Elliott Wave analysis for constant-dollar gold 1980 - 2001

October 30, 2005

I just think with gold and silver that are both political metals, they are manipulated, so Elliott Wave analysis is null and void for these metals (manipulation does not make for a good count anywhere). - David Petch, April 1, 2005

If traditional Elliott Wave analysis is worth anything (and that's a big if in a rigged market), - Nelson Hultberg, May 2, 2005

This essay addresses the concern quoted above from two highly respected and clear-headed analysts, by demonstrating that - irrespective of whether the market is manipulated or not, which is an issue completely beyond the scope of this work - a strikingly neat Elliott Wave count can be assigned to the action in the price of gold in constant dollars from 1980 to 2001 which not only is valid according to Elliott rules but also is clearly the preferred interpretation, and has extremely bullish implications for the future.

As a fair warning, this work is of strict technical nature, may feel quite hard and dry for anyone except hardcore Elliotticians, and is strictly based on Elliott Wave theory as taught by Robert Prechter in The Elliott Wave Principle (EWP, "the book"). Its conclusions, however, are completely opposite to the views held lately by Prechter.

In the field of Technical Analysis, within which the Elliott Wave Principle (WP, "the theory") is perhaps the most comprehensive approach, the key for forecasting the future evolution of the price of a security is to look back at its past action. In WP, moreover, it is crucial to take into account not just the recent past, but also - and first of all - a longer timeframe, in order to identify market junctures and - more importantly - anticipate market moves. Quoting Prechter (The Elliott Wave Theorist (EWT), 1982 Sep. 13):

"The task of wave analysis often requires stepping back and taking a look at the big picture and using the evidence of the historical patterns to judge the onset of a major change in trend. Cycle and Supercycle waves move in wide price bands and truly are the most important structures to take into account." (emphasis added)

It should be noted that the context for this statement was Prechter's accurate prediction of the end of the bear market for stocks (cycle wave IV in his count) in August 1982 and the start of a new, long-lasting bull market.

Specifically about the end point of that cycle wave IV (in Prechter's count), in a recent comprehensive review of WP practitioners' past performance at identifying major market junctures (EWT, 2004 Dec. 10), Prechter left open two possibilities: December 74 using the nominal Dow or August 1982 using the constant-dollar Dow (deflacted using the PPI). But in the same issue of EWT he gave an important clue for choosing one over the other: the DJIA dividend yield, for which the termination of waves of higher degree have produced more extreme figures. Since the dividend yield was noticeably higher in 1982 than in 1974, it lends considerable support to the notion that 1982 (i.e. the inflation-adjusted low) was the actual bottom of that corrective wave.

But more importantly, when considering the whole stock market history since 1790 and adjusting the Dow for inflation, there are some significant changes to the wave count, which are discussed in detail in the page http://yelnick.typepad.com/yelnick/2004/03/constantdollar_.html

The conclusion of that discussion is that "the inflation-adjusted Dow better matches the real economy. It also better matches Elliott Wave theory, giving us a clearer picture of the wave count, especially after a period of severe deflation (as in the '30s) or severe inflation (as in the '70s) when the actual Dow is adrift from real value. This adjusted picture has profound implications for the current wave count." (Specifically for the broad stock market, what might have topped in 2000 or more likely will top soon is not cycle wave V of a supercycle wave that started in 1932, but cycle wave III of a supercycle wave that started in 1949.)

A particular security for which it is most relevant and timely to apply this conclusion is gold. Using nominal prices, since November 2004 Prechter has been interpreting the price action since January 1980 as a yet unfinished complex corrective pattern that calls for a deep decline to historical new lows, first from the 454 top on December 2, 2004, and now from the 480 top on October 12, 2005. However, if we use constant-dollar prices, a very different picture emerges.

For that purpose, we will bring all past prices (i.e. inflate them) to a common date, for which we chose Dec. 01, 2004. It is evident that any date could be chosen as common date without affecting this analysis. For the date chosen, the value of the inflation indices are:

CPI 2004-12-01 190.3

PPI 2004-12-01 150.8

Doing that, we will explore the possibility that the price action from January 1980 to April 2001 is a complete motive wave, i.e. that it subdivides into five waves. Since it lasted 21 years, this prospective wave should be of at least Cycle degree. Therefore, we will tentatively assign Cycle degree to the whole wave and call it C. Its five subdivisions, then, would be of Primary degree and denoted as ((n)).

Before delving into price numbers, it is important to notice that, from the viewpoint of Fibonacci time relationships, the case for a five-wave motive pattern could not be better: the three actionary waves last about 5 years, the two reactionary waves last about 3 years, and the whole Cycle wave lasts 21 years. 5 and 3 are Fibonacci numbers, and so is 21.

Regarding wave formation guidelines, the alternation rule is satisfied since wave ((2)) is quite sharp while wave ((4)) is fairly sideways.

Using constant-dollar prices, the price action satisfies the rules for a motive wave, with wave ((3)) travelling further down than wave ((1)), and with wave ((3)) being, in percentage terms, greater than wave ((5)).

That motive wave, however, is not an impulse because Primary wave ((4)) overlaps wave ((1)). Therefore, the Cycle degree motive wave must be a diagonal triangle, either a leading diagonal or an ending diagonal. This is a very important point, and we'll get back to it later.

Since waves ((3)) and ((5)) are almost equal, it is interesting to check their relationship to ((1)), the biggest primary wave. The results are striking.

Relationship between waves ((1)) and ((5)), inflating using CPI:

46 / 75.2 = 0.612

Relationship between waves ((1)) and ((3)), inflating using PPI:

44.8 / 72.8 = 0.615

Summarizing, the interpretation of the price action from January 1980 to April 2001 as a motive wave is supported by the following facts:

  • fulfillment of rules for motive waves
  • perfect Fibonacci time relationships
  • perfect Fibo amplitude relationships
  • fulfillment of alternation guideline.

Furthermore, it is hard not to arrive at the conclusion that such a confluence of facts makes this interpretation not just valid, but also preferred.

The next crucial step is to check whether this is actually a complete motive wave, because it could potentially be still developing, in which case:

((3)) above would actually be (1) of ((3)): 432.15 393.41

((4)) above would actually be (2) of ((3)): 509.60 482.28

((5)) above would actually be (3) of ((3)): 275.34 271.62

and the December 2, 2004 high (4) of ((3)): 454.20 454.20

The numbers above do not support the case for a still-developing motive wave because of the observation that the price action from April 2001 to December 2004 is a motive wave, which implies that it cannot be a complete correction itself, and, if it is the first part of a correction, that correction has to be a zigzag, so we should expect a higher high for wave (4) of ((3)). Since wave (4) of ((3)) has already (at the time this essay was originally written: July 2005) retraced 76.3 % (CPI) or 86.7 % (PPI) of wave (3) of ((3)), if it still had to make a higher high, it would eventually retrace so much of wave (3) of ((3)) as to make this interpretation highly implausible to say the least.

Consequently, the price action from January 1980 to April 2001 is indeed a complete motive wave.

At this point, and for the sake of the thouroughness of the analysis, we will solve a couple of minor issues.

The first minor issue is that primary wave ((1)) is not an impulse if we use nominal prices, because wave (4) overlaps wave (1). But since part of wave ((1)) took place during a period of high inflation, we will inflate its prices as we did with the Cycle degree wave and see what it looks like then.

Here we see that, after inflating with both CPI and PPI, wave ((1)) becomes a clean impulse, i.e. (4) does not overlap (1).

The second minor issue is that the double bottom on July and August 1999 was, in nominal terms, lower than the postulated bottom in April 2001. This issue can be resolved in two ways. One is to again inflate the prices and see what happens.

Here we see that the 2001 bottom was lower, in constant-dollar prices, than those in 1999.

The alternate way of solving this issue is by simply interpreting the ((5)) wave as ending in a truncation.

Either way, the orthodox bottom was in April 2001, thus allowing for the Fibonacci time relationships mentioned above.

Additionally, for this issue the truncation interpretation is preferred because the period involved had a low inflation value (142.1 / 133.3 = 6.6% per PPI, 5.9% per CPI), substantially lower than inflation during the period involved in the issue considered before (101.0 / 86.2 = 17% per PPI, 21% per CPI). This interpretation is supported by the observation in Prechter's EWP that "truncated fifths ... imply the same thing: dramatic reversal ahead." And that's exactly what happened at Primary degree since April 2001, with the subsequent bullish Primary wave having ended in December 2004.

Now we can come back to the important point left pending above, namely that the 1980-2001 complete bearish motive wave of Cycle degree must be a diagonal triangle, either a leading diagonal or an ending diagonal.

Quoting from Prechter's EWP, "the main key to recognizing a leading diagonal is the decided slowing of price change in the fifth subwave relative to the third." In this case, given that the time length of waves ((3)) and ((5)) is almost the same, all we need to do is to compare the waves' amplitudes. And since the amplitudes are almost equal, as we have shown above, there is obviously no "decided slowing of price change." (Actually, as can easily be checked in the chart http://www.sharelynx.com/chartsfixed/SpotAUlog.gif for its most part (Feb 1996 to Aug 1999), wave ((5)) had an acceleration of price change relative to wave ((3)).)

Therefore, the only remaining possibility is that the complete pattern should be an ending diagonal. But now we are faced with an apparently unsurmountable obstacle: all actionary subwaves in an ending diagonal subdivide into a "three", whereas we have previously shown that primary subwave ((1)) is a clean impulse. (On the other hand, primary subwaves ((3)) and ((5)) do support this interpretation, as they can perfectly - and actually more easily - be interpreted as "threes", specifically zigzags. That can easily be verified by looking at the chart above.)

At this point, it is important to remember why we are interpreting the whole Cycle wave as an ending diagonal in the first place:

  • First and foremost because subwave ((4)) overlapped subwave ((1)), which in turn was because either subwave ((3)) travelled down too little, or subwave ((4)) moved up too far, or both.
  • Secondly because of the internal structure of subwaves ((3)) and ((5)).

In other words, the whole wave should be an ending diagonal because of the mood of investors from subwave ((3)) onward. While the obstacle for interpreting it as an ending diagonal is the internal structure of subwave ((1)). In other words, the whole wave should not be an ending diagonal because of the mood of investors during subwave ((1)).

The key for reconciling this discrepancy is simply to consider the time scale involved: subwave ((1)) ended in February 1985 while subwave ((3)) started in December 1987. And a three-year's interval looks long enough for investors to substantially change their mood from impulse-like to ending diagonal-like.

Thus, we are postulating that the Cycle wave started as an impulse (according to the internal subdivisions of subwave ((1))) and then "mutated on the fly" to become an ending diagonal (according to the amplitudes and internal subdivisions of subwave ((3)) and following). While that kind of dynamic transformation of a wave's nature during its course of development is not explicitely contemplated in the current state of WP, we hold that this concept remains true to the basic tenets of WP and is a perfectly valid possibility if - and only if - the wave in question is of a degree high enough that sufficient time can elapse between subwaves so as to allow investors to substantially change their mood. (Besides, the transformation itself has to make sense. A terminal motive wave that starts as an impulse and ends as an ending diagonal does make sense. The inverse transformation does not.)

This interpretation of the Cycle degree motive wave from January 1980 to April 2001 as an ending diagonal has an important consequence according to Prechter's EWP. Quoting now in its entirety a paragraph partially quoted before: "Fifth wave extensions, truncated fifths and ending diagonal triangles all imply the same thing: dramatic reversal ahead. At some turning points, two of these phenomena have occurred together at different degrees, compounding the violence of the next move in the opposite direction."

Thus, the fact that April 2, 2001 marked the end of both

  • a truncated fifth wave of Primary degree, and
  • an ending diagonal triangle of Cycle degree,

sets the background for the violence of the subsequent bullish move of Cycle degree. (Previously we had arrived at a similar conclusion but just for Primary degree.)

To sum up:

First, we showed that price actions spanning long and/or high-inflation periods are best analyzed using constant-dollar prices.

Secondly, we applied the above criterion to the price action of gold since January 1980 to April 2001 and showed that its interpretation as a complete (al least Cycle degree) motive wave not only was valid but also should be the preferred interpretation because of the strikingly perfect Fibonacci relationships - in time and amplitude - between its subwaves. So far, that implies that the price action from April 2001 onward should be a bullish wave of the same Cycle degree, which in principle could be motive or corrective.

Third, we substantiated the case for that 1980 - 2001 Cycle degree motive wave to be specifically an ending diagonal. That implies that April 2001 marked the end of not only that Cycle degree wave, but also of a larger bearish wave of the next higher degree (i.e. at least Supercycle). Which in turn implies that what started in April 2001 is a bullish wave of at least Supercycle degree, which again in principle could be motive or corrective.

In order to figure out the exact nature of the Supercycle degree bearish wave that ended in April 2001, the fact that the 1980 - 2001 Cycle degree bearish wave was an ending motive wave leaves open the following three possibilities for the larger pattern: impulse, zigzag or flat.

In order to sort out these possibilities, we should take a look at an even bigger picture of past gold price action. Again, and with even more reason, the analysis should be done in constant-dollar terms. Doing that, a very basic and important fact shows up, namely that the January 1980 top was higher, in constant-dollar terms, than any real gold price since the XVI century. That excludes the possibilities of interpreting the Supercycle degree ending bearish pattern as an impulse or a zigzag.

Therefore, the 1980 - 2001 bearish ending motive wave was effectively wave C of an expanded or running flat of at least Supercycle degree that ended in April 2001. For this flat:

  • downward wave A ended in January 1970 and started no later than 1932-1934. If it did start in 1932-1934, then the bearish flat was definitely of Supercycle degree, consistently with the timeframe for the last bullish Supercycle in stocks (1932 - 2000 according to Prechter, 1949 - 20?? in constant-dollar count).
  • upward wave B ended in January 1980. In line with the flat interpretation, wave B was an unmistakable zigzag, whose subwave ((B)) Apr 1974 - Aug 1976 was in turn an expanded flat, as counted by Prechter in EWP.

Measured in constant dollars, this Supercycle degree flat was running, because point A in January 1970, at nominal USD 35, was real USD 175 inflating by CPI or 135 by PPI, in both cases much lower than point C in April 2001 (real USD 275 or 271 respectively).

This last conclusion is very interesting because, quoting from Prechter's EWP, in the case of running flats, "the forces in the direction of the larger trend are so powerful that the pattern becomes skewed in that direction."

The direction of the Supercycle degree running flat that ended in April 2001 was down. The direction of the larger trend that started then, in which "the forces are so powerful", is up.

In constant dollars.

The purity of gold is measured in carat weight.