January 2017 Issue
© January 20, 2017


Question: which Elliott wave analyst labeled the following chart?

I’ll give you some hints. This analyst:

  • Does not read the news
  • Has no political opinion
  • Does not get tired, scared, bored or excited
  • Does not take bathroom breaks
  • Is available 24 hours a day, 7 days a week
  • Constantly improves over time
  • Will analyze any market (or data series), no matter how obscure
  • Applies the Elliott Wave Principle (WP) without bias, subjectivity or emotion
  • Has such a long name, that we usually use an acronym

If you can’t think of any human analyst with these traits, then you’re on the right track, because this particular analyst is made of silicon. The above chart was produced entirely by a machine running the latest and most advanced incarnation of our Elliott Wave Analysis & Validation Expert System (EWAVES) artificial intelligence software. That’s right: after several years in the making, including incalculable hours of research, building, rebuilding, testing and debugging, the next major release of EWAVES—version 2.0 beta—is finally here. For this revision, we rebooted the entire project from the ground up.

It’s Charty Time

Take a look again at the chart on the front page. The subdivisions involved in it, especially of wave

(iii) of 8, detail the wave structure at four different degrees, providing context for both short and long-term market opinions. The wave labels are nearly exactly where a human expert would put them.

Nit-picking turns up an imperfection: the first (ii) label (within 6) should be shifted to the low a few trading days later. Such an adjustment would further bolster the case for a motive wave 6 by making zigzag subdivisions clear for both (ii) and (iii) ((i), (iv) and (v) are already clear zigzags), which are typical of a leading diagonal.

The good news is that the fractal nature of the market allows for the impact of these kinds of short-term imperfections to be erased through the context that superstructure provides. The subsequent steep and deep wave 7 retracement and the large wave 7 support the interpretation provided. Therefore, for all practical purposes this analysis was effective in providing a template that would have correctly anticipated further rising prices.

Of course, we at Qualitative Analytics (QA) are perfectionists, so we’re already busy working on improvements for a future EWAVES 2.X version, with the ultimate goal of exiting beta. Nonetheless, it’s important to realize how far we have come: All in all, the technology underlying this preliminary 2.0 release is a leap ahead of version 1.1.

Directly above is the same chart again, but this time with our brand new labeling scheme. This format is both what we use internally and also what we will use henceforth for publishing live charts to Flash Services subscribers coincident with the release of each new Flash recommendation.

Instead of stylized numerals that denote absolute degrees (Cycle, Primary, Intermediate, etc.), our new labeling scheme uses simple Arabic numbers and Latin characters, varying only by size and color to denote relative degrees. This simpler format allows for more rapid visual orientation, which is helpful in a

real-time market environment. The increased symbolic uniformity is truer to the fractal nature of the Wave Principle (WP).

Our new format also disambiguates Elliott wave patterns through subscript notation, a feature missing from traditional wave labeling. For example, a flat is af bf cf whereas a zigzag is az bz cz, and so on. This preserves the traditional symbols while also adding explicitness.

Gold through December 31, 2015

Gold Analysis

This above chart shows EWAVES’ analysis of the decline in gold that began in 2011. The data was not loaded past the end of 2015, so EWAVES had no knowledge of what was to come. This chart is interesting because it shows two patterns that did not appear on the front-page chart: a triangle (subscript “t”) in the wave 4 position, which precedes an ending diagonal (subscript “e”) at the wave 5 position. This is a powerful combination of patterns indicating an impending market reversal of at least one degree larger. Triangles almost always occur prior to the last wave within a trend one degree larger. Rarely, they can appear at the end of double threes, prior to the last wave within a trend two degrees larger. The ending diagonal indicates that the ensuing move should be swift in the opposite direction, ideally retracing at least the price span of the entire diagonal. It also provides a logical place for a stop-loss order, since wave 5e cannot meaningfully break the trendline produced by waves 1e-3e without invalidating the wave structure.

EWAVES’ machine analysis matches what experts at Elliott Wave International (EWI) were expecting in December and January 2016. Here is an excerpt from the January 8, 2016 issue of the Elliott Wave Financial Forecast (EWFF):

According to multiple news accounts, it’s official: gold has lost its luster. The precious metal was down 10% in 2015 for the third consecutive yearly decline, its longest annual losing streak in 18 years. No one noticed at the time, but gold’s luster—which should refer to its prospects, not its past—actually disappeared in September 2011, when prices peaked at $1921.50 and EWFF was resolutely bearish. Now that gold is down over 40% from its peak and nearly everyone is predicting even lower prices for 2016, it’s time to be extra attentive to the wave structure and sentiment measures in anticipation of a significant countertrend rally. The Federal Reserve’s December rate hike is considered a negative for the metal in many quarters. In the wake of that move, the leading daily financial newspaper lent support to the myth that “higher rates tend to weigh on gold, which pays its holders nothing.” “There is some fear that there will be further rate- hike moves,” worries the president of a large commodity-focused hedge fund. Last month, however, EWFF carefully dispelled the notion that gold cannot rally amidst rising interest rates (see p.7). The claim is simply rationalization of bearish sentiment on the metal. Pessimism toward gold remains extreme, as the CFTC’s weekly Commitment of Traders report shows that Money Managers hold a record net-short position in gold futures and options contracts, betting on a continued price decline. This cohort of speculators is usually caught in extreme wrong-way bets at significant trend reversals.

As published in EWFF, December 4, 2015

Chart from the December 4th, 2015 EWFF, which was completed the day of the bottom in gold and published the following morning
Gold Through January 26, 2016
Never trust a black-box system

If we pull in just a few more weeks of data and rerun the analysis, the chart on the top of this page shows EWAVES recognizing the initial stages of an upward wave of the same degree as the entire move down from 2011. The software’s minimum expectation is for gold to launch the largest rally in over four years. (The above wave count also implies a specific, objective fail-point if the forecast is incorrect, located just below the 5e label.) This is clearly not your typical black-box “buy” or “sell” recommendation. EWAVES is glass box, showing subscribers what it is thinking, why it is thinking it and what would invalidate its conclusions. Its output is detailed and nuanced. A subscriber is free to examine the chart and its implications, and choose to accept, reject or amend the chart’s message.

The chart below shows what happened next. Although “nearly everyone predict[ed] even lower prices for 2016,” the metal experienced its largest rally since 2011. The WP offers the only basis to forecast such an event, which EWI’s team delivered. And now we know our new silicon friend at QA would have generated the same forecast.

Gold through August 1st, 2016

Although in the above case EWAVES’ analysis matched that of EWI’s analysts, it is important to note that we did not sit around thinking, “How can we get EWAVES to label gold a certain way in late 2015?” The program’s core components were developed long before this pattern manifested. More fundamentally, we do not practice curve fitting. We developed (and continue to develop) EWAVES based on the theory and practice of Elliott wave analysis, focusing solely on pattern recognition irrespective of any specific data series, time period or scale. So, it is not surprising that at times its message will parallel that of talented human analysts.

One final point: Going through all the news prior to the surge in gold—worries about Federal Reserve rate hikes, the lack of yield on gold and the consensus of the public—it’s amazing to think that none of it matters. All popular commentary is simply rationalization of non-rational herding impulses. EWAVES is immune to such banter. It simply interprets patterns in the data. The patterns reflect the unfolding mass psychological dynamics, which ultimately create events and therefore the news.

Dow Jones Historical Analysis

Dow Jones from January 1932 to September 1937

Now let’s time-travel to nearly a hundred years in the past. Above is EWAVES’ labeling for the DJIA from the Great Depression low in 1932 to a post-depression recovery high in 1937. The chart contains an extended fifth wave, which completes the highest-degree structure visible here, pointing the way towards a multi-year bear market and economic contraction.

The main point of showing this chart is to demonstrate that nearly a hundred years ago, the social mood patterns that drive financial markets were exactly the same as they are today. These patterns hold going back a hundred years, a thousand years, or quite possibly even a million years (depending on exactly when the specific capacity for shared social mood evolved within humans). Therefore, contrary to the opinion of many economists, the remote past is just as relevant to financial theory as the present is. Not only that, but centuries-old bull and bear markets also provide context that informs an analyst’s judgement of wave patterns unfolding in the present. Therefore, ancient history is relevant not just for theory but also for practice.

One potential reason that many financiers regard the distant past as irrelevant is that financial markets exhibit quantitative non-stationarity. Simply put, any quantitative statistical property of the market is transient and can disappear at any time in the future. The very first issue of EWAVES Flash provided LTCM as an example. Non-stationarity always eventually causes model burnout for traditional quantitative (quant) methodologies.

EWAVES is based on qualitative extrapolation. It does not depend on any quantitative regularities in the data. Rather, it relies on the fact that while financial markets are quantitatively non-stationary, they are qualitatively stationary. Said another way, the general form of Elliott waves has not changed in all of recorded history, despite the fact that, like snowflakes, the quantitative properties are so diverse that no two Elliott waves are identical.

How did we build EWAVES as a qualitative rather than quantitative program? One way is by upholding very powerful principles throughout the entire codebase. One of these principles is scale invariance, a central tenet of EWAVES. This property is strongly enforced through our test infrastructure. Our software upholds several other major invariants (which are proprietary yet ultimately derivable from WP), which combine to place it into a fundamentally different paradigm than that of conventional quant systems, which do not adhere to such principles. EWAVES is unusual in the realm of analytical software in that it is qualitatively consistent while being quantitatively highly variable. This is the exact opposite of the norm in the quant paradigm, which searches for quantitative consistencies irrespective of qualitative consistencies.

The subsequent decline following the peak

Individual Stock Analysis

Elliott waves also manifest in individual stocks, especially when they are widely followed. Below is EWAVES’ analysis of Goldman Sachs (GS) through November 2015. Even an astute market technician could have missed this setup simply because of the sheer volume of individual stocks to watch. But EWAVES, a tireless machine, can pay full attention to a vast number of stocks each day.

Goldman Sachs through November 2015

The next chart, below, shows how the expected decline unfolded. In May 2016, CNBC ran the headline “It’s the end of Goldman Sachs as we know it,” as GS was to begin “shedding traditional staff, like traders and bankers, and hiring [software] developers.” As always, news follows market action. (This socionomic fact is convenient, since it is due to this theoretical insight that technical analysis can be utilized!) Several months after this decline, GS eventually completed a larger pattern and rallied to new highs.

Goldman Sachs through February 2016

Flash Services

We have upgraded our Flash Services analysis engine to EWAVES 2.0 beta, running on dual high- end workstations totaling 16 processor cores and 64 gigabytes of memory. Starting today, all new Flash recommendations will include an Elliott wave chart. These charts are entirely machine generated by EWAVES 2.0 beta.

Each day, our strategy module examines EWAVES’ analysis and looks for certain qualitative setups. When it identifies a high-probability setup, it issues a Flash recommendation to subscribers, complete with entry information, stop criteria and a wave count. It watches the recommendation each day, issuing stop adjustment notifications as necessary. It concludes the recommendation with either a stop-met or exit notification. Flash gives the best of both worlds since it (1) provides black-box style, complete support for the subscriber while also (2) delivering glass-box style transparency and customization by providing a wave count with each recommendation.

Flash Services cover ETFs, Futures and individual Stocks. Futures and ETFs cover two timeframes, designated Traders and Investors. The Traders timeframe issues more frequent recommendations that are shorter in duration vs. Investors. For all services, each recommendation may last from less than a day to over a year. The range is wide because the market is in charge. If the program decides quickly that a recommendation should be jettisoned, we recommend an exit. But in a trending market, our software lets it run as long as possible.

We will continue to post an audio update every week on Fridays by 5PM EST. The update is available for subscribers on the myEWI Flash portal, or for your convenience you may bookmark this direct link on your device: ewaves.com/audio. The audio update is recorded regularly. Our recordings discuss progress in EWAVES, newsletter topics, trading theory and more. These updates provide more frequent insight into the day-to-day workings of the EWAVES team and our thought processes.

Click here for more information on Flash Services

What’s in the Box?

Behind the scenes, EWAVES is divided into four main components: (1) the fractal-analysis engine, which performs all of the necessary work to analyze nested patterns, (2) the Elliott wave specifics that codify each pattern we care about, (3) the graphical interface or GUI and (4) various separate components that perform tasks such as dealing with financial data, running trading simulations, dynamically generating charts and issuing Flash recommendations.

Subdividing the project in this manner makes the process of programming EWAVES manageable, since each team member specializes in different areas. One developer’s focus has been the fractal-analysis engine along with the strategy module; another owns the GUI, data, notification, IEWs and charting; and another primarily develops the module handling Elliott wave specifics.

EWAVES uses a client-server architecture. This means that instances of the core analysis engine run on high-end servers, each waiting and listening for remote commands. The graphical interface (the client) is completely independent from the analysis engine and can be run from nearly any computer, tablet or smartphone. It can connect remotely to EWAVES’ servers and issue commands to analyze data and report results. The ability to control an “army” of machines from a single interface is immensely powerful when it comes to running large test simulations. Eventually, this design will also allow remote users to issue on- demand analysis requests on any streaming, static or custom data set.

Man vs. Machine

Many of the charts in this newsletter appear similar to the work of human analysts. But they have a radically different effort profile. While it takes significant time and mental exertion for human beings to analyze a plethora of markets and label charts, EWAVES automates most of this task, reducing the human capital cost of analyzing even a hundred thousand datasets to near zero.

Does this mean that EWAVES’ analysis is costless vs human analysis? Far from it. Of course there are equipment costs, but the real cost of its analyses are untold man-years of research and development effort. This includes both the past and the future as we continue to improve its algorithms. In other words, at QA we do not focus on analyzing markets per se, rather, we focus on the process of analyzing markets (i.e. we are second-order oriented). Within the second-order realm, there are nearly infinite research areas for yet the next version on which we are already hard at work. Our foci range from strategy improvements all the way up to machine learning modules that allow the program to self-improve through experience.

The famous tech venture capitalist Marc Andreessen noted in 2011 that software is “eating the world.” We agree. But building the software, especially when it involves inventing new techniques to solve a problem that nobody else has heretofore solved, is a massive undertaking. Our engineering team, backed by analysts with decades of Elliott wave experience, is in a unique position to provide a solution. Today we reached a major milestone with the release of EWAVES 2.0 beta, laying the foundation for the future of Elliott wave and socionomic forecasting.


The EWAVES Flash publication is open-access. Feel free to share the link http://ewaves.com/monthlypublication/1701 for this issue, or visit ewaves.com and click on the “newsletter” tab to read other issues or to sign up for alerts when new issues are published. To subscribe to Flash Services, use the “services” tab or go directly to ewaves.com/flash-services

‘Like’ us on Facebook to receive exclusive updates & videos.




1Aronson, D., & Masters, T. (2013). Statistically sound machine learning for algorithmic trading of financial instruments: Developing predictive-model-based trading systems using TSSB. Place of publication not identified: CreateSpace Independent Publishing Platform.

2Andreessen, M. (2011, August 20). Why Software Is Eating The World. Retrieved January 05, 2017, from http://www.wsj.com/ articles/SB10001424053111903480904576512250915629460