explicitClick to confirm you are 18+

I’ve astounded even myself!

Deep CaptureMar 11, 2020, 7:57:12 PM
repeatthumb_up1thumb_down

Originally posted on 02/22/2010 by Judd Bagley

Last week I dared to predict the number of then unreleased delivery failures in shares of Sears Holdings (NASDAQ:SHLD). The figures I ultimately settled upon (after two minor tweaks in the days to follow) were:

1/29/2010 879,444

1/28/2010 873,222

1/27/2010 870,570

1/26/2010 851,904

1/25/2010 848,742

1/22/2010 865,266

1/21/2010 857,106

1/20/2010 1,535,508

1/19/2010 1,540,914

This morning, the SEC having yet to release the numbers, I explained the basis for my prediction, which is in simple terms, based on a pattern of apparently manipulative naked short “reset” transactions the Deep Capture team has observed in SHLD, and how those trades consistently predict how many shares of SHLD will fail to deliver after two days.

Well, literally moments after I published my explanation, the SEC released the numbers.

I asserted that my prediction would be accurate to within +/-2.5%. I’m embarrassed to reveal that in the end, my predictions were on average accurate to within 2.55%.

Sorry. I’ll try harder next time ;).

For the record, here’s a table comparing my predictions and the actual numbers.


At this point, the most relevant question is: will the SEC do anything about this obvious violation of the law?

Sadly, I predict that no, the SEC will not.

The next most relevant question is: will the manipulator continue manipulating, knowing he or she has been spotted?

Because I suspect the manipulator is operating without much concern over being held accountable, I must also predict that no, this pattern is unlikely to change.

However I can confidently predict that Deep Capture will continue identifying and reporting on these sorts of abuses, in the trading of SHLD and several other companies, in the very near future.

Mostly inconsequential postscript: Deep Capture team member Patrick Byrne points out to me that instead of average error, I should have used weighted error, in which case my prediction was off by just 2.51%.