The drawing of horizontal trend lines, support and resistance levels, Fibonacci levels and so on is a common technique for analysing price behaviour, setting stop losses and making trading decision. For examples of what I mean, see here, or here. However, in the case of GLD (and pretty much all metal ETFs) drawing them horizontally can be misleading.
The reason is that almost all precious metal ETFs pay for their management fees by selling gold from the fund to pay for the fees (while the number of shares outstanding remains the same). The result is that the amount of metal backing each share declines over time. This means that a share today is not equal to a share from the past – the prices are not equivalent. You are not comparing apples with apples, actually, you are comparing an apple with a fractionally smaller apple.
Most funds disclose the amount of gold behind each share on their website, usually listed as net asset value (NAV) in ounces. For the largest gold ETF in the world, GLD, its website was showing “NAV (in gold oz) Per Basket” as 9,585.06 ounces on the 14th. A basket (which is used by market makers) for GLD is 100,000 shares and when GLD started in November 2004 each share was equal to 0.1 ounces, so a basket in 2004 was equal to 10,000 ounces. Compared to 9,585.06, that is a loss of 414.94 ounces, or one London Good Delivery bar over the past 10 years – a fair bit when you think about it.
That loss is equal to 4.15%, which is basically GLD’s management fee of 0.4% per annum over 10 years. You can see the decline in the real underlying value of GLD’s shares by downloading its historical data. According to that spreadhseet, GLD’s closing price on the 14th of July was $110.74. Simplistically dividing by 0.1 would give you $1107.40 per ounce, yet we know that gold never got that low. If you divide the $110.74 by GLD’s actual gold backing of 0.09585063 you get $1155.34, which makes a lot more sense.
For a graphical example, the chart below takes GLD’s closing prices and draws some Fibonacci retracement levels from gold’s high, and from the 2008 low, but corrected for GLD’s declining gold backing.
You can see how they slope downwards over time, reflecting the loss of gold from the fund. It might not look like much on the chart, but consider the table below which shows the difference between the figures that would be normally be drawn horizontally and what those levels are adjusted to today to reflect the reduced gold backing.
| Level |
August 2011 Figures |
Adjusted to Today |
| 100% Fib |
$184.59 |
$181.71 |
| 61.8% Fib |
$114.08 |
$112.30 |
| 50% Fib |
$92.30 |
$90.85 |
| 38.2% Fib |
$70.51 |
$69.41 |
| 2008 Low |
$70.00 |
$68.17 |
A $2 to $3 difference on a $100 share is a fair margin of error, causing you to make a decision earlier than you should. Given the relatively small management fees, the effect is not material if you are doing analysis for period of less than a year. Something to keep in mind if you do draw levels on ETF charts, or the next time you see a technical analyst doing the same.
