Uploaded by lahcen.ezzerhouni

Fractal Market Mastery: Trading with Fractals

advertisement
Fractal Market Mastery
Andrew McElroy
Published by Andrew McElroy, 2023.
While every precaution has been taken in the preparation of this book, the
publisher assumes no responsibility for errors or omissions, or for damages
resulting from the use of the information contained herein.
FRACTAL MARKET MASTERY
First edition. January 28, 2023.
Copyright © 2023 Andrew McElroy.
Written by Andrew McElroy.
Preface
Fractals are all around us and prevalent in financial markets. The same patterns
are repeated over and over as humans repeat the same behaviours.
What if you could anticipate how and when this will happen? Then you’d have a
fractal ‘map’ to future price action, and when you have this, you have a powerful
trading tool.
‘Fractal market mastery’ is a practical guide to trading with fractals. Written by
a trader for traders, it covers all aspects of using fractals and profiting from
them. There is no ‘holy grail’ and I can’t promise secret techniques to make
millions. I can, however, guarantee an enhanced understanding of price action,
trends, market behaviour and even some completely new theories which have
never been seen before.
I first heard of fractals when I happened upon the late, great Ed Matts on Twitter
in 2014. I found the concept intriguing, but I didn’t quite grasp its power until I
had a ‘light bulb moment’ several months later. It came while I was preparing to
trade the S&P500 and a feeling of déjà vu came over me. Had I made a similar
trade somewhere before? I looked to the left of the chart and at previous sessions
and there it was – the same pattern I was about to trade. To my alarm it
suggested my trade would soon lead to a loss, so I waited and watched, and I
was amazed as the market repeated the same behaviour all over again.
Such was my interest in fractals, Ed Matts became my unofficial mentor and I
ended up working with him at Matrixtrade in 2017. Around this time, I also
wrote prolifically for Seeking Alpha and published over 300 articles. Many of
the charts in this book come from these articles as they provide evidence of how
fractals were used in real-time to make actual trades and calls. I also link to
Twitter calls if they support the veracity of a theory.
Most of the content in this book is aimed at practical use. Having read some of
Benoit Mandelbrot’s work and struggling through ‘The Misbehaviour of
Markets,’ I questioned what I could practically do with all this new information
and how I could fit it in with my methods of trading. This book shouldn’t pose
such questions. It does touch on some theory, but only to further understanding
of the logic behind fractals, their formation and their uses.
Repeating behaviour and price action is a consistent theme throughout, but not
all chapters are about fractals in their strictest sense and the methods are
expanded into all areas where they have practical uses. The aim is for all the
concepts to be useful to traders and most important of all, to make money. There
is no ‘holy grail’ and I can’t promise secret techniques to make millions. I can,
however, guarantee an enhanced understanding of price action, trends, market
behaviour and even some completely new theories which have never been seen
before.
Contents
1. Introduction to Fractals
2. Fractal Formation
3. Market Fractals
Self-Similar Fractals
Finding Fractals
Fractals in Trend – A Dow Jones Industrial Average Case Study
Fractals in Triangles
4. Elliott Wave and Fractals
A Symbiotic Relationship
A Different Perspective
5. Other Types of Fractals
Analogues
The Bubble Template
Event Fractals
Black Swan Events
6. When Fractals Fail
About the Author
1. Introduction to Fractals
Afractal is a mathematical set that exhibits a repeating pattern displayed at every
scale.
We see fractals constantly, particularly in nature. Ferns repeat the same pattern
through smaller and smaller fronds.
Source: Wikimedia Commons
Dandelions have amazing fractal symmetry, as do most trees. In the fractal tree
coded below, two branches are coded from any given branch until they get so
small, they can’t be seen.
Source: rossettacode.org
When translated into the markets, fractals look like this:
This isn’t a perfect repetition, just as a real-life tree is not as perfect as the coded
fractal. It is important to never expect perfection in anything with so many
moving parts. In the same way that a dandelion pappus may blow away from the
head, or a tree branch may snap, market patterns can be ‘blown’ around by a
number of external factors (news, large orders, lack of liquidity etc...).
Nevertheless, we can still identify a repeating structure and it can be useful to
think about the broader character of the fractal.
In the oil (CL) example above, the first failed break out above $50 led to a
reversal lower and a decline to $26 where a ‘V’ bottom reversal led to a rally and
a new high above $50 again. Oil then repeated this behaviour in a smaller
iteration and it could be speculated it would do the same again (the arrows on the
chart). This is where the magic (and profit) happens – we have a potential ‘map’
based on the previous pattern and we can base trades around it. This article on
Seeking Alpha used the fractal to predict a rally and advised to get long. Oil
soon reversed at $42.95 and followed the arrows higher, reaching $56.18 by the
end of the year.
2. Fractal Formation
When committing money to any method of trading, it is prudent to question why
it should work. What is the logic? What is truly going on?
Fractals form in nature as the most useful and efficient structure is repeated.
Human lungs and trees have remarkably similar anatomies as they both aim to
maximise surface area for gas absorption. Furthermore, trees manage to do this
without compromising strength and stability.
Leonardo da Vinci recognized nature's ability to from efficient structures 500
years ago when he wrote "all the branches of a tree at every stage of its height
when put together are equal in thickness to the trunk." This suggests if the tree
were to have its branches gathered and compressed, it would have the same
thickness from top to bottom. The rule was tested in 2011 by physicist
Christophe Eloy, from the University of Provence in France who used computer
simulations to model trees with intricate fractal structures. A virtual wind tunnel
put each branch under pressure and found the proportions of the tree and the
diameters of the branches were exactly the right size to limit snapping. The
proportions between different sizes of twigs, branches and trunks were always
the same regardless of wind speed or the height of the tree, which was exactly
what da Vinci’s rule suggested.
A selection of Professor Eloy’s work. Source: Leonardo's Rule, Self-Similarity,
and Wind-Induced Stresses in Trees
The way fractals form in markets can be thought of in similar terms to nature.
Every trade is made at a price where the largest number of buyers and sellers
agree to meet. The market constantly moves in its search for this most efficient
price, and in doing so, creates patterns on the chart. This is a natural process and
each pattern is formed by the reactions of humans (and algos programmed by
humans) to external factors.
We can assume each pattern was beneficial/successful for some of the
participants. Given the opportunity, those traders would repeat the same
behaviours again. In other words, if traders made money by doing certain things
at certain times, they will be inclined to repeat the same behaviour. Moreover,
decisions are driven by similar conditions and emotions including greed, fear,
complacency, and indecision. Not all are judgements are carefully considered;
rather they are triggered, often subconsciously. At times you can see the
repeating behaviour and the causes of it all over the charts.
The Global X MLP ETF (MLPA) chart below illustrates this point as moments
of panic and fear create the same repeating pattern: price accelerates lower into a
‘V’ bottom and then partially recovers in a slower rally to lower highs.
3. Market Fractals
Self-Similar Fractals
Aself-similar market fractal in the S&P500 (SPY) is shown below.
It is a similar pattern to the previously shown oil repetition but is on a 1-minute
intra-day chart. Fractals can occur in any timeframe and any instrument. They
can also repeat any kind of pattern. Below is a weekly fractal in Unicredit
(UCG).
All these fractals form first in a higher timeframe pattern and then repeat in
smaller iterations. A pattern can also start small and then repeat in larger
patterns, but such behaviour is unusual. This is logical as larger patterns are
more important than smaller ones and exert control over them. Market
participants who create patterns in the higher timeframes are the largest
institutions and their actions are seen in every chart. However, the opposite is not
true - even if a small retail trader were able to briefly shape a market on a 1minute chart, they would never be able to replicate the same pattern on a weekly
chart.
The large dominant templates get smaller as they repeat as they lose power or
efficiency. It’s comparable to an echo getting quieter on each repetition until no
longer audible. At times the pattern repeats multiple times and gets so small it
can only be seen on intra-day charts. A few more iterations and it will disappear
from view completely. This doesn’t happen often, however, as patterns usually
fade and are replaced by a new dominant market phase and pattern long before
they get a chance to get so small. Even in nature, fractal patterns are usually
substituted by new structures after a few smaller iterations. It cannot be assumed
that a pattern keeps repeating indefinitely as the more times it repeats, the
weaker it becomes. Furthermore, the more times a pattern repeats, the more
noticeable it will become and when anything is obvious and easily seen by the
crowd, it is prone to failure.
Ideally, a fractal is spotted early or mid-way in the first repeating stage. i.e. a
new pattern is created on a higher timeframe and the potential for a repeat is
identified as soon as the early stage of the next pattern looks similar. This allows
some trades to be taken before the crowd has seen it. Of course, the risk is that
the pattern is not actually repeating and some other pattern is eventually formed,
but this is true of any fractal trade and every trade based on technical patterns.
Assuming it does repeat successfully, the second iteration can also be traded, but
may abort before completion. If a third iteration starts to form, be careful.
Gold (XAUUSD) formed a large consolidation and rally pattern from 19802011. This was repeated from 2011-2020 and repeated again in 2020-2022. A
third iteration stated in 2022 but failed towards the latter stages.
The gold chart shows repetitions of the same pattern, but the smaller ones cannot
be shown with much detail. Often it is better to split the iterations onto separate
planes on the chart so they can be viewed and compared side by side on similar
scales.
In the chart below, the top pane is the template; the highest timeframe weekly
chart showing over 20 years of price action. The middle pane shows the 20112020 pattern on a 2-day chart, and lastly the 12 hourly chart shows the smallest
pattern. Each pane may be on a completely different timescale, but they contain
a similar amount of detail and bars which helps the comparison.
This side-by-side comparison can be a powerful trading tool. If, for example, the
repeating pattern was identified in early 2021, the chart below could be
constructed and act as a guide for future trades.
It would suggest longs could be taken below $1700 for a move back to the highs
around $2000. It would also warn there could be a long consolidation at the lows
before the rally gets going.
Gilead (GILD) provides another example. The reaction to Q2 2016 earnings was
very similar to the reaction following Q1 earnings, but on a smaller scale.
While the repeat is evident, it is more useful to use a side-by-side view and to
use a lower timeframe on the smaller repetition so the scales and number of bars
are similar. The same comparison is made below with a 4hr chart on the top and
a 2hr chart on the bottom.
The two charts can also be superimposed–
Source: Prorealtime
An accurately placed comparison can reveal a number of trades. The above
example was used for a long recommendation from $76.8 to $82 and followed
up later.
Finding Fractals
Hours spent trawling through charts may throw up some similar patterns, but it is
unlikely to yield a reliable result. Similarly, computer programs which search
massive amounts of chart patterns will eventually find something resembling a
match. But is it really a fractal? Machines may have the edge on speed and
computing, but they are no match for a human with feeling and intuition.
Perhaps if you have to search really hard, it isn't really there in the first place.
So how are fractals found? The best ones are usually patterns remembered from
a previous occurrence and provoke a feeling of familiarity, even déjà vu. When
experiencing today’s price action, you may find yourself saying, ‘I've done / seen
/ felt / traded this before'.
Often the trigger comes from inconspicuous action. For example, the S&P500
broke to new lows in late September 2022 while the Dax held a higher low. This
divergence seemed familiar and a quick look to the left of the chart confirmed
the same thing happened around a month earlier. Furthermore, the patterns
happened to be nearly identical. Comparing the two occurrences provided a
guide to the bounce and an eventual reversal lower again.
Source: Tradingview
This ‘feeling’ of likeness or familiarity is one of the reasons fractals form and
repeat. It provides comfort, and in an environment that often provokes painful
losses, comfort is something traders seek, either consciously or subconsciously.
We, as fractal traders, may be conscious of the repeating action, but most
participants are not.
Were S&P500 traders aware they were repeating similar actions to the previous
week in April 2022?
Source: Tradingview
Probably not – they were just reacting to market stimuli as they saw fit at the
time – but the successes (or otherwise) of the previous week likely influenced
decisions. Those who recognized the repeating pattern could well have profited
from it.
Of course, traders cannot simply wait and hope for a feeling of déjà vu to inspire
them. They can, however, regularly ask, ‘what is the market doing and when
might it have done something similar?’ This is a logical way of finding fractals
rather than a random search and will provide improved results when trading
them.
A basic knowledge of Elliott Wave also helps location as it breaks down the
trend and corrections into specified parts. If we are trading a wave 5 currently,
then it makes sense to look at previous wave 5s (or wave 1s) in different
timeframes. Chapter 4 will deal with Elliott Wave in more detail.
For many traders, events are easier to categorize than price action and easier to
remember. When, say, North Korea fires a missile for the second or third time, it
is logical to ask, ‘what did the market do the last time this happened?’ Naturally,
the context could be slightly different each time, but since price action is a
reflection of context, if it is similar leading up to the missile launch, then there
are good odds it will continue the likeness. Bear in mind, though, reactions and
patterns will likely get smaller (as fractals do) each time an event happens as it is
less of a surprise and more traders will be prepared for the outcome.
When Japan entered a bear market in 2015, the media were talking about an
impending crisis in Asia. Anyone with a knowledge of market history will have
known there was a crisis in Asia in the 1990s, so it was fairly obvious where to
look for a guide. The chart below was recognized in early July 2016 and was the
basis of this article calling for a 30% Nikkei rally.
Source: Prorealtime
The rally exceeded expectations and went on to gain around +60% over the next
18 months:
When oil crashed in early 2016, history not only provided a guide to where it
would bottom, but also how it would reverse and rally. The below tweet went out
on February 11th 2016, the day the 77% crash bottomed.
Once oil started moving higher from the 11th February low, it continued to
follow the 2009-2010 template with a strong initial recovery followed by a
sideways consolidation. The comparison shown below formed the basis of this
call in January 2017 for a pullback into the $40s followed by a large rally into
the $60s.
Source: Prorealtime
The two oil crashes were a good match, but it is also possible to compare a crash
in one market to a crash in another. After all, it can be assumed the sellers in
both markets are fearing for their capital and are behaving in similar ways. The
same can be said about bubbles and other patterns, or indeed events. Specific
events and analogues in chapter 5.
Fractals in Trend – A Dow Jones Industrial Average
Case Study
The Dow Jones Industrial Average (DJI) has a fascinating long-term chart and its
fractals make it a good candidate for a case study and long-term projections.
These are more of an intellectual pursuit than a useful trading guide. If a mistake
is made on this scale, it can be 100s of percent out. Even so, most traders have
questioned the fate of the long-term trend in stock markets and this question is
best answered through fractals and the framework of Elliott Wave.
Here is the monthly chart of the entire DJI trend going back to 1896.
It may not look all that interesting, but its general shape is a classic example of
trend and is repeated through many timeframes and sections. The first instance
of this seems to come from the ‘roaring twenties’ into the 1929 top and this set
the template for most of the trends since.
Source: Tradingview
These trends show the same distinct stages - acceleration and steepening in the
central area transitions into a recognizable consolidation pattern near the highs
(in the rectangle) which then signals one final rally into a top.
Being able to recognize the DJI trend fractal has a number of advantages as it
not only repeats in the higher timeframe charts, but in the lower ones too. A 15minute chart from 2022 with the same distinct phases is shown below and
anyone trading it should have known when to sell at the top.
Once the sequence ends, the subsequent corrective drop also tends to follow a set
pattern. In fact, 4 out of the 6 patterns in the first comparison have similar
corrections after the top. Only the 1929 crash is clearly different while the 19322022 chart is very long-term and is yet to correct. Furthermore, it can be
speculated that if the 1932-2022 trend does correct significantly, the path could
follow a similar pattern to the others. The chart below shows a closer view of
one of these comparisons.
Source: Tradingview
This suggests a return to the 2009 lows over the course of around 20 years. It
sounds fantastical (and it isn’t a trade recommendation), but at some point, the
market is likely to correct in proportion to the 1929-32 correction. If and when it
does, it should be in a different way to the sharp crash of 1929-1932 (the Elliott
Wave rule of alternation) so a sideways to down drift over many years is the
most likely.
The projection above would retrace the 1932-2022 wave 3 by 23.6%; the most
common Fibonacci retrace for a wave 4 after an extended wave 3. It is also
similar to the path projected by the comparison below which utilizes the
correction following the 1942-1973 trend.
Using the same principals, the 1987-2002 chart (or even 1942-1973) can be
matched with the recent 2009-2022 bull market and guide the correction from
the 2022 top. Zooming in to the price action from 2000-2003 highlights how the
corrections have the same sideways-to-down character and provides a more
detailed visual of the projection.
This comparison has a lot of supporting evidence, from the US dollar also
matching 2000-2002, to the interactions of growth and value stocks during both
periods.
Source: Tradingview
Fractals in Triangles
Triangles can be one of the most rewarding patterns to trade as they provide
defined risk / reward in and can be traded in both directions with tight stops. The
first challenge, however, is to anticipate the pattern forming in its early stages.
Again, Elliott Wave helps: triangles only form in the wave 4 position of a trend
or as a linking pattern in a correctional phase.
Source: author’s drawing using Tradingview
Fractal clues are another way to identify and trade triangles from the early stages
and eventually in the direction of the break. Each leg inside the contracting
pattern is related to the others and the same traders are likely involved in their
creation. This often means the downswings are similar in structure, as are the
upswings. In the chart above, this would mean waves A, C and E are fractals,
with waves B and D also likely related.
The Tesla (TSLA) triangle which developed throughout 2017 is a good example
of this.
Source: Tradingview
Zooming in to the three declines within the triangle (waves A, C and E) show
they have similar characteristics.
Source: Tradingview
The structure of the rallies in the triangle (waves B and D) were very different to
the declines but were comparable to each other.
Source: Tradingview
Note the timeframes are getting shorter. The pattern is contracting and getting
smaller as the legs repeat until a new dominant pattern takes over, i.e., the trend
continues.
Identifying these repeating legs helps us in a number of ways. Firstly, it confirms
suspicions of a triangle in the early stages. If wave C starts looking like wave A,
then most likely a triangle will form and each leg can be traded. Secondly, it
helps with trading each leg as we have a visual guide as to how it should unfold
and when it is likely to complete and reverse. Lastly, and most importantly, it
tells us when the triangle is likely to finish and break-out into a new phase of the
trend.
4. Elliott Wave and Fractals
A Symbiotic Relationship
Elliott wave relies on the fractal nature of markets. In fact, the very first fractal
discovered in 1872 by the mathematician Karl Weierstrass looks very much like
an Elliott Wave pattern and has the same qualities.
Source: Wikimedia
The ‘Weierstrass function’ is continuous everywhere but differentiable nowhere.
In other words, it has detail at every level so zooming in on a piece of the curve
does not show it getting progressively closer and closer to a straight line. Rather,
between any two points, no matter how close, the function will not be monotone.
The chart below from Elliott’s original 1940 essay, ‘The Basis of the Wave
Principle,’ is remarkably similar.
A basic 5 leg trend sequence has fractal qualities: zooming in to one of the
individual waves should reveal the same basic 5 wave structure. In theory, the 5
wave structure should be present from the smallest timeframe right up to the
highest.
It is by no means necessary to know Elliott Wave to use or recognize fractals.
However, some concepts help in locating fractals and guide expectations. For
instance, an Elliott Wave sequence often contains wave equality so that waves 1
and 5 are comparable in size and wave 3 is usually extended in length.
Furthermore, waves 1 and 5 are often similar in structure. If you think a wave 5
is underway, the previous wave 1 is the obvious place to look for a guide.
Additionally, due to the fractal nature of the waves, zooming in to look at the
smaller waves inside waves 1 or 3, or even zooming out to the larger structure
can also reveal guides. This may sound complicated, especially for those without
knowledge of Elliott Wave, but the concepts can be approached from a different
perspective which not only simplifies the methods, but reveals additional
insights too.
A Different Perspective
Elliott Wave can be difficult to master and reliably identifying the waves is a
challenge. One of the problems is that trend sequences come in so many shapes
and sizes they rarely follow a textbook example. Occasionally, though, a perfect
trend is formed, and it reveals a number of important aspects.
Trends build (in wave 1) and then unwind (in wave 5), with the same
participants creating similar patterns in both phases. Often this happens around
the central part of the trend (wave iii of 3’s mid-way gap) which provides a
marker and can help significantly in navigating the trend and in anticipating
patterns. All of this can be thought of in non-Elliott Wave terms as when a clean
trend unfolds, labelling individual stages is easy and logical.
Netflix (NFLX) provides a good example and the below chart was tweeted on
May 2nd 2022, 10 days before the major bottom.
Source: Tradingview
Positions built in the top section – the initial drop of wave 1. This was followed
by a huge gap down in wave 3 and it could be speculated the mid-way point of
this gap was also the mid-way point of the entire trend sequence. Positions were
then unwound at the lows and this created a similar pattern to the to the one in
the initial stages.
The Netflix chart has no Elliott Wave labels, and they were not needed as each
stage of the trend was clear and it could be speculated the decline was very
nearly complete. Indeed, a price target for the low could have been formulated
on the day of the large gap down; using the price of the central point in the gap
of $290, and the price of the start of the trend of $396, assuming $290 was
indeed the middle of the trend gave a target of $185. i.e., 290 - (396-290) = 184.
This did not play out perfectly, as although there was a bounce at $185, it
reversed for one further low and a capitulation to $162. Even so, anyone with the
above chart could anticipate the trend completing and be prepared to buy for a
large recovery.
While stocks and ETFs often gap down in the central part of the trend, this is not
a requirement and futures markets contain no gaps throughout the week.
However, acceleration / steepening in the trend can still be identified with a
central point as shown below.
Source: Tradingview
As well as containing a clear central acceleration point, the Dax chart above
shows waves 1 and 5 are equal in price and in time (the two orange boxes are
identical). However, the structure of waves 1 and 5 are quite different, at least
when viewed linearly. The problem is that initiating positions and closing them
are opposite actions and can have opposite effects on structure. Taking this a step
further, if we consider everything before the central point as positions being
initiated, and everything after as the opposite action, unwinding, the structure
should be opposite too, like a mirror image.
This theory can be tested by manipulating the chart. If the first half of the trend
up to the is copied and rotated to create a mirror image based around the central
point, it looks almost the exact same as the original Dax chart.
Source: Tradingview and MS Paint
The fact that a trend can unfold as a mirror image around a central point adds to
some of the concepts of Elliott Wave. Elliotticians call the central point the
‘point of recognition’ where it finally becomes obvious a trend sequence is
playing out. Somewhat ironically, it is this recognition then flips the pattern into
the opposite polarity and a mirror image.
The Nvidia (NVDA) downtrend in 2022 again illustrates this point and the
symmetry can be seen clearly.
Source: Tradingview
If the chart is manipulated to create a perfect mirror image by copying and
rotating the top section, it remains comparable to the original chart.
Source: Tradingview and MS Paint
This price action makes sense when we think of smart money and their
counterparts, dumb money. Smart money is initiating positions in the first half of
the trend and accumulating at the expense of dumb money. Later, when the trend
finally gets going and accelerates through the central point, the point of
recognition, dumb money realize their mistakes and do the opposite by initiating
positions in the direction of the trend. This creates further acceleration, but of
course, by this point, smart money is already slowly taking profit and unwinding
their positions in the mirror image of the first stage.
Note: it is perhaps simplistic to think of two separate groups of participants and
label them smart money and dumb money. These are the two extremes, and in
reality, there is a trader for every notch of the scale in-between. While the really
smart trader is the seller at the high and the really dumb trader is the seller at the
low, there are more traders with small profits and small losses who trade in the
middle of the trend. While we can’t always be the smart trader, as soon as the
central point is recognized, there are clear expectations and therefore trades to be
made.
Classifying the stages of trend in the above way can help us understand why
waves 1 and 5 are often similar in size and structure. It also explains why wave 3
is usually longer and stronger as more traders recognize and chase the trend.
Unfortunately, though, not every trend unfolds cleanly and with these
parameters. Indeed, it is rare to find a truly perfect trend like the ones shown
here as there are so many possible drivers that can blow it off course.
Take the post-2016 election rally, for example. The initial stages of the Dow
Jones trend unfolded cleanly enough and the acceleration / central point at
19,400 seemed clear. Using this to create a mirror image chart projected
continuation to around 24,000 (as shown by the chart shown in purple below).
However. President Trump’s tax cuts led to a very extended last phase of wave 3
and all symmetry was lost as the DJI went on to peak at 26,951.
Traders could still have made good money on the above projection, even if longs
would have been closed early. Certainly, no-one should have taken a loss as
another key point is that the central point provides a clear inflection point and
stop level. Once it is formed it should not be tested again until the trend has
completed.
If there is no clean symmetry or central point, we can choose not to get involved
or just be careful and conscious of the potential for 'messy’ patterns. Even if the
trend starts cleanly and appears to be playing out like textbook, events can take
over and lead to divergent patterns. A false expectation that everything should
happen perfectly can lead to mistakes and losses.
The Amazon (AMZN) downtrend shown below shows a very clean structure
until the very last stage when wave 5 reversed higher again before the ideal
target based on wave 1.
Source: Tradingview
The premature reversal higher means late shorts may have been caught out or
longs waiting for the perfect entry may have missed the trade. There is never any
room for complacency.
Sometimes we can find clarity by zooming in to the most central trend and
excluding waves 1 and 5. These are often choppy and diagonals / wedges are
common. Wave 3, however, must be clean.
If the price action in the Amazon chart above is viewed in a lower timeframe,
zooming in from a 1-hour chart to a 15-minute chart to focus on the box
surrounding wave 3, it looks like this:
Source: Tradingview
This time there is good symmetry and an equality between waves 1 and 5. Often
the cleanest action is rooted within wave 3 and its importance cannot be
overstated.
5. Other Types of Fractals
Analogues
So far, this book has focused on self-similar fractals as these are fractals in their
truest definition. However, the concepts can be expanded into other areas and
there are many types of ‘fractals’ in a looser sense. For instance, ‘analogues’
form when a particular stock or market repeats the price action in another stock
or market. This is of no use if the patterns are made at exactly the same time, as
is so often the case in heavily correlated markets, but if one market leads by
months, days or even just a few hours or minutes, it can be used to guide trades
in a lagging market.
Analogues are often misused and have been given a bad name by the large
number of sensational looking charts circulating on Twitter and elsewhere. A
scary projection can be made by matching a current chart to the early stages of
the 1929 crash or 1987, but just because some price action matches, there is
nothing to say the similarities will continue. Let’s face it: calling for a crash is
not a high probability activity.
Try to put the odds in your favour: analogues should have more in their favour
than a small portion of matching price action. Ideally there should be some kind
of logic behind the comparison. Are the drivers or narrative comparable? Are the
participants the same (the same people tend to do the same things)? Is enough of
the price action matching to confidently say positioning and sentiment should be
similar? Correlated markets can also help increase the odds of a match e.g., the
S&P500 in 2022 compares to 2000-2002, as does the US dollar. Other sectors,
stocks and ratio charts can also provide a piece of the puzzle.
In the chart below, Goldman Sachs(GS) and Bank of America(BAC) were not
only repeating previous declines in 2016, but the relative strength / weakness at
both lows gave extra confidence to call the reversal.
Source: Prorealtime
The analogue below shows Apple (AAPL) lagging Amazon (AMZN) by around
a year when the comparison was first spotted in mid-2016.
Source: Prorealtime
The premise for this comparison was simple and logical: Apple and Amazon
seemed to take it in turns to rally. Amazon consolidated from 2014-2015 in a
30% range while Apple rallied. When Amazon finally broke out and started
trending higher again, Apple stopped rallying and consolidated in a similar 32%
range. It was as if the same participants were trading both stocks and buying
interest flowed from one stock to the other. After a whole year trading in a range,
it was proposed on Seeking Alpha that Apple was finally ready to follow
Amazon higher.
As can be seen below, this played out very well over the next two years.
Repeating Gilead (GILD) earnings reactions were already described in chapter
three, but there was further supplemental evidence available at the time. This
came from a different stock in the same sector called Valeant (VRX) which had
similar earnings reactions a few months prior to Gilead and provided additional
evidence to the buy signal.
Source: Prorealtime
A similar lead could be seen in Netflix (NFLX) and Meta (META) in early 2022.
Again, the two stocks are quite different fundamentally, but had similar rallies
from the 2020 lows into the late 2021 highs before reversing lower under
mounting headwinds. Netflix then made a huge gap lower on earnings on the
21st January; a gap that did not fill. With the similarities up until that point,
Meta looked dangerous going into earnings on 2nd February.
The comparison didn’t exactly make Meta a short – after all, anything can
happen on earnings day. However, it certainly warned not to buy or hold.
Moreover, once Meta had made its gap down, the behaviour of Netflix continued
to act as a guide. Its post-earnings drop took it –50% from its 2021 high before
bouncing; Meta ended up dropping –51% from its 2021 high before its first
proper recovery in March 2022. Meta than made a +27% gain which was again
comparable to the +31% rally Netflix had made earlier.
Comparing two different markets often draws criticism – why would one market
follow another when there are no fundamental similarities? Unfortunately, there
is no definitive answer to that question, but when we consider price patterns are
created by human emotions and reactions, then the notion of them repeating is
not so far-fetched. The exact triggers may be different, but the emotions and the
responses can be the same. This is evident in the way different markets follow
the classic bubble pattern.
The Bubble Template
The observation that many different markets have followed the same boom and
bust pattern is testament to how fundamental disparities don’t always matter as
long as long as they trigger the same emotions – in this case, fear and greed .
Rodrigue’s classic bubble chart breaks down the many stages and drivers
involved.
Source: Dr Rodrigue / Wikimedia
Even when the bubble is obvious to many, it can continue to play out to the same
set pattern right in plain sight.
Source: Prorealtime
Needless to say, traders with any knowledge of bubbles or fractals were not
investing in the ARK Innovation Fund (ARKK) in 2021. Moreover, by
comparing ARKK to the Nasdaq, shorter-term trades could be taken in both
directions. This was highlighted in an article in May 2022 where it was said
“there could be a bounce developing, but that the low is not in for ARKK.”
ARKK then rallied from May to August, but by October 2022 had made further
lows.
Even after bubbles pop and crash, amazingly some markets go on to repeat the
same behaviour all over again. Many traders will insist “this time is different”
but when the same conditions are spotted in the early stages, we know it
probably isn’t.
When Bitcoin crashed in 2018, previous action in gold and the Nasdaq not only
provided a guide into how it would bottom, but also inspired the call for what at
the time seemed an improbable second bubble.
Source: Tradingview
As we can now know, this played out well. The chart below shows how the
second Bitcoin bubble developed in line with gold and the Nasdaq.
Source: Tradingview
This comparison is still ‘active’ in that it can still provide a useful guide at the
time of writing. Will gold and the Nasdaq now follow Bitcoin’s lead? Or to look
at it from another perspective in the chart below, will ARKK follow Bitcoin,
gold and the Nasdaq and form a second bubble? Due to the differing timeframes
in each market, all are possible.
Source: Tradingview
Note: although the above chart suggests a second bubble in ARKK is possible, it
also warns it will take many, many years to form and ARKK will be ‘dead
money’ for a long time.
Bubble patterns are created by increasing interest and increasing greed. The fear
part comes later and is driven as much by the crash in prices as it is by any
fundamental event. After all, any events which occurred during the crashes in
Bitcoin, gold or the Nasdaq were all significantly different.
But what if the events themselves are comparable or even identical? Can markets
be expected to repeat the same price action when the same trigger repeats?
Event Fractals
Similar events and outcomes can trigger similar market reactions. This is a
straightforward assumption, but there are a number of other inputs that
complicate things. A look back on NFP data and subsequent market reactions
over decades will show no set pattern for say, the S&P500, when NFP beats and
when it misses. There are simply too many events, reactions and other variables.
Another consideration comes from positioning and sentiment leading into an
event. If the S&P500 rallies over 5% in the week leading up to FOMC, could the
same reaction be expected when it falls over 5% before the event even if the
outcome is the same? Probably not. Filters therefore need to be applied to try
and even up the data, but this is more an area for quants and back testing
software.
An alternative – and relatively simple - solution comes from a comparison of
price action leading into an event. This accounts for the fundamental drivers and
all external factors as it assumes they are reflected in the action itself. If a market
makes a sharp drop several days ahead of a release and then forms a bear flag,
the pattern reflects nervousness about the outcome. If the outcome is more
positive than feared and the market rallies again it will provide a guide should
these parameters repeat again. It doesn’t really matter if the worry was a too high
inflation print or too weak a jobs number – the action before an event can guide
the action following it.
Of course, if there is a recent occurrence of similar price action into the same
event (and in the example below even at the same time of year), the odds of a
similar outcome are increased. Gold (XAUSUD) ahead of the December 2016
rate hike is a good illustration as it was compared to the December 2015 hike in
this article.
Source: Prorealtime
The low came on the same day the article was published and led to a strong rally
of over +12% by the end of February.
Intra-day charts are useful for spotting repeating patterns in the days before an
event. A 15 -minute chart picked up similarities in S&P500 between the May
FOMC and the lead-in to the June FOMC and projected an unsustained spike
higher followed by further lows.
Source: Tradingview
This was pretty much what happened.
Fractals and repeating patterns can undoubtedly give us an edge around certain
events. Donald Trump’s election in 2016 and the UK Brexit referendum were
very different incidents on different sides of the Atlantic, but both took the
market by surprise and caused brief panic selling in very similar ways. Since the
Brexit drop soon reversed and led to a strong recovery, when the S&P500 futures
dropped in the immediate aftermath of the election results on November 9th, the
below fractal suggested not to panic but to buy.
Source: Prorealtime
Stock markets soon recovered all of the drop and went on to develop one of the
strongest rallies in years.
Source: Tradingview
Not all panic selling will create the same post-event pattern. The 2016 US
election and Brexit referendum had some similarities in that they were scheduled
events and had binary outcomes. Trump could either win or lose, Brexit would
either be agreed or rejected. Because of this, markets were not completely
unprepared and the reactions had good odds of being similar. However, if there
was a complete surprise like a terrorist attack or an assassination, the reaction
would presumably look different as markets would have to improvise a quick
response. Something completely unexpected like this could be classed as a Black
Swan event, but even these lead to predictable and repeatable behaviour.
Black Swan Events
Black Swan events come in various forms and can provoke very different market
reactions, at least in terms of scale. However, the basic stages are always the
same and knowing them provides a way to profit from the panic.
Black Swan theory was developed by Nassim Nicholas Taleb. Without going
into too much detail, a Black Swan event can be identified by three common
traits.
The event is a surprise (to the observer).
The event has a major effect.
After the first recorded instance of the event, it is rationalized by hindsight, as if
it could have been expected; that is, the relevant data were available but
unaccounted for in risk mitigation programs. The same is true for the personal
perception by individuals.
The term has evolved a somewhat looser meaning over the years. According to
this Business Insider article, these are '9 black swan events that changed finance
forever':
The Asian financial crisis
The Dot Com crash
9/11
The global financial crisis
The European sovereign debt crisis
The Fukushima nuclear disaster
The 2014 oil crisis
The 2015 Black Monday
Brexit
Covid-19 could be added to that list, but some of the entries are questionable.
For instance, it is debatable if Brexit was a Black Swan; everyone knew the
exact date and timing of the vote. The result was surprising to some, but the
markets had plenty time to prepare. You could also say the Dot Com crash and
other crises were not completely out of the blue. 9/11 and the Fukushima disaster
on the other hand were. This is an important distinction to make as the reaction
of the market is often completely different depending on if it has had time to
prepare or not.
The initial reaction to a Black Swan is an obvious one – sell. This happens
quickly as the instinct is to sell first and ask questions later. Once the questions
are asked and partly answered, assets can recover again, but there is no real way
of knowing where the bottom will be. Obviously, some events are more
significant than others and buying the panic can be very risky. Even so, Black
Swan events provide several trading opportunities in the aftermath and relatively
safe ones too.
9/11 certainly fits the definitions of a Black Swan event. It shocked the world,
and the markets. If you were unlucky enough to have bought the day before the
planes hit, you would have had to sit through a market shut down, then watched
the S&P500 plunge 13% in the next five sessions. To make matters worse, this
happened in the middle of an already severe bear market. Could a reversal
pattern be trusted?
The short answer is, ‘yes’. What all Black Swan events have in common is that
once price reverses, the recovery returns back to the level it was at when the
event first happened. This ‘point of origin’ or ‘scene of the crime’ is an important
target as it also acts as resistance when price returns there.
Source: Tradingview
As a Black Swan event is completely unexpected, even inside traders at the very
top of the market food chain cannot position for it and will therefore be hurt by
the downside reaction. Perhaps this is why price always recovers to the point of
origin - doing so allows smart (or connected) traders out relatively unharmed.
The 2011 Japanese earthquake and subsequent Fukushima nuclear disaster had a
significant effect on many markets, especially the Nikkei (EWJ).
Source: Tradingview
Prices crashed 20% in two sessions, but yet again price returned to the daily
range where the crash originated. This level provided a clear target for longs.
Furthermore, when price eventually reached the point of origin, prices reversed
sharply and crashed 20% again.
The logic behind this behaviour is simple; as traders take the chance to break
even, the selling creates resistance and price falls. We don't know how much of a
reaction there will be, but the vast majority of times there is at least some
downside move.
Looking further back, there are some other memorable Black Swan events such
as Pearl Harbour in 1941.
Source: Prorealtime
And the assassination of JFK in 1963.
Source: Prorealtime
Again, we see the repeated price action and the return to the scene of the crime.
A more recent example comes from the currency pair EURCHF following the
de-pegging crash. The return to the point of origin took years but once it
returned, there was a great opportunity to short and replay the entire crash again
(even if the move down was much slower second time around).
The Swiss Franc debacle is perhaps less related to the previous examples of
attacks and natural disasters and more in tune with Taleb's definition. Some
people knew the peg would be pulled (it had to be arranged and planned for),
and there were warnings. You could say 'it is rationalized by hindsight, as if it
could have been expected; that is, the relevant data were available but
unaccounted for in risk mitigation programs'.
6. When Fractals Fail
It would be dishonest to say fractals never fail. In fact, the truth is all fractals
must fail eventually as they either get so small they disappear, or more likely,
another pattern becomes dominant. This concept was covered earlier in chapter
3.
One of the main challenges is to know when the pattern has diverged enough to
say it is not repeating, or when to stick with the comparison despite a divergence
as we can’t expect perfect repetitions each time. Unfortunately, there are no hard
rules on this subject and decisions must be taken ad hoc and depending on
personal parameters. Fractals can diverge for a number of bars and then recorrelate so this should be taken into account when constructing trade ideas.
It can be helpful to use complimentary techniques, and this is again where Elliott
Wave comes in handy as it gives a much more rigid framework for validation.
For example, if a fractal suggests a falling market should reverse higher again,
an accurate Elliott Wave count should be able to say exactly where this view is
invalid and the fall is likely to continue.
The breakdown of a fractal can be seen below in the Gilead (GILD) chart below
which is a continuation of the example in chapter 3:
Source: Prorealtime
The initial projections worked well but the patterns diverged on October 7th
when a leaked tape of Donald Trump's comments hurt his election chances and
all Biotech stocks sold off.
Events like this can knock fractals off course, either temporarily or permanently
as a new dominant pattern and new driver take over. In the Gilead example, the
post-earnings move repeated almost perfectly for two months and over several
major pivots. However, the influence of earnings naturally started to wane and
trying to trade the projected reversal on October 7th was perhaps pushing it too
far as a new macro driver assumed control.
About the Author
Andrew McElroy
Andrew McElroy is chief analyst at Matrixtrade.com and a regular contributor at
Seeking Alpha. Born and raised in Scotland, he now lives in Bulgaria with his
wife and two children.
Daily Analysis: Matrixtrade.com
Email: andrewmcelroybg@gmail.com
Twitter: @elroytrader
Seeking Alpha: https://seekingalpha.com/author/andrew-mcelroy
Disclaimer: All the information contained in this publication is provided as
general commentary and does not constitute investment advice or
recommendation. Andrew McElroy will not accept liability for any losses which
may arise directly or indirectly from use of or reliance on such information.
You are advised to conduct your own research before making a decision. Trading
financial instruments carries a high level of risk, and may not be suitable for all
investors. Before deciding to invest in any financial instrument you should
carefully consider your investment objectives, level of experience, and risk
appetite. You should be aware of all the risks associated trading financial
instruments, and seek advice from an independent financial advisor if you have
any doubts. Past performance is no guarantee of future gains.
©2023 Andrew McElroy
Download