Studying changes in GLD bullion inventory levels is potentially another example of a Frank-Holmes-like-card counting strategy.
When playing Blackjack, the house has the edge in winning ... unless you know how to count cards, a strategy proven to increase a player's chances. But did you know you can apply this concept to the market as well?
While card-counting may get you barred from a casino, you can freely use patterns to your advantage in investing.
Take the Purchasing Manager's Index (PMI), which gives us a reliable indication of the manufacturing sector's well-being...
Resource Investors Who Use this Strategy Have Seen Significant Gains
Specifically, I believe that watching the GLD bullion inventory levels provides an important signal for anyone trading gold equities. For NEM in particular, changes in GLD bullion inventory levels is an excellent signal to enter/exit long/short positions. The signal also works, albeit with diminished results, for other equities such as Agnico Eagle Mining (NYSE:AEM), Randgold Resources Limited (NASDAQ:GOLD), and NYSE ARCA GOLD BUGS INDEX (HUI).
In this article, I present a simple trading rule based on GLD bullion inventory levels, along with backtest results that come from applying it to NEM, AEM, GOLD and the HUI.
A recent article by Steven Saville, What Do Changes in GLD'S Bullion Inventory Tell Us About the Future Gold Price?, concludes that the answer to the question in the title is "nothing". To make his point, Mr. Saville provides an authoritative overview of the internal workings of ETFs in general and GLD in particular and his article is well worth reading to learn more about ETFs.
What I found most interesting in the article was the general shape alignment between the gold price chart and the GLD Bullion inventory chart. In fact, it is this general alignment that potentially makes the GLD signal useful for anyone trading gold producers.
To showcase the similarities between the price of gold and GLD bullion inventory levels, I've prepared a number of normalized gold price and GLD bullion inventory charts for various time scales.
As the first chart shows, from the introduction of the GLD to the present, the general shape of GLD bullion inventory levels is quite similar to the shape of the price of gold plot. They generally rise and fall together.
By narrowing the display timeframe down to start from 2014, then to start from 2015 and finally to start from September 2015, you can see visually that there is a very good degree of correlation in the two plots.
A visual review suggests that the GLD bullion inventory chart is a smooth version of the gold price plots without the day-to-day variance inherent in the price of gold.
Please see Using Normalized Closing Price in Anticipating And Trading The Agnico Eagle Mines 2015 Full Year Results to learn more about normalized plots, their construction, and related cautions.
GLD Bullion Inventory Levels As an Entry/Exit Trading Signal
My prototype trading algorithm runs as follows:
If there are two consecutive increases in GLD bullion inventory levels, enter a long position the day after the second increase occurs.
If GLD inventories decrease, exit a long position the day after the decrease occurs.
If there are two consecutive decreases in GLD bullion inventory levels, enter a short position the day after the second decrease occurs.
If GLD inventories increase, exit a short position the day after the increase occurs.
Otherwise maintain position or remain out of the market until a new entry/exit signal is generated.
Note that between two consecutive increases/decreases there typically is a period of days during which the GLD bullion inventory levels remain constant.
To avoid look ahead bias, all entry and exist costs/profits/losses are computed using the equity closing price on the following day. The starting balance is assumed to be $10,000 and fractional shares are used in this simplified model. Position size is a fixed percentage of the current balance. Current balance is computed using closing prices. No attempt has been made to optimize the algorithm. Rather, the algorithm is a raw interpretation of the visual clues provided in the charts above.
Applied to Newmont Mining Corp. (NYSE:NEM), this simple rule yields the following plots. In each, the allocate percentage refers to position size and to the commission fee per trade. The "buy and hold price change" is the percent change in share price over the time period and the black line is the linear regression of the Account Balance time series, shown in blue.
As expected, applying a transaction cost per trade also reduces returns.
Using a more conservative capital allocation figure (i.e. applying 50% of balance to each trade instead of 90%) also reduces returns.
However, the returns over the full time frame of GLD data are still potentially interesting. Also, if one looks at a subset time frame from say 2009 to present day, the algorithm results look significantly better.
Performance results are not as good for HUI or AEM but still potentially interesting if one looks at the 2009+ window.
For AEM, if one picked January 2, 2009 as the start date, the AEM returns over the time window would have been -34% while the results the GLD flow trading rule results would improve.
For Randgold Resources Limited, the trading rule fails does not fare as well as a buy and hold strategy, although using a January 2, 2009 comparison start date would make results almost equal (GOLD gained 81% from Jan2/2009 to Feb5/2016).
Although I am not suggesting that trading gold producers based solely on this unoptimized, simplistic algorithm is a good idea, I do believe that the evidence presented above is enough to make serious traders consider whether the GLD bullion inventory deserves a place in their trading algorithms.
I have made the day-by-day changes in GLD bullion inventory levels I computed available as a CSV file for download by Pro Subscribers. The file gld_action.csv contains the action indicator ( 'open long', 'open short', 'close long', 'close short' or 'no action') for each day since 2004-11-19. The action is to be executed on a subsequent day since GLD data is only released after current day market close.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: I will look for PUT option purchase opportunities once I see GLD bullion inventories start to decline. I do not have any long positions now given the unprecedented GLD holdings increases from the start of the year.