Sunday, January 24, 2010

Gold-At-A-Discount Indicator: A Cautionary Note

I've picked through the workings of the gold-at-a-discount indicator, and have found it wanting for its original purpose. Over the last six-month timeframe, a simple and intutive accumulation plan - buying an ounce of gold at the beginning of each month - was less expensive than a plan that uses the indicator as a timing aid.

There are two graphs below these two explanatory paragraphs. First of all, a graph with a green line put on the 10 line. The first chart is the spot price of gold divided by the price of one share of GLD. When this ratio is below 10, physical gold is selling at a discount to paper gold as represented by GLD. Ten shares of GLD correspond to one ounce of gold.

Just below it is a chart of gold using the same timeframe. The vertical green and blue-green lines show the buy dates given by the gold-at-a-discount indicator. The buy dates are the day after the indicator goes below 10. That date-shift is crucial in understanding why it's not suited for its purpose:

At first blush, the indicator looks fine - if laggy in places. However, its purpose was to act as a timing aid for a buy-and-hold plan. In testing it for this purpose, I assumed two monthly accumulation plans.

Plan #1 was buying one ounce of gold on the first business day of each month, at the London PM gold fix as shown here. For North American buyers, the PM fix is the most likely price that'll be had. [It kicks in before lunch break, North American time.] I ignored transation costs because I assumed they'd be the same for both plans.

Here's the schedule for the straight monthly-accumulation plan:
Aug. 3rd: 1 ounce @ $959.75
Sep. 1st: 1 ounce @ $955.00
Oct. 1st: 1 ounce @ $1,004.75
Nov. 2nd: 1 ounce @ $1,062.00
Dec. 1st: 1 ounce @ $1,192.50
Jan. 4th: 1 ounce @ $1,121.50
Total: 6 oz for $6,295.50.

Plan #2 is a little more involved, as it uses the gold-at-a-discount indicator. If the indicator goes below 10 in a month, and it's the first time that month, then the ounce of gold is bought the next day using the same London PM gold fix.

If it goes below 10 a second time during the same month, the signal is ignored. The gold budget for the month is already blown.

If it doesn't go below 10 in a month, then no gold is bought during the month. Instead, the funds are saved for the next time. As of the next trigger day, a cumulative buy is made that brings the plan up to speed again. In the approximately six-month period, there were three duplicate signals [marked in blue-green vertical lines] and one month with no signal. Since there was none in December, two ounces are bought in January at the January price.

Here's the schedule for the indicator-assisted monthly-accumulation plan:

Aug. 3rd: 1 ounce @ $959.75
Sep. 3rd: 1 ounce @ $983.00
[Sept. 4th: second monthly signal; not acted upon.]
Oct. 7th: 1 ounce @ $1,040.25
[Oct. 30th: second monthly signal; not acted upon.]
Nov. 4th: 1 ounce @ $1,090.00
[Nov. 30th: second monthly signal; not acted upon.]
[None for December]
Jan. 5th: 2 ounces@ $1,121.50 each

Total: 6 oz for $6,319.50.

The straight monthly-accumulation plan costs $6,295.50. The indicator-aided plan costs $6,319.50. The monthly-accumulation plan is $24 cheaper.

Using the AM fix on the same days makes the indicator-aided plan fare worse. The straight-monthly costs $6,268.25. The indicator-guided plan costs $6,326.00. The straight monthly-accumlation plan is $57.75 cheaper.

Rather than drag it out by shifting the dates around on the straight accumulation plan, even though refraining from doing so makes the above analysis a comparative price history of a relatively short period, I'm going to stop here. The beginning-of-the month accumulation plan is the most plausible choice of dates, as people typically get paid at the end of the month or nearabouts. If the indicator can't beat at first blush, then there's something wrong with it.


The gold-at-a-discount indicator sounds, and even looks, plausible. Why, then, was it found wanting?

My initial interpretation of it, that physical gold sells at a discount relative to SPDR Gold Shares when gold's oversold, was wrong. A better interpretation is that GLD leaps up higher than gold itself when the gold train gets rolling again. This difference makes the gold-at-a-discount indicator, for the practical purpose for which it was aimed to aid, a laggard. It lags too much, relative to the next day London gold fix, to be of real help to a regular gold buyer.

Therefore, I'm going to change the disclaimer to read "This indicator is only intended for reader interest." And/or, "A quick test has shown that it does not help with gold accumulation plans."

In certain market conditions, it regularly sinks below the 10 line. This daily graph for the last 3 years shows that, in late '08 and early '09, it wasn't helpful for much:

I've put a daily graph of gold just below it, for comparative purposes. For the sake of reader interest.

And, I'll continue tinkering around with the concept from time to time.

For the sake of reader interest.

No comments:

Post a Comment