The oil/gold ratio does it again!

There are many reasons to be dissatisfied with my commodities system from yesterday, but the most important one is that it does not even pretend to address the questions I had laid out for any such system when I started going on about commodities, namely how to explain the behavior of the all-knowing but utterly mysterious oil/gold ratio.

The peculiar relationship between precious metal/industrial commodity ratios, such as gold/copper and especially silver/oil, and interest rates that we mentioned yesterday has been especially difficult and irksome. To repeat, these ratios appear to have a long-term positive correlation but a negative short-term or intermediate correlation. As a proxy for gold/copper and silver/oil, I used the gold/oil ratio, simply because that is the one I have data for.

I believe I have come up with a solution to this paradox. As I have come to expect when dealing with these kinds of market questions, the answer—if that’s what it should turn out to be—is both simple and unexpected. Moreover, it raises still more questions about the mechanisms at work. And, for all the progress I think this represents, I still am unable to explain the engine behind the oil/gold ratio, although I feel rather confident that this is a major step forward. Moreover, it says some interesting things about the future of the markets.

But, let’s get to the solution, and you can decide for yourself.

Yesterday, I presented the following chart comparing the gold/oil ratio along with short- and long-term yields, as well as the yield curve. Even from this point of view, you can see that there is a relationship of some kind, but it is hard to pin down.

notice the long-term positive correlation versus the intermediate negative correlation

The solution is the following chart.

notice the mostly positive correlation between gold/oil & 10-year

I have moved forward the gold/oil ratio by sixteen months against interest rates. In other words, this is the gold/oil ratio ‘predicting’ interest rates sixteen months in advance. Sixteen months was the number that jumped out at me; perhaps fifteen or twelve or some other number would be better, but it doesn’t really matter (I think). You can see it is not a perfect fit, of course, and I strongly—very, very strongly—suspect that the silver/oil ratio will turn out to be a better fit, but apart from that, you can see that there is a bit of a problem with the fit, even allowing for the difference between gold and silver.

For example, when the oil/gold ratio broke above the critical level of 0.12 (see my ‘Supersystem‘ about the importance of these levels)—or below the 8.33 level on the gold/oil chart here—the correlation turned negative, and the gold/oil ratio began to ‘forecast’ the yield curve instead. This is a typical problem when trying to get the oil/gold ratio to speak to you. It correlates perfectly with the yield curve at one moment and with year-on-year oil another, but nothing really seems to stick.

But, I think this time, it’s somewhat different. Here’s a rolling five year correlation between gold/oil and the yield curve in real time (i.e., without the 16 month advance).

70s and early 80s: positive correlation matched by inverted yield curve

In comparison, here is the correlation between gold/oil shifted 16 months forward and the 10-year yield.

yield curve appears to follow correlation between gold/oil (shifted) & 10y yield (current)

The chart with the shifted gold/oil ratio seems to offer a better fit over the long term, but it is a bit abstract. Why would the yield curve follow the correlation? Why would a negative correlation between gold/oil and yields result in an inverted yield curve or a positive correlation in a steep one? Frankly, I am not sure at the moment, but I am sure that it means something.

For me, as suggestive as these charts are, the clincher is the ratio between gold/oil (shifted) and the 3-month yield (current), because it more or less sets the tone for the yield curve.

Even when the correlation btwn gold/oil and the 10y yield breaks down, it keeps ticking

same chart as above but I pulled some mathematical shenanigans to get the log of the yield curve

Considering that this is the gold/oil ratio shifted forward 16 months and how volatile that ratio is, not to mention how volatile the 3-month yield is, this is a remarkable correlation, I think.  It is especially interesting that it works so well, even when the correlation between gold/oil and the 10-year rate appeared to break down since 2000. The greatest divergence during that period appears to have been the period of Greenspan’s “conundrum”, namely the unusually low interest rates of the 2004-2005 period. If the gold/oil ratio is any indicator, long-term rates should (setting aside the fundamental causes for now) have been even lower. In fact, it appears as if short-term rates may equal the gold/oil ratio (shifted forward) subtracted from the current long-term rate.

can we model short-term yields?

Somewhat unexpectedly, it appears that subtracting the current gold/oil ratio from the yield curve generally gives us a better approximation of the 3-month yield than does the gold/oil ratio shifted forward. But, does that mean the 3-month is the difference between the gold/oil ratio sixteen months ago and the same ratio today? Glad you asked.

not a perfect fit, but more predictive than I'd have guessed

Assuming this chart has any meaning, what is especially interesting is the dead end we seem to have come to. The gold/oil-gold/oil number has been narrowing into a triangle and (at least as of November 2009) broken sharply downwards, but if this number is predictive, there is not much room to run for yields, especially the three-month.

What could this possibly mean?

The only significant upticks in yields that were not preceded by an uptick in this  gold/oil-gold/oil number were those of the 1970s. The similarity we have noted before between the mid-1970s and the post-financial crisis world is perhaps salient, but I am completely exhausted and am too tired to engage in speculation about where interest rates are going to be headed now.

As I write this, however, the two-year rate is at 0.257%.


No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: