A Bubble in the Pot of Gold?

Wed, Sep 29, 2010

Market Commentary

The comment I posted in response to Jared’s latest feature article, “Why Conventional Hedging Methods Fail”, alluded to my opinion that gold is the current leading candidate for what I’ll call “bubble watch”. A growing number of analysts are jumping on the gold-bubble bandwagon, but predicting when a bubble will burst can be frustrating (and capital-crushing).

My comment referred readers to Didier Sornette‘s book Why Stock Markets Crash: Critical Events in Complex Financial Systems, and my interest in applying his model to gold futures. Sornette has a team of mathematicians and a super-computer at his disposal (not to mention the proprietary details of his latest models), so my efforts are feeble in comparison; nevertheless, a chart of gold prices, overlaid with the four most recent market bubbles (technology, housing, the financial crisis [represented by SPX], and oil) tells an undeniable cautionary tale.

What I see in the above chart is an exponentially accelerating price trend, with increasingly frequent sharp sell-offs—a crude, but fairly accurate, description of Sornette’s bubble model. Granted, chart analysis can be subjective, to say the least, but in combination with that indescribable intelligence we call “intuition”, the chart of oil prices in June of 2008 told me the price of oil was in a bubble that was in imminent danger of bursting. The pop came less than a week later.

Now, I’m not predicting that gold will crash next week, next month, or even next year. I merely wish to point out how precipitous the run-up has become, and that the chart suggests it would be wise to be cautious. (If you invested in gold back in 2002 [or 2003 or 2005, or even 2009], would it be smart to sell some of that stake now? Uh…yeah.)

Getting back to the application of Sornette’s model to gold, the latest information I could find indicates that it accurately (but not all that precisely) predicted the critical inflection point in October 2009. However, the bubble (yes, I’m sticking my neck out and calling it what it is—but I won’t go so far as to predict exactly when the bubble will burst) began reflating less than three months later—so, clearly, there’s a larger-scale pattern playing out.

For want of a super-computer…

P.S. – A recent article from Reuters Hedgeworld (free registration required), entitled “5 points to consider with Gold at new all time highs (in USD)”, presents some good arguments in favor of continuing to ride the gold bull market (as long as you don’t do it with gold coins).

Homepage photo courtesy of Flickr member Artnow, under Creative Commons license.

Tags: , , , , ,

12 Comments For This Post

  1. Josh Says:

    If you look at gold on a log or semi-log chart it doesn’t look like faster than exponential growth, which would be a more appropriate way to visually evaluate such a thing. In fact plotting a slope of the log weekly close shows that the past year’s growth rate is less than that of the May 2006 and March 2008 peaks.

  2. Frank C. Says:

    Well, Josh, as I admitted in the post, technical analysis has more than a grain of subjectivity to it—but your comment prompted me to take a closer look at a log-scaled weekly chart for gold, and I see a very different picture.

    I constructed a log-scaled weekly chart of gold prices since October 2000 and overlaid linear-regression channels for three distinct periods. The slope of the channel from 2001 through 2005 shows a healthy exponential growth rate. But things started to change in 2006.

    From 2006 through 2009, the gold market was, understandably, much more volatile. But when you average out the violent ups and downs, the linear regression slope, on a log scale, increased significantly. After the credit crisis, volatility in gold contracted, which, to me, indicates increasing correlation among traders. More important, the linear-regression slope (again, on a logarithmic scale) steepened even more.

    Another warning sign you’ll see on the linked chart is the increasing frequency of speculative peaks. I’ve numbered them 1, 2, 3 and 4—and it sure looks to me like their frequency is increasing at an exponential pace.

    Again, it’s very, very hard to time the peak of a bubble—but when all the red flags are flying, one could do worse by not exercising caution.

  3. Josh Says:

    I noticed that as well but I actually plotted the slope of the log converted prices so as to try and bring (some) objectivity to my view. Depending on the calculation length for slope or momentum, it is either increasing, flat, or decreasing. Personally, I think at least an intermediate top is due but I’m not so sure about longer term. But at this point I don’t think much would surprise me in markets.

  4. eber terandst Says:

    In a previous article, Sornette predicted the big bear market that would start in 2003. That was a bit wrong.
    Other than that, your site is my first reading every morning.
    Excellent job ! !

  5. Frank C. Says:

    Thanks for the info, Eber. I’d appreciate any additional details or references you might be able to provide.


  6. eber terandst Says:

    Re. Didier Sornette.
    Please see his paper in SSRN 454960. Or you might follow this link directly to SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=454960
    Or, to save you time go directly to page 21, Fig 1, in the paper.
    Beware of matematician bringing predictions . . .

  7. gappy Says:

    I am a regular reader of the blog, but I think you’re off on this one. For several reasons. First, because Sornette has a poor track record (for a damning indictment, see http://falkenblog.blogspot.com/2010/06/dishonest-forecasters.html).He has already called for a gold collapse (http://www.technologyreview.com/blog/arxiv/25269/) and it turned out to be a false alarm. Second, because he is not original. The discipline of Econophysics is overcrowded with second-class physicists dabbling in finance. In 15 years since its founding, it has produced nothing, really. Same thing for physicists working on networks and complex systems a la Barabasi or Huberman. Same trendy ideas and methods(self-organized criticality, power laws, fractals, Levy Processes), same empirical disconfirmation. And as a physicist, Sornette is certainly not Giorgio Parisi or Phil Anderson.

    Lastly, although I vaguely believe in bubbles, I am more inclined to believe in them when they are explained by a theory, or at least a narrative. There was one for 2000, and one for 2008. Those who called the top had consistent and similar narratives. Because of the confidence given by the internal consistency of these theories, they were comfortable staking their reputations and fortunes. I am waiting to audit Sornette’s investment record and verify his willingness to stick to his shorts when he’s down 30%.

  8. Frank C. Says:

    Thanks for the reference, eber. I probably won’t have time to read the whole paper for a week or so, but based on a cursory glance, I have two comments:

    - In this paper the authors are talking about an anti-bubble theory, which is less developed and less tested than the bubble model;
    - Sornette’s bubble model attempts to find critical inflection points–he explicitly makes it clear that a critical point does not predict a crash…the equations only model a time when market instability peaks, marking a point at which the probability of a crash is highest. The model includes the strong possibility that the market will level off, or even go higher, from that critical time point.

    More to come…

  9. Frank C. Says:

    Thanks again, Eber–now that I’ve had a chance to skim over the paper, I see that Sornette’s anti-bubble curve did not play out as predicted. Nevertheless, I still think log-periodic patterns in time and amplitude are useful for spotting potential bubbles (and for tracking anti-bubbles). You can be assured that I’ll be writing more about this in the future as I do my own research.

  10. Frank C. Says:


    Admittedly, technical analysis leaves more than a little room for interpretation. FYI, I’ve uploaded my log-scale chart of GLD, highlighting steepening linear-regression channels and exponentially more frequent price peaks: GLD, Weekly (http://www.condoroptions.com/wp-content/uploads/2010/10/Gold-Exponential-20101012.gif).


  11. Frank C. Says:


    Thanks for the insights and references. Clearly, applying models from physics and mechanics to financial markets willy-nilly is a bad idea. Nevertheless, I do think Sornette has made some pretty compelling arguments showing how his premise is consistent with trading behavior and traditional market theory. Two quick points:

    As I’ve already noted, Sornette’s model is designed to predict the critical time when a crash is likely, but a crash is not necessarily the only outcome. IMO, any representation that he claims to be predicting crashes is off-base. Even traditional technical analysis, as understood by an experienced chartist, is a game of probability based on identifying extreme conditions; it’s not a crystal ball, but rather a gauge of risk. As the time-honored saying goes, “Markets can remain irrational longer than you can stay solvent.” If M. Sornette is actually shorting the market solely based on his model, I would agree that an audit of his returns probably would be rather dismal.

    I don’t know what your criteria are for bubble-mania, but personally, I see a strong narrative around irrational bullishness on gold. More than a few otherwise rational traders I associate with believe that a complete collapse of the global financial system is imminent and that gold somehow will be the only asset that comes out unscathed. Small shops that offer to buy your gold jewelry (at a huge discount) are popping up all over my area like those Halloween stores that suddenly appear and are gone the next time you drive by. Countless fly-by-night boiler-room operations are pushing gold coins (at a huge premium) through direct mail, late-night commercials, ads in publications (especially ones with an older demographic), and telemarketing. All that is evidence enough for me.

    So, with all due respect (and, again, appreciation for the citations), I’m not convinced that buying gold right now isn’t enormously risky.


  12. Frank C. Says:

    If proven correct, Contrarians will mark this day as proof of their model:

    Gold settles at record high of $1,370.50

Leave a Reply


Jared Woodard specializes in trading volatility as an asset class. With over a decade of experience trading options and other volatility products ... Read More


Open All | Close All