Dust flux, Vostok ice core

Dust flux, Vostok ice core
Two dimensional phase space reconstruction of dust flux from the Vostok core over the period 186-4 ka using the time derivative method. Dust flux on the x-axis, rate of change is on the y-axis. From Gipp (2001).

Wednesday, July 10, 2013

Gold's changing anticorrelation to the dollar

This article posted a few days ago struck me as interesting, as it seemed so counter-intuitive that I thought it worth a look.

On the basis of weekly charts of gold and the US dollar index over the past five years, Charles Hugh Smith concludes there is no correlation. And at first glance, there seems much to support his view.

The principal argument (as I have never been a believer in peaks and troughs) is his second point--that there are at least three significant intervals where gold and the US dollar rose in tandem since late 2008.

Over the past few years I have attempted to show that most economic data are nonlinear and best studied by methods suitable for complex systems. Such systems are not easily analyzed using methods like linear regression or fourier analysis. In fact I would go so far as to say that such methods can lead you to the wrong conclusions.

The world's situation is complex and changing. Change can drive unpredictable variations in market preferences--so while it would seem logical that people's preference for US dollars and gold might normally vary inversely, perhaps there are some circumstances when the market equally seeks both.

We consider a scatter plot of USDX vs gold (weekly) from January 2007 to the end of last month.

There are a few segments suggesting correlation. From early 2007 until late 2009, the two data series appear to be negatively correlated. From late 2009 until about mid-2010, they appear positively correlated (they rise in tandem). Since mid-2010, they appear to be negatively correlated.

The gold price appears to be far more sensitive to the USDX in the second phase of negative correlation compared to the first phase; by which I mean that a small change in USDX correlates to a much larger change in gold price presently than was the case before mid-2009.

From a dynamics perspective, I would argue that the three areas of the graph represent different "states" of the (US? world?) economy. Finding the triggers for changing from one operational state to another is of key importance.

QE1 occurred during the V, from the top of the first negatively correlated segment through the positively correlated segment. QE2 occurred during the advance (lower line) in the second negatively correlated segment. QE3 occurred during the period of decline (upper line) in the second negatively correlated segment.  This all may be coincidental.

I expect we will continue to see relatively large fluctuations in the price of gold relative to changes in the USDX.

It looks like the system will have to drop down to the lower line before a steady advance in the gold price. If so, we would see a small, sharp drop in USDX without movement in the gold price, prior to a major move in gold (up) and the USDX (down).


  1. interesting - did you see this? http://chartistfriendfrompittsburgh.blogspot.com/2013/07/of-two-minds-has-temporarily-lost-its.html

  2. This is really Interesting stuff. But I wonder if your graph is distorted and should instead be plotted with Au price on a log scale, because USDX is a ratio, and because if I am buying or selling gold a doubling in price from $600 to $1200 should have the same magnitude as a doubling from $1200 to $2400. If you did plot it with Au on a log scale the yellow lines would be concave-down-to-the-left curves and would all be more or less parallel, and 3 more segments with positive slopes (green lines) would be more apparent. More systematic patterns fall out, I think.