Gold Inching Ever Closer into what Will Likely Be a Major, Multi-Year Low
We’ve been through PMs fairly extensively as of late (this is our third on the topic in a month vs. eight total in the previous two years). Our most recent work on the space has generally been time-focused. Specifically, we’ve looked at the % of time in a trailing number of weeks the YoYs on PMs have been < 0% and in negative territory. By contrast, this post is an attempt to cut the analysis in a different, but corroborating fashion to our previous work by focusing on the price aspect of things.
To wit, we’ve discussed the importance of the 4 year/48 months/208 week cycle period in the past. Put simply, the average U.S. macro market cycle has lasted this long, on average, in the post WW2 world. Thus, why is it not reasonable to think about PMs like gold in a similar fashion since their machinations are largely driven by policy measures that are seeking to encourage or counteract the particular 4 yr macro cycle trend in play at a given point in time.
In effect, what we want to know is if PMs are oversold on a time basis, might they be the same on a price basis when compared to their macro cycle 208 week MAs?
For instance, in its current 4 yr sell-off the first time gold closed at least 20% < its 208 week MA was the week of 10/3/14, almost a year ago. At its worst, on 7/31/15 it closed ~24% (specifically 23.8%) < the same MA. As it stands, gold is currently ~22% below its 208 week MA.
We care about the above because a close for gold at least 20% < its 208 week MA is rare. In fact, dating back to the early 70s it’s happened only 79/2,269 weeks, making it a 96th percentile event. A close at least 22% < the 208 week MA is a 98th percentile event; at least 24% < as we saw the week of 7/31/15 is a 99.3rd percentile event, occurring in only 16/2,269 weeks.
Visually, the chart below shows where gold has closed at least 20% < its 208 week MA for the first time in at least a year (blue) and at least 24% < as well (red). Note that these signals are incredibly rare with the most recent set in the past year representing the 5th (at least 20% below) and 4th (at least 24% below) signals since the early 70s, respectively. Also take note of the fact that the signals have historically and effectively come at lows (1976, 1982) or close to them (1983/1984, 1997).
The table below depicts how gold has traded after getting either one of the latter signals. Our draw-down #s (i.e., ‘% loss’ column) are from the closing value as of the week of the signal vs. intra-week swing lows as of the trough week.
The most recent signals where gold closed at least 20% and 24% < its 208 week MA are in yellow. Below those signals you see estimated price targets (Px) assuming gold were to experience a loss similar to the ones it has in the past after generating these signals.
For instance, gold generated the 20% signal on 10/3/14 when it closed at $1190.7 and so far at its lowest point it has lost 8.1%; 49 weeks have elapsed off that signal. Historically, and on average for the three other signals that preceded the most recent, gold should trade down to $1050.7 and is theoretically 7 weeks past the date we would have expected it to bottom (the 1st of the four signals, 1976, came at its absolute low). However, if we assume a worst case scenario based on history, gold could trade down to a minimum of $901.9 and still has 37 weeks to bottom.
On the right, gold generated its 24% signal on 7/24/15 when it closed at $1098.4 and so far at its lowest point it has lost 1.6%; 5 weeks have elapsed off that signal. Historically, and on average for the four other signals that preceded the most recent, gold should trade down to $1023.6 and still has 26 weeks to bottom. However, if we assume a worst case scenario based on history, gold could trade down to a minimum of $985.3 and still has 79 weeks to bottom.
Despite some of the downside Px outlooks in the table above, “averages” across history are just that.
How can we drill down with more precision to derive a better, more accurate Px?
Well, recall from above that trading at least 20% or 24% < the 208 week MA is incredibly rare (i.e., 96th and 99.3rd percentile events). In addition, let’s now add to the discussion the fact that since the early 70s gold has never closed more than 27.3% < its 208 week MA (to reiterate, it is currently 22% below) Thus, should history repeat itself, we have a worst-case scenario for the lowest weekly closing print that gold should put up vs. its current 208 week MA of $1418, that being $1030 [i.e., $1418*(1-.273)].
Therefore, in referring back to our table above, some of the price targets for downside are simply too bearish. For instance, using the 20% threshold group from the table, if gold were to trade as it did following the most bearish result of this group, (11/18/83 = -24.3% peak-trough), it’s resulting low price of 901.9 would put the metal 36.4% below its current 208 week MA, a degree of bearishness vs. trend that has never occurred! The same logic applies to the 24% threshold group – the most bearish move was a 15.2% peak-trough loss after the 7/27/84 signal which would put gold at a low of $931.2 this time around but also 34.3% below its 208 week MA, also a degree of bearishness that has never occurred.
However, if gold were to bottom at ~$1030 on a weekly closing basis, it would have experienced a 13.4% peak-trough loss vs. its $1190.7 price as of generating the 20% signal and a 6.1% peak-trough loss vs. its $1098.4 price as of generating the 24% signal. Because the latter losses refer to weekly closing lows, they could be larger on an intra-week basis. Regardless, losses of 6.1%-13.4% would be in-line with all of the losses experienced in the table above on both the 20% and 24% signals which range from 0% to -24.3% and average 8.6%.
Moving on, I think that the latter analysis of gold vs. its 208 week MA can be further refined by seeing if the latter signal work regarding price-based trend properties (i.e., gold at least 20% < 208 week MA) meshes with the work from our recent post regarding time-based trend properties (i.e., gold’s YoY < 0% in at least 88% of trailing 165 weeks).
For instance, dating back to the 70s are there any examples of gold being at least 20% < its 208 week MA in the same week that its YoY has been < 0% in at least 88% of the trailing 165 weeks? As it stands, it has accomplished both as of last week and this week.
It turns out that the answer is ‘yes’ but only one such previous signal exists: during the week of 6/11/99. Gold closed that week at 260.4 and put in a secular intra-week low of 251.7, or 3.3% lower, only 11 weeks later during the week of 8/27/99. The rest was history for the next 12 years into its swing peak of ~$1900 in 2011. If gold were to decline 3.3% off last week’s close of $1122.6, it would imply a forthcoming low of $1085.
All of the above historical statistical data is well and good in terms of providing incremental clarity into how gold may act in the coming future as we search for what could be a durable bottom. That said, in my view, very little supercedes the clarity provided by a simple chart-based, technical look at the metal in determining upcoming support zones we should be considering in making an estimation of where such a bottom might come into play.
Consider the chart below and note first that long-term support line (1) – hugely important in my view as it runs atop gold’s former all-time high from 1980 and then across multiple of its congestion highs through all of 2008 during the formation of and depths of the global crisis – runs to ~$1035 as of today on a weekly closing basis. Second, recall that in another recent PM post we were comparing the technical similarity b/t gold’s recent move off its 2011 high until now with its 1974 high into its 1976 low; that specific comparison can be seen in this chart. There we suggested that if the technical similarities hold to create similar peak-trough declines of 47% this time around it would imply gold moves as low as $1000 in the coming weeks.
However, there’s nothing written in stone that suggests gold’s current peak-trough decline need be precisely 47% again as it was 1974-1976. For instance, a decline to $1,000 would put gold ~30% below its current 208 week MA and as I said before, since the early 70s gold has never traded more than 27.3% below that MA, or ~$1030.
So, more likely than not, based on both line (1) support and the historical MA data, I just don’t see gold, at least on a weekly closing basis, closing < $1030 in the intermediate-term as a durable swing low. There is potential for that level to be breached intra-week, potentially to $1000, but much less so on a weekly closing basis in my view.