More on Our Gold Model

Michael Stokes at MarketSci has a great post where he takes a closer look at our Gold Model. He found that using 2.4% as the break-even point produces a slightly better fit. (Here’s my original post.)

Michaels sums it up with three points; the good, the bad and the geeky:

The GOOD: there’s clearly something good about the CWS model. The analysis is made a little difficult by the fact that we have so little data to look at (the price of gold didn’t float prior to the late-1960’s), but the model definitely makes sense conceptually and fits the macro trends in gold over the last 40+ years.

The BAD: during both of those periods (in grey) when the model and gold diverged, long-term real rates were behaving very different than short-term real rates (and better matched the changes in gold). I’m wondering if the model could be improved by incorporating both short and long-term rates. Food for thought for future analysis.

The GEEKY: the model “works” by trying to predict month-to-month changes in gold, but has only done well modeling the absolute price level of gold. That’s awkward. A model that predicts monthly changes, but is only effective in terms of absolute price, is especially prone to curve-fitting because each monthly prediction impacts all future data points. Just something to roll around in the noggin’.

Posted by on November 1st, 2010 at 8:41 am


The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.