Elliott Wave: Solving The Probability Problem
"Pride of opinion has been responsible for the downfall of more men on Wall Street than any other factor."
- Charles Dow
Without a doubt, the greatest drawback of using Elliott Wave theory (EWT), and the reason most traders avoid it, is its high degree of subjectivity. Even the most experienced Elliott experts can have trouble agreeing on wave counts and forecasts on the same issue, index or commodity. Where there is subjectivity, there is uncertainty. Overcoming this uncertainty requires the guidance of solid probabilities determined by statistical analysis. Let's take a look at a development that, through computer power, has helped take the subjectivity out of the Elliott Wave theory.
Adverse Effects of Opinions
As all successful traders have learned, solid rules are essential to long-term success. The possible variations in deriving an Elliott Wave count while either strictly or strongly adhering to the original rules make it hard to know which count is best. Ultimately, the analyst chooses the count with which he or she feels most comfortable, but that is often based on little more than an educated guess or past experience. As such, the analysts may be prone to get "married" to the opinion, even when logic might dictate otherwise.
Opinions never hurt when it comes to the markets until there is money at stake. If the market goes against the trader, unquestioning loyalty to an opinion can be very costly. Fear of being wrong, combined with pride of opinion, is a deadly handicap in the trading business; emotional gremlins are more responsible for traders' failings than any other single factor.
In his book, "Trading In The Zone" (2000), trader Mark Douglas helps traders break the emotional habit. All great traders who have sustained success have learned to think in probabilities, realizing that trading is nothing more than a numbers game. Successful traders have made it a habit to make decisions only if they know the risk/reward ratio and if they have backtested and recorded the past success of their system. Emotions don't control these decisions. Probabilities do.
Until the late twentieth century, however, Elliott traders did not have the luxury of knowing the precise probabilities of success or failure of a forecast. Because of the complexities of Elliott Wave, there was no way of knowing what to expect with any degree of mathematical confidence from even a single Elliott Wave pattern, let alone a complicated one spanning a number of years. No public databases providing that information were available.
Putting Elliott to the Test
In 1994, a small team from Perth led by Rich Swannell began designing Elliott Wave computer programs for traders. Swannell was a programmer first and trader second. Very few in the world of trading are good at both.
During the early years, the team consulted with Elliott veterans, conducted intense research, and developed what Swannell claims was the world's first comprehensive software program designed to analyze price data using the rules and guidelines of Elliott's theory. The problem with the software was that it was based on observations and not an exhaustive statistical analysis of wave reliability. And, while results from the software were respectable, without probabilities the trader was still trading blind. How could a way be found to overcome this weakness?
The team came up with a novel solution. Swannell developed a screen saver in 2001 that would work in the background on the computers of more than three thousand volunteers. While not being used by their owners, these machines would be scanning a universe of stocks, commodities and indexes to search for and analyze Elliott Wave patterns. The goal was to determine once and for all which patterns worked, which did not and even whether the Elliott Wave theory itself had sufficient merit to trade it with confidence. It was all based on mathematical probabilities.
After eighteen months and hundreds of thousands of hours of computer time, the team had enough data to start analyzing it. For those interested in more details, Swannell wrote a book about the experience, "Elite Trader's Secrets: Market Forecasting With The New Elliot Wave System" (2003); it includes a good analysis of Elliott patterns. Here is a summary of what they found:
- Not only did the Elliott Wave theory prove to be statistically sound, the research was able to generate the probabilities of a forecast being correct. In other words, the trader could now know the chances of a wave pattern and the resulting forecast with a low margin of error (statistical significance).
- The most common Elliott Wave patterns were often significantly different in both shape and frequency than the previous conceptions of them. Some patterns that were previously believed to be reliable did not work often enough to be used with any degree of confidence.
- The team confirmed Murray Ruggiero's finding that a correct wave count is not the most important factor in trading. Even with the help of a good program, all Elliott forecasts are, at best, an educated guess: a trader can never be certain because there is always a larger pattern that cannot be included in the analysis unless he or she goes back to the beginning of time. Swannell's team found that since many alternate counts result in similar forecasts, this problem of possible inaccuracy is not as critical as many previously thought. As long as a count is arrived at logically, adheres to the rules and is confirmed over various time periods, it doesn't matter what the larger degree (next largest wave pattern) is. In Swannell's findings, the most probable scenarios gave exactly or at least very similar forecasted results. This finding is crucial to a trader's success and means that, as Ruggiero says, the count is of less importance than the penalty for being wrong, which is the loss on the trade.
- By performing forecasts in various time frames, the team separated the issues that worked from those that didn't. Forecasts for those that exhibited no consensus over various time periods were deemed unreliable (see our example below for a more detailed explanation). The probability for failure in most cases was greater than the probability for success, so why take the chance?
Of the thousands of equities, indexes and commodities tested, Swannell's team found that in about 65% of the cases, Elliott Wave theory proved too unreliable to be used to trade with any degree of confidence. In other words, using the theory to trade the instruments included in this 65% would prove a losing proposition. It means that traders should limit their focus to the 35% that proved to be viable trading candidates.
But why did only about one-third of the candidates work using Elliott? It has to do with the basis of the Elliott principle, which quantifies market crowd behavior. Elliott Wave theory works best in equities that (1) have lots of volume (liquidity) and (2) move according to key forces of fear and greed on the part of many participants. When a security is not prone to this crowd behavior and is controlled instead by a few strong hands, Elliott patterns begin to break down. Issues traded by a few are more subject to manipulation and control and, therefore, are more difficult to forecast.
Elliott warned us that his theory worked best on indexes and very liquid securities, so Swannell's finding was not all that surprising. But now the notion was proven and quantified and a list of trading candidates was identified. In the process, a large amount of subjectivity and uncertainty was removed. All this information was now stored and available in a large database for immediate computer reference.
Coding and Applying the Lessons Learned
Through ongoing research and data from the screen saver program, Swannell's team further discovered that certain techniques, when consistently applied, generated impressive forecasting results. A new proprietary indicator based in part on the Elliott Wave oscillator also greatly assisted the trader in recognizing and confirming key reversal points.
The new discoveries were in part based on a prime tenet of technical analysis that if a pattern or method of analysis works in one time frame, it should also work in others. Moreover, the more time frames in which patterns confirm each other, the higher the probability that a forecast will be correct. For example, if a pattern within daily data agrees with one found in weekly data, the trader can have greater confidence in the pattern.
Swannell's team also found that a pattern confirmed in the same time period (that is, one day) over multiple date ranges was much more reliable. For example, using a starting point of the Oct 1987 low, let's say that we find an impulse wave consisting of three up-waves and two down- (corrective) waves in an uptrend. If this impulse wave agrees with patterns we find using a low from 1998, a low from 2002 and one from 2003, the reliability of a forecast made using these four time periods is substantially higher than one made in only one or two time frames. This confirmation of patterns has become the basic premise of forecasting using the program called the Refined Elliott Trader.
Taking the process one step further, the software Swannell's team developed rated each pattern, and those exhibiting a rating of 80 or more were reliable enough to use in a forecast. Those with scores above 100 were most reliable.
Let's look at an example analyzing the S&P 500 Index using end-of-day data.
The following charts show how the program is used to produce market forecasts. In this example, head trader Mark Lindsay takes us through the analysis process of locating confirmation Elliott Waves over four different time periods. We are looking for parts of the same wave patterns. The more closely they confirm each other, the more confidence we can have in the forecast. .
The Refined Elliott Trader looks for statistically significant matches and rates each pattern it identifies. Note on the left-hand side of each chart the list of numbers showing the rating of each pattern. We are looking for ratings (at the top of the list) of 80 or better. A rating of 100 is excellent and means that the pattern on the screen shows a strong correlation with similar patterns found in the database.
|Figure 1 – Long-term chart of the SPX from 1990 to 2004 showing large impulse wave and forecasted price using RET.|
In Figure 1 we start with a chart going back 15 years from 2004. It shows the longest-term chart with an impulse wave starting in late 1990. Wave 1 peaks in mid-2000, and wave 2 bottoms in October 2002. Wave 2 is a corrective ABC pattern. According to this chart in 2003-2004 and going into 2005, we were in a wave 3. The dark red rectangular pattern, which has a target area between 2000 and 3500 (indicated by dark-red vertical line), is the longer-term forecast.
|Figure 2 – Second shorter-term chart of SPX focusing in on the same impulse pattern shown in Figure 1.|
Next we isolate the wave 2 from March 2000 to October 2002 (Figure 2) to see the pattern in more detail. Remember, we are looking for pattern confirmation in each step of the process. As we take a closer look at each pattern, we see each wave in greater detail.
In Figure 2, we take a closer look at the period from late 2000 to late 2003 and isolate impulse wave 3. Impulse waves occur in waves of five while corrective waves like the one we see if Figure 1 between 2000 and 2002 occur in waves of three. Also note that forecasts generated in each chart confirm one another. This is important if the trader is to have a high degree of confidence in the ultimate forecast.
In Figure 3, we focus in closer, looking at the period from February 2003 and December 2004 showing impulse wave 3 in greater detail. It shows the first part of wave 3 followed by a ‘double 3' (sideways corrective wave) with the start of a smaller wave 3 (at the buy arrow).
|Figure 3 – Third shorter-term chart of SPX gives a closer look at the impulse pattern showing a similar forecast to the above charts.|
The final screen (Figure 4) shows the latest wave 3 from August 2004 to December 2004. The smaller parts of this wave consist of even smaller impulse waves 1, 2, 3, 4, and what looks to be the start of an impulsive wave 5 with an immediate price target from Dec 3 between 1220 and 1290.
This program also produces expected time ranges for each target (not shown in these figures).
|Figure 4 – Smallest time frame chart of the SPX showing end of impulse wave, forecasted price and the proprietary Refined Elliott Oscillator (lower window) used to help traders pick potential entry and exit points.|
It is important that the waves found by the program in each of the four charts confirm one other. If the overall pattern in the first chart is not found in the following three charts in a lesser degree, something is wrong and it's time to go back to the drawing board. If after performing a detailed search, the patterns don't agree, it's better to discard the prospect of trading the security than risk a bad trade.
Challenges and Solutions
The program developed by the Australian team may have solved a number of the challenges that existed, but it is not for those looking for an effortless trading system.
As a word of warning, the Refined Elliott Trader demands a thorough understanding and recognition of Elliott Wave patterns. A minimum of 50 hours is required to learn the 60 modules in the first level Elliottician course and then pass the four wave-recognition speed tests. Those with a phobia for learning or with little interest in probing the nuances of Elliott Wave theory are advised to look elsewhere.
But in the final analysis, all Elliott traders should take heart in the findings of this research even if they have no interest in using a computer program. It proves mathematically that the theory developed more than seventy years ago by R.N. Elliott is based on sound principles of market behavior. Actions taken by investors in the past do have chart pattern ramifications in the present, regardless of the reasons for these actions. The scope and duration of these reactions can be used to trade or invest longer term with greater confidence.
The total amount of cash and unaffiliated holdings compared to ...
The ratio of developed premiums to net premiums earned over a ...
The ratio of an insurance company’s net income to its policyholder ...
An increase in the value of an asset or account expressed in ...
The difference between an asset’s book value and the amount received ...
The assets of a mutual insurance company minus its liabilities. ...
Explore the use of accumulation areas in the analysis of traded securities. Learn about on-balance volume and its role in ...
Learn how the accumulative swing index is calculated using swing index. Explore absolute value and how it is used in the ...
Explore the difference between bar and candlestick charts. Learn how technical analysts use charts in the analysis of supply ...
Arbitrage and speculation are very different strategies. Arbitrage involves the simultaneous buying and selling of an asset ...