Index | Recent Threads | Who's Online | Search

New Thread
This topic has been viewed 102 times and has 5 replies
Historical *EPSStdDev

I would like to access historic CurQEPSStdDev. What's a reliable method for getting the most recent historic value for both CurQEPSStdDev and CurFYEPSStdDev?


Aug 13, 2019 12:13:12 PM       
Edit 1 times, last edit by wwasilev at Aug 13, 2019 12:23:35 PM
Re: Historical *EPSStdDev

I think that FHist should work:


The comma following the one is the number of weeks in the past that you can get the figure.

This is a bit of a last-ditch solution -- FHist reads historical values and isn't necessarily adjusted in the same way that just about everything else on the site is -- but it's the only one that will work.

You know that's the standard deviation of the estimates that form the mean, by the way, not estimated future standard deviation, right?

Aug 13, 2019 3:29:26 PM       
Re: Historical *EPSStdDev

Thanks Paul

I don't know if FHist will map reliably to the data set I want. My intention is to calculate Standardized Unexpected Earnings (SUE). If EPSStdDev(offset, type) existed, I could write;
(EPSActual(0, QTR)-EPSEst(0, QTR))/EPSStdDev(0,QTR)
(EPSActual(0, ANN)-EPSEst(0, ANN))/EPSStdDev(0,ANN)

I'll look at FHist and maybe WeeksIntoQ to get the index right.

Aug 13, 2019 3:47:10 PM       
Edit 2 times, last edit by wwasilev at Aug 13, 2019 5:02:52 PM
Re: Historical *EPSStdDev

I'll look at FHist and maybe WeeksIntoQ to get the index right.

This is probably what you mean (and may have looked at this) and I have not tried this but I would look at this for 4 weeks ago:

Eval(WeeksIntQ > FHist(WeeksIntQ,4),FHist(CurQEPSStdDev,4), FHist(NextQEPSStdDev, 4))

Presumably if WeeksIntQ > FHist(WeeksIntQ,4) is true there has not been an announcement in the last 4 weeks so FHist(CurEPSStdDev,4) should be referencing the current quarter.

If WeeksIntQ > FHist(WeeksIntQ,4) is false, then there has been an interim announcement and FHist(NextQEPSStdDev, 4) references the current quarter: what was the next quarter 4 weeks ago.

Probably what you were referring to in your quote (if it is useful at all). And, of course, will depend on how reliable WeekIntQ is: I have only played with this for announcement and never implemented or checked the reliability.


From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Aug 13, 2019 7:03:29 PM       
Edit 3 times, last edit by Jrinne at Aug 13, 2019 7:15:57 PM
Re: Historical *EPSStdDev

Thank Jim, I think WeeksIntQ > FHist(WeeksIntQ,4) will work for detecting the start of a new quarter. It's actually kinda neat. I don't think WeeksIntoQ is anything but a counter, but that works here.


Aug 13, 2019 8:11:07 PM       
Re: Historical *EPSStdDev

Strangely enough, I think this works most of the time;
(EPSActual(0, QTR)-EPSEst(0,QTR))/FHist("CurQEPSStdDev",WeeksIntoQ+12)

At least, under Fundamental charts, the time alignment looks good between;

and if the alignment looks good for FHist("CurQEPSMean",WeeksIntoQ+12), then I hope it's good for FHist("CurQEPSStdDev",WeeksIntoQ+12)

ANN SUE may be trickier to get right.


Aug 13, 2019 8:54:53 PM       
Edit 3 times, last edit by wwasilev at Aug 13, 2019 9:32:07 PM