the price changes during the day on a spreadsheet. In fact, I am likely to give this a trial run at some time in the future. But the point is: how much would that cost?

I’m just thinking of the P123 subscriber entering the new fundamental or earnings estimate data during the day himself (possibly with P123 realtime price data) for the watchlist as it comes in during the day: keeping those stocks with no manual entries static. And just a watch list that needs no more computing power than a laptop.

-Jim

28

Just an update to close the loop and/or keep alive the faith.

The question is how to tell how long the data has typically been public before it’s put into Compustat and made available for us as data consumers. For this, we must know only two things:

  1. How long since the earning/press release ‘n’ periods ago.
  2. How long since the database was updated with the information.

I can infer the # number of days from the last report released by said company as:

@DaysSinceLastAnnouncement = DaysSince(LatestPeriodDate) - AnnounceDaysQ

I cannot think of an elegant way to recover how long its been since the data has been updated. Chaining together a bunch of FHist functions seems overly brutish and error-prone.

However, it is very facile to screen for companies that are experiencing reporting lags through the use of “StaleStmt” and other conditional statements. This tells us that there are lags, but provides limited information about the typical lag. Moreover, when the data is input into the backend system, it makes everything seem copacetic – as though there never was a lag. I.e., we can infer nothing about how stale the data is by the time we get the info.

I am very interested in being able to retrospectively determine the lag period between the data becoming public to when it was ingested into either P123 and/or Compustat. I also believe this information would be highly valuable to other investors who use this service. Since I realize there may be not be a “clean
'n easy” way to timestamp each update because – on one level – P123’s database is updated each day by CompuStat, I would like to clarify it should be sufficient to simply stamp the most recent change to any fundamental datapoint as of each day.

Thoughts very welcome…

//dpa

29

Just a quick update on my thinking about Capital IQ. I no longer care about whether we use SnapShots (unless Marco has a preference either way). I am not particularly worried about whether the sims and ports match. Let me start with an analogy.

If you want everyone in a class to perform the same do you handicap the brilliant students or do you improve the performance of those who are struggling? Well it is pretty clear that it depends. But I vote for helping those who are struggling.

So, with P123 do I want to handicap the sim that may have additional public information not available in the database at the time of rebalance for the port? Or would it be better to see how the sim does in hindsight and start to look at the information sources that provide information in the morning and during the day that does not get downloaded into P123 the previous night. Remember, Capital IQ pretty much guarantees the information existed and it is PIT in that sense. It does show what you might have done if you had that information available.

I would prefer to know what is possible. What to aspire to. Which means learning where to look for additional data in the morning and throughout the day.

I will probably never be able to combine all that information during the day to get new ranks. But I probably can learn about some stocks that I want to stop buying: if for example, there is a new earnings estimate release. This would be an earnings estimate that will show up tomorrow in the P123. That new information will kick the stock out of the first 5 (10 or 15) stocks tomorrow but I continue to buy the stock today in blissful ignorance (using VWAP best-efforts for example). I might even be able to get a crude assessment of how the rank has changed with price changes during the day: maybe it is no longer as much of a value. Then there are morning earnings releases etc that could be incorporated into the question: do keep buying this stock after lunch?

Capital IQ provides a real-time service just for that purpose. Marco has discussed it as something he was considering purchasing for P123. But it may not be feasible to make this option available thru P123…

In any case, I trust Marco on this and I may not really care either way about SnapShots. But with my real money I want to improve my ports and do not need to handicap my sims: if there there is, in fact, a discrepancy and if it turns out to be the case that there are some improvements to be made with additional research in the morning and during the day.

Also further discussion of this may lead to the false impression that I want to change the data sources in some way. Actually, I want no change in my morning rebalances at all.

-Jim