Trading Strategy: Gold vs Gold Stocks
Last month I shared a strategy that used the ratio between oil stocks and the price of oil itself to trade the oil sector (read part I and part II).
Keeping with that theme, in this post I’ll look at using the ratio between gold stocks and gold to trade very short-term moves in gold. I’ve been wrestling with this observation for a while, and I’m sharing the basic concept here because I think the strategy is not ready for primetime (more on this later).
First, a look at the ratio of gold (represented by the ETF GLD) over the gold sector (XAU) from 2005:
Over the last 4+ years, the two have traded in a fairly narrow range versus the other, but in mid-2008 the ratio exploded as investors embraced the “safe” (good for gold) and abandoned all things equity-related (bad for gold stocks).
Note: I’m using the ETF GLD to represent physical gold, and the index ^XAU to represent the sector, because I think those two are the most familiar to readers, but the observations in this post have more or less held for other vehicles such as futures (gold) and the ETF GDX (gold sector).
The Strategy
The ratio of gold vs the gold sector has exhibited a fairly strong contrarian tendency. The following graph shows the results of two strategies, the first (green) going long gold at today’s close if GLD underperformed XAU for the day, and the second (red) going long if GLD outperformed XAU, frictionless from 2005 to present.
[logarithmically-scaled]
The observation hasn’t been foolproof (note late 2005, early 2006, late 2007, and early 2008), but generally speaking, gold has been consistently stronger tomorrow when yesterday it underperformed gold stocks (and vice-versa).
For the number-lovers:
This strategy is exploiting a very small daily advantage (similar for example to adaptive daily follow-through), and therein lays a problem.
Most of the strategies that I talk about on this blog could be traded using leveraged mutual funds (not to be confused with leveraged ETFs). These are the only thing that I trade. Because they incur no transaction fees or slippage, most of the tests I’ve performed on this blog could have been duplicated, for all intents and purposes, as well in the real world.
Not so here. To the best of my knowledge, there are no mutual funds suitable for active trading that track gold (the gold sector yes, but not gold itself). Trading this strategy with ETFs/futures would bring trading frictions that would close an already very fine advantage.
I’m struggling now with a way to improve upon this advantage enough to make it tradable. I share it here in hopes that I’ll generate a spark amongst fellow quant’ish folks who frequent the MarketSci Blog. As always, more to follow.
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Stock Market Sectors, Trading Strategies | 10 Comments
Time to rethink my investment strategies.
Hi Michael, why not experiment with GDX (goldminers) and GLD (gold) using the “ETF Rewind” Pairs Analysis tool? I agree it goes through periods of flat versus stellar performance. But… try the following parameters just for kicks, a three day look back and a z-score of 1.1. Cheers, Jeff
RE to Jeff: very slick sir.
Is there a way to see that test over a longer period of time? The problem is that the sector predictability is very unstable (i.e. only the gold side exhibits predictability). I fear that the big runup the PA tool is showing is purely a result of a one off event (mid to late 2008).
P.S. to readers who might be reading this comment, if you trade ETFs, you need to be keeping your ears to ground for Jeff’s new ETF Rewind tool which will be launching soon. It’s not a strategy in the sense of what I offer at MarketSci Blog – it’s an analytical tool (but a very powerful one).
I agree that it’s “what’s working now”… and that an adaptive version is the key. Because these are mean reverting pairs and conceptually should remain so, however, it’s worth tracking. I’ll see if I can’t get a longer study going! Best, Jeff
Mike – Ernie Chan did something similar in his book – ‘Quantitative Trading: How to Build Your Own Algorithmic Trading Business’, but it was more of an example of trading a pair of cointegrated items.
Its worth a look & may provide some new ideas.
Regards,
Eric
what about differential RSI2 — subtracting the RSI2 from the best performer of the pair vs the worst performer. Creating a percentile ranking of the difference with a lookback of 100 days would make thinks somewhat adaptive. entry could be at a percentile ranking below 10th and exit at 50th. this would give a holding period of 3-5 days and produce higher gains per trade
dv
Although this is somewhat off topic, does anyone know how the leveraged ETFs actually achieve whatever results they achieve? (There has been lots of talk lately about their failure to achieve their results long term. But that’s not what I’m asking.)
How does one know, for example, how much FAZ (the 3 times triple inverse of XLF) should sell for. What makes its price go up and down? What’s in the FAZ portfolio that would give it a NAV (net asset value)?
Thanks.
– Russ
RE to David: I presented the strategy here as a “follow-through reversal” because it’s easy to wrap the head around, but I tried approaching it using short-term RSI as well. Based on what I found (which is not to say it’s the end all be all answer) it was good, just not good enough considering trading frictions. But yes, I agree, this type of short-term observation is very suitable I think for a short-term RSI strategy.
RE to Russ: I have a general idea of the answer, but it’s not something I’m particularly versed in. Would love to hear from one of our savvy readers on this one.
michael
another interesting pair to consider would be oil v. gold, however it’s mean reversion tendencies play out over years.
russ,
i have used the leveraged ETFs for a while but primarily for trading…..although i am very familiar with how they work. A lot of the confusion and anger over tracking error is in general misplaced because these were designed to be trading instruments only. They achieve their returns by daily rebalancing using futures, options or stocks on the underlying. As a consequence because of the nature of compounding, they can rise far above, or fall far below an equivalent investment in the underlying instrument. Also because of market microstructure (imbalances in supply and demand) they can also trade above NAV temporarily if their is a shortage of hedging instruments as their was during the ban on short-selling. In general, i wouldn’t worry about these drawbacks as a trader (1-10 day horizons) but by no means should you use these for longer hedging purposes. Instead for longer term purposes you are better off with options. If you want ETFs to buy and hold that track the underlying more closely, focus on ETNs– exchange traded notes— which do not rebalance daily and are required to match the underlying at some future maturity date—and therefore are arbitraged to stay close to fair value.
dv