HUI/Gold Ratio Limitations
This latest buy signal is once again highlighting the HUI/Gold Ratio, or HGR, indicator. It became fairly popular a few years ago for a variety of reasons, not the least of which was the ease of building simple ratio charts at StockCharts.com. The effortlessness of charting the HGR led to entire trading systems being developed around it, and some of these systems yielded excellent major buy and sell signals for the HUI.
I find myself interested in the HGR again this week for a couple reasons. Some of our subscribers at Zeal use the HGR as their primary trading indicator, so they are excited about the latest buy signal. The HGR is also interesting though as its evolution of effectiveness illustrates the typical lifespan of an indicator. Most trading indicators are not perpetual in their utility, but work best for a finite period of time before fading.
With newfound interest surrounding the HGR's latest buy signal, it is an ideal time to examine both its utility and limitations. In order to grasp the significance of the HGR though, it is important to have a rudimentary foundation in the basic mechanics of ratio analysis.
A ratio is simply one number divided by another. In the case of the HUI/Gold Ratio, naturally it is the daily close of the HUI divided by the daily close of the price of gold. When this ratio, or any ratio, is charted over time it provides an excellent running representation of relative strength and weakness between the two variables.
When a ratio is climbing on a chart, it means the top number is outperforming the bottom number. In the case of a climbing HGR, the HUI is either rising faster than gold like this past week or falling slower than gold in order for the HGR to rise. Conversely when a ratio is falling, it means the bottom number is outperforming the top. For the HGR this happens when gold rises faster than the HUI or far more typically when gold falls slower than the HUI.
Generalizing, usually if this ratio is rising significantly it is during a major HUI upleg, as the HUI gold stocks tend to rise much faster than the gold they mine. If this ratio is falling significantly though, it usually means a major HUI correction is underway. Leverage is a double-edged sword, so gold stocks fall faster than gold in their periodic corrections. If gold falls slower than the HUI it is outperforming the HUI and lowering the ratio.
Like any indicator, the HGR has limitations. I've been well aware of these since I first started following this indicator, and in past essays on the HGR I have said that it is best used only as a secondary confirmation, not as a primary trading signal. Nevertheless, due to its ease of use and seeming clarity a lot of investors and speculators have been using it as a primary. It is crucial they understand its limitations.
Before we delve into these limitations, it is useful to consider the bull-to-date history of HGR buy and sell signals. There have been seven of each since 2001. A buy signal is defined as a ratio resistance breakout, when the HGR establishes an upper resistance line but then breaks above it early on in the next major upleg. A sell signal is defined as the ratio falling under its 50-day moving average.
While this clever HGR trading system sounds complicated in the abstract, it is easy to understand when drawn on this chart. Viewing all the HGR buy and sell signals in context also reveals how much more common they have become in the past year or so. Where HGR buy and sell signals used to be rare, now they are happening with increasing frequency. Each of the seven sets of buy and sell signals are numbered below.
Most trading indicators, including this HGR, have a mathematically precise component. If the HUI closes at X on a given day and gold closes at Y, the HGR is absolutely X divided by Y and is perfectly accurate to multiple decimal places. One of the problems with virtually all indicators is some traders equate this mathematical precision with the ability to predict with accuracy. This assumption is inherently flawed though, always.
While the HGR can be known with precision at any time, two elements can wreak havoc with its predictive ability. The first is probably the most important and is outside the control of even the most adept student of the markets. Past market behavior is not necessarily predictive of future market behavior. Since all indicators and trading systems rely on past data and observations to some extent, they can all fail when the future doesn't remain congruent with the past. Unless you are God and know the future, there is no way around this limitation.
Second, even precise indicators have to be interpreted and these interpretations are fraught with all kinds of subjectivity that cannot be eradicated. In the case of the HUI/Gold Ratio, defining both buy and sell signals, though they seem straightforward, requires all kinds of subjective judgment. Obviously the more subjectivity is injected into interpreting an indicator, the more its signals risk becoming nothing more than pure opinion.
Starting with HGR buy signals, they require a "ratio resistance breakout" to trigger. Seven blue ratio resistance lines are rendered above. While they look pretty straightforward at a glance, they are all hand-drawn like the vast, vast majority of all technical lines. The chartist, in this case me, has some discretion on where to draw a given line. This can affect the slope of the line, and hence its intercept with the ratio on the breakout. If 10 technicians drew one of these lines, we would have 10 slightly different slopes and intercepts.
Now there are ways to draw mathematically precise chart lines, but even in this computer age they are barely ever used. And even if a computer does draw a line, subjectivity still exists. Take ratio resistance line 2 above as an example. At the very dawn of 2003 there is a HGR spike above resistance. Should we tell our computer to connect the very top of this spike with the initial spike in Q2 2002 which would flatten our resistance line and raise our breakout buy signal intercept? Or should the computer draw a best-fit line slicing through all the high points? No answer is absolutely right and all are subjective.
So the blue resistance lines, even if computer drawn, require human judgment and decisions. More decisions are required for defining a breakout that marks a buy signal. For example, if the HGR closes 0.1% above the resistance line for one day and then falls back under it for ten more, was this brief foray above a breakout? What if it closes 2%+ above for three days and then falls back under for three more? Even the breakouts have to be interpreted injecting subjectivity into HGR buy signals.
One method to minimize subjectivity is to wait a few days or weeks to watch an evolving breakout. If the HGR breakout stays above resistance, it is real and a buy signal. If it doesn't, it is not a buy. This sounds good on the surface, but the longer we wait to define a buy signal in real time the worse our actual entry point will be. If a trading system's signals can't be identified and acted upon right away, their utility rapidly collapses. So ex-post-facto interpretation adds more subjectivity and worsens entry and exit points.
Lots of limitations already! While I am using the HGR as an example here, very few trading systems designed to catch big swings, major uplegs and corrections, are purely mechanical. And even for those handful of mechanical systems, the assumptions that go into defining the underlying mechanics ensure that subjectivity always enters the mix. Human judgment is everywhere in trading, and inexorably and irrevocably intertwined in it.
Thankfully HGR sell signals are not quite as subjective as the buy signals since no technical lines need to be drawn. All that has to happen for an HGR sell is for the ratio to break under its 50-day moving average. Once again though how is this failure defined? Is it 0.1% under for one day? 1%+ under for five consecutive days? What if the ratio quickly goes back above its 50dma? Was it not a real failure then? No matter what decision criteria are used, it is subjective.
It is this great degree of subjectivity, capricious human judgment, that is the greatest limitation of the HUI/Gold Ratio and most other trading indicators and systems. No matter how careful you are with this indicator you have to make many assumptions and they will affect its utility without a doubt. There is absolutely no way around this fact. Thus the HGR is probably best used as one of many indicators, not as a one-ratio show.
Another problem with the HGR is evident above in its frequency of flashing signals. Prior to mid-2005, HGR buy and sell signals were rare. The buys really tended to flash near the early days of major uplegs while the sells really tended to flash near the early days of major corrections/consolidations. As such, back in the HGR's glory days it really was quite useful to help investors and speculators understand when big swings, major uplegs or corrections, were highly likely.
But during the past year, the HUI/Gold Ratio buys and sells as defined by this once-popular trading system are getting far more frequent. Instead of just marking major uplegs and corrections, they are flashing often during minor rallies and pullbacks. This is a big problem if you are still using this HGR system as it was originally intended, solely to catch the major swings. The more often it fires, the less likely its signals will be useful and profitable.
In order to get a quick visual overview of how the usefulness of HGR signals has been fading over the past year, these signals are superimposed on my next chart along with the actual HUI itself. This chart is interesting as it illustrates yet another limitation of ratio analysis. If you look at the chart below, you will notice that the HGR mirrors the HUI incredibly well. It is, in a very real sense, just a distorted projection of the HUI itself.
In any ratio analysis, the most volatile of the two variables tends to overpower the less volatile. Since the gold stocks are far more volatile than gold, their movements are more defining for the ratio than those of gold. With unequal volatility, there is never parity between the two variables in terms of their ultimate influence on the final ratio. As such, in many ways using the HGR to trade the HUI is like using the HUI to trade itself. Gold's role is minor due to its relative lack of volatility.
The bull market in the gold stocks began back in November 2000, so the first HUI/Gold Ratio sell signal of the bull occurred after this initial upleg in mid-2001. As you can see above, this was a good sell as the HUI consolidated flat to lower for the second half of 2001. As 2002 dawned the first buy signal triggered which was also excellent as the HUI was launching its second massive upleg which was the largest of this bull, carrying it up 145% in 127 trading days.
HGR sell signal 2 in mid-2002 was also good. It suggested getting out of the HUI just before the index crashed back down shortly after the sell triggered. This particular signal is also useful in illustrating the great danger in dawdling in declaring a signal occurring. If a trader had waited even a week or two to verify that the failure of the 50dma really occurred, odds are he would have been caught in the subsequent sharp HUI crash.
The second HGR buy signal in mid-2003 was also fairly solid. It did not flash at the very bottom of the consolidation but it did occur early on in the HUI's next massive upleg. Traders could have ridden the HUI all the way up in the second half of 2003 and then sold on the third HGR sell signal in late 2003. This sell would have taken capital out of harm's way while the HUI consolidated into the third buy in Q3 of 2004.
Interestingly the HUI's late 2004 upleg was a consolidation upleg, the smallest of this entire bull at a mere 45% run higher. Yet buy signal 3 still did a good job of announcing its beginning. Unfortunately though, the HGR's sell signal 4 in late 2004 happened after the HUI started correcting sharply so it didn't lock in as much of its preceding upleg's gains as the earlier HGR sell signals.
The next HGR buy in mid-2005 was once again excellent, just after the bottom of a major HUI correction. This would have got investors and speculators positioned in the HUI early on in what would grow to be the second largest upleg of this entire gold-stock bull to date. The HUI was starting to soar 137% higher in a massive upleg ultimately cresting in May 2006. It was one heck of a run!
This brings us through the first four sets of buy/sell signals of the HUI/Gold Ratio. While not terribly precise in terms of picking exact interim highs and lows, they did do an exceptional job of getting traders into the HUI early on in major uplegs and out of the HUI early on in major corrections. These initial four sets of buy and sell signals are what gave this HGR trading system its formidable reputation.
Unfortunately though, its efficacy is fading. Recall that this particular HGR trading system was explicitly created to capture major uplegs and corrections, to harness the big swings for traders. It was designed to be robust enough, even with its many subjective elements, to not be easily swayed into giving false or premature signals due to minor rallies or pullbacks. Yet over its most recent six signals, minor HUI moves have whipsawed it into flashing noisy signals.
The fifth HGR sell and buy signals happened within a month or so of each other, the sell triggering on a minor pullback and the buy triggering a little later once the sixth major upleg of this HUI bull to date resumed. Now the argument could be advanced that making this series of trades really doesn't do any serious harm since a trader following these signals would still be long the HUI for the rest of the upleg. But the sell was actually lower than the subsequent buy, so there is definitely some slippage of profits potential in this particular case.
The sixth set of HGR signals happened in much the same way in early 2006. The HUI started falling, rather sharply this time, and the HGR sell triggered suggesting a major correction was underway. But shortly after this signal tripped the HUI started rocketing higher again following the renewed vigor of gold. These signals occurred at similar levels in the HUI so there wasn't much slippage, but there were definitely transaction costs and probably unwarranted anxiety spawned by two signals so close together.
The seventh and final set of HGR signals occurred very recently. In early May when the HUI was within 10 points of topping a week or so later, the HGR sell signal triggered. This particular sell signal was excellent as the HUI would only travel a bit higher but then plunge 31% into June. So a trader would have been out near 385 on the HUI. The seventh buy just triggered in mid-August, around HUI 340 or so. This was much higher than the latest mid-June interim lows under 275 though, certainly not a great buy relative to these lows.
So the evolution of the HGR over the lifespan of this HUI bull has been quite interesting. Up until the middle of 2005 the HGR buy and sell signals generated using this trading system were meaningful and provided some good strategic timing cues. While they didn't catch the exact interim highs or lows, at least they triggered relatively early in major uplegs and corrections. Back then a HGR signal was a big deal since they only tended to happen a couple times a year at most.
But since mid-2005, the HGR signals have been noisier. The great volatility of the HUI relative to gold is causing thrashing, signals triggering midway through major uplegs and major corrections. Since none of the recent thrashing signals have been horrible, one could argue that the HGR still has value for traders. But on the other hand, it was originally designed to catch the big swings and now it is catching minor swings too.
Why is the frequency of HGR buy and sell signals changing? I am not sure, but there have been a couple potential factors on my mind. Technical trading systems often have a lifespan. It is quite normal for them to be very effective for a season and then to gradually lose effectiveness. This can be caused by more people knowing about the system and trading it, which reduces its edge. But it can also be caused by changing market conditions.
In summer 2005, about the time the HGR signals started getting noisier and thrashing a bit, gold entered Stage Two of its bull market. In Stage One it was driven primarily by dollar weakness, and hence its uplegs were generally orderly and modest. In Stage Two though it decouples from the dollar and is driven by global investment demand. Global investment demand is far more powerful and changes far more rapidly than the dollar bear, sparking a lot more volatility in gold. And since gold is the primary driver of the HUI, the leverage of gold stocks amplifies gold's volatility.
So it may just be that this particular HGR system was better suited to the relatively calm and orderly Stage One gold bull conditions than the far wilder and more unpredictable Stage Two now upon us. With the gold bull having an entirely different character today and the HUI mirroring those more volatile characteristics, perhaps the increased volatility will continue to exceed the parameters required by this HGR system to limit signals solely to major uplegs and corrections.
I think these observations about the HGR are also very valuable in a broader application. As investors and speculators, we have to realize that markets change and hence the efficacy and utility of our trading indicators change. Over time we shouldn't expect trading systems to work totally statically without adjustment for new market conditions. We have to expect them to age and fade, and hence keep looking for new indicators.
This also reminds me of some great Biblical wisdom out of Proverbs. "Where no counsel is, the people fall: but in the multitude of counsellors there is safety." Trading without indicators, with no counsel, is pretty silly, as it is like running blind and it encourages emotional trading that is the bane of successful investment and speculation. But at the same time, it is not a good idea to use just one indicator. Many indicators, a multitude of counselors, flesh out a more accurate picture of prevailing probabilities than any one in isolation.
At Zeal we attempt to follow this wisdom in our trading. We are always trying to study the markets from new perspectives, and sometimes this yields new indicators that have proven extremely profitable. But at the same time we realize that we need to follow many indicators, not just one. If the HGR gives a buy signal but an array of other indicators don't concur, then we have a problem. We only want to trade when probabilities are wildly in our favor, when many indicators line up pointing to the same likely near-term market outcome.
If you are interested in following the markets for commodities stocks including gold and silver stocks, and you want cutting-edge analysis on what a variety of indicators are suggesting is likely imminent, please subscribe to our acclaimed monthly newsletter. Not only are we constantly trying to better understand the markets and push the envelope in timing trades, but when the odds for success appear highly favorable we launch and recommend new stock and options trades as appropriate.
The bottom line is the HUI/Gold Ratio trading system that has been popular, and very effective, in recent years does have limitations. Like every trading system it is built on assumptions that add layers of subjectivity that can never be squeezed out. As such, neither it nor any other indicator should be used in isolation for trading. The best odds for success exist when a multitude of indicators agree on a particular likely path ahead.
Therefore prudent investors and speculators will take a signal from any one system with a grain of salt. If only that system is signaling an entry or exit, it should be carefully investigated before capital is committed. This applies directly to the recent HGR buy signal we saw in mid-August as well. Be careful here.
Adam Hamilton, CPA
September 8, 2006