We've had a lot of talk lately about gold begining it's 3rd phase (especially HoBo-Jo's excellent analysis here). The counterpoints to that have been to point out golds' strong fundamentals, for example, growth of emerging markets (China, India), massive and increasing world debt levels, relatively small proportion of global investment into gold and so on. These fundamentals certainly make the long term outlook bullish but they really can't be used to rule out a bubble. The problem is that such arguments aren't sufficiently quantitative to tell us whether the price growth we're seeing is within reasonable bounds. In other words just because the fundamentals are strong doesn't rule out that manic buying might be taking place. So I was fascinated to come across a recent article by the World Gold Council that uses quite a different type of argument, a purely statistical technique (z-scores) to show that gold is not in a bubble. Here's the key chart from the article: Several markets are displayed above and notice how a z-score > 2 makes an excellent predictor of a market crash. In fact for gold in 1980 the z-score reached almost 5 (that's actually off the chart!). So based on z-scores a crash isn't imminent today. Perhaps more importantly by monitoring the z-score going forward, if you see it go above 2 that's an early warning that a crash could happen soon.
I'm not sure a "normal distribution" applies to our present situation of gold priced in fiat, which is not a closed system and the population of fiat is essentially limitless. The limits of extraction of gold from the earth might be a better place to apply Z.
It is the Z score of rolling annual returns, and they surely are not limitless. Looks like a useful indicator.
As THUCYDIDES79 points out it is the Z score of 'rolling annual returns', not the price of gold in fiat. I should have been clear about that. So for example if a market delivered 30% annual returns every quarter over a long period of time, the price is rising at a rapid (indeed exponential) pace and yet the z-score would be near zero. But if that same market suddenly saw a 100% and then 140% return in it's last two quarters the z-score would spike because of the rapid change of pace.