Monday, September 2, 2013

Unscheduled stop


Hold everything. I was right.

I went thru the motions yesterday for this morning's post. As described, I got the FRED TCU data and plugged into Lars Christensen's spreadsheet in place of, actually in place of his S&P500 data. Excel picked up that data, reworked the numbers, and adjusted the lines on the graph to agree with the revision I made. All very nice, and the error that I claimed would appear on the graph did not appear. So I had my tail between my legs then, and tried to make the best of it with a funny post title.

But just now, or about 20 minutes ago really, it occurred to me: it is the NGDP series average that Lars adds in to his Market Indicator in the wrong sequence. Not the S&P500 series average. I went running upstairs, turned on the old computer, made another copy of Lar's spreadsheet with my notes (from 1 September) and plugged in the Total Capacity Utilization data in place of Lars's NGDP data. Here is the resulting graph:

Graph #1: The Red Line is Lars Christensen's Erroneous Calculation
The blue line is the original source data, in this case the TCU data. The orange line tangled up with it is my version of Lars's calculation. The red line up high is Lars's market indicator with the TCU numbers plugged in, in place of the NGDP numbers.

My numbers are down near the source data, where they are supposed to be. Lars's numbers are ridiculously large and obviously in error.

This is what I expected to show earlier today. Lars's calculation adds the TCU series average to his index number first, and multiplies by the TCU standard deviation after. His sequence of operations pushes his market indicator line way up high, as expected.

//

Here's the spreadsheet. It contains my VBA code for formatting my graphs. But you can disable that, as it is not used in any calculations and the graph is already done.

2 comments:

Jazzbumpa said...

If you subtract some constant in the range of 125 from Lars' numbers, do the lines then overlay?

Maybe that is the explanation for Lars' apparently arbitrary subtraction of 1.5 [IIRC] that got you so piqued in the first place.

Cheers!
JzB

The Arthurian said...

But Christensen's calc already has a -1.5 fudge factor in it. Now we need another fudge factor? Anyway it is the arbitrariness of the fudge that I object to. If we need to eyeball the graph and make the adjustment, then why bother with all the preliminary calcs to line things up?

Besides, my version of Christensen's calc is in the right place and doesn't need the fudge factor. Because I got the sequence of operations right.