It is a fact that market patterns are the result of composite cycles. The summation of several cycles produces the distorted patterns we all see on the price charts.
By means of mathematics, it is possible and being accomplished today the ability to extract the underlying cycle patterns with a high degree of accuracy. Not necessarily 100% accuracy, of course, but a very high percentile.
In the work I do, market data is processed within a very special set of programs that produces Cycle Dates. These Cycle Dates are used to determine when to expect a market swing as seen on our bar charts. Each date is expected to locate when the market either makes a swing top or bottom. In some situations, it will merely mark an Acceleration Day, a day when the market moves fast in the direction of the trend producing an above average size price bar. This happens a very small percent of the time, however.
Some have argued that Random Date generating is just as good as Cycle Date generating. However, this idea is flawed. One very big fact against Random Date generating for predicting market swings is that the actual market data is not part of the equation. In generating Random Dates, one merely decides on a number of random dates to generate, or even let the number of the dates to be produced be randomly decided on. Obviously, there is no connection here between the actual market and the random quantity being selected.
Therefore, suppose in a given year that there are 100 market swings. Unless the random generator decides to select approximately 100 random dates to begin with, it will not come close to representing the actual market swing pattern. To randomly pick 250 dates within a year that only has 100 swings immediately exposes the weakness of the Random Date theory for prediction. To only pick 50 random dates would be just as bad. Unless the Random Date generating can somehow, randomly, decide on the correct number (or very close estimate) of dates to represent the actual turns that WILL occur in the next year, it will fail miserably and always does.
Cycle Date generation uses actual market data for its calculations. Cycles by nature repeat themselves, so past data is valuable in determining future market swings. By properly detrending the market patterns into cycle patterns or maps, an excellent estimation of future market pattern (swings) can be determined. The number of Cycle Dates will approximate the actual number of market swings if calculated correctly. This is far superior and more reflective of actual market patterns. If one year there are 100 market swings, but the next year there are 200, it is important that the number of actual dates produced dynamically determines this. Random date generation cannot possibly arrive at matching this change in quantity when its input is based on pure randomness.
Another shortcoming to the Random Date theory is the ‘spacing’ of the dates. Each market swings is distanced from one to the next by a variety of day lengths. While two swings may be close in distance this week, next may only contain one or none, and the distance is greater until the next swing. With Random Date generating, the distances produced are not based on anything relating to the market itself. They are just randomly selected. To see two random dates produced that are 7 days apart would not produce any confidence in the trader using them that the next two swings in the market are likely to be about 7 days apart, because the trader understands that the value of 7 was randomly chosen.
True Cycle Date generating is based on using real market data to produce real market results. Not only are the number of dates to be closely matching the actual number of swings to occur, but the individual distances between the dates are to closely match the actual distances between the forecasted swings themselves. Therefore, if the math being used to calculate the Cycle Dates is valid and accurate a high percentage of the time, which back testing can easily determine, noting that the next two cycle dates are 7 days apart will be taken more seriously than with random dates, and the trader gains this confidence by noting the accuracy leading up to the current time period.
The evidence is logical and reasonable, that Random Date generating cannot and will not resemble the actual market pattern that will occur for the period of time being forecasted. Although the random pattern will look just like a market chart pattern, which is why many believe the markets are random, that pattern will not match the intended target market it was meant to forecast. Cycle Date generating is the art and science of using real market data to produce a real market forecast.
Now, suppose one was to agree with the logic of this information, but only concedes this IF it really were possible to produce Cycle Dates that mimics the actual market swing pattern with a high degree of accuracy. But because many do not believe this is possible, it is assumed that this Cycle Date generating is no better than Random Date generating and thus the foundation to their Random Date argument.
I concur that if Cycle Date generating, or simply Cycle Analysis did not produce highly reliable information, that they would not be any different in value than Random Date generating. Unless you get highly accurate information from any approach, it too will fall under the same uselessness as Random dates for swing forecasting. However, Cycle Analysis is real and it works. There are even various approaches to extracting this information, each with their plus and minuses. But the bottom line is, real and solid Cycle Analysis or Cycle Date generation exists and is far and away better than Random Date generating. They simply do not fall under the same column of worth in market forecasting.
With Random Date generating producing your trading signals, you can certainly expect random results. But with a fixed method to Cycle Analysis that proves to have better than a 75-80% accuracy to patterning after actual market patterns or swings, although falling short of 100% accuracy it will help produce more consistent results when properly used in ones trading plan. In addition, by utilizing other techniques in conjunction with Cycle Analysis one can minimize any negative effect of the small percentage of error that exists between actual statistical accuracy and 100%.
Note: About the Author
Rick J. Ratchford is President of ProfitMax Trading Inc. He is a full-time commodity trader for his own account as well as assisting other traders. He has been a computer programmer for more than 20 years and a trader since 1990.
Â