Turtle says things that sound sensible, but while guideworthy, you still have to ask, what is the foundation of truth to them? All complex models must say that they approximate the observed change, not that they know the cause as the hammer to the nail. One must look to a degree where their prejudice has led them to reject things out of hand.
Cycles are an observed phenomena. Because they are complex does not mean that they are not regular. Of course their noise is disconcerting, but in that it can be seen, it belies the dismissal of an underlying regularity.
We must eschew however any easy approach to cycles. Some patterns are insecapable and that is why I recommend the dormant-stock-revival patterns found by sweeps of volume/price lead changes, such as found by canadianmarketwatch.com Others factors and cycles are well known, such as the weather influence and clima(c)tic correlations. Others are harder to divine a parallel for, but are legendary, such as the 10 and 40 month cycles. It is hardly surprising that the average life of a man equals twice the most commonly found long cycle of markets and of solar change, that of 35.6 years.
The incredible truth, often scorned by markteers of note, is that value and fundamentals analysis, are really dependent on two things, an imperfect or not universally divined market, which is moot, and the assumption that the models of causation are perfectly divinable. In essence then, the forecasting method of causation based systems is saying, is that due to factors theoretical, the market will follow the trend assumed. What observations support this? In other words, how do they know that if a stock will pay a dividend equal to is earnings, that, if its P/E is low, the stock will eventually rise? Or that given that there will be a buying spree, the stock will rise. It seems logical, but what if, in certain sectors, let's say, no stock ever did that?! So the basis for all theory is that it must fit observation. So we are back to TA or price-trend-other-factor analysis again. Obviously the most comfortable system would utilize both to fit or reject the other. (So we do factor analysis to find the real factors, then TA perhaps)
It is easy to reject cyclical analysis as being too complex, or seemingly unproven, or not able to divine what you are told are "random systems". But hold a minute. What systems are random? If you accept fundamentals and causation, then can you accept randomness for the same system? Or to put it twicely complex, if many things vary "randomly" in the same system, is there not some degree of repeated functionality to their variance? If there is any interrelation to the variables, they have to start to oscillate together in patterns. Hamiltonian dynamics demands this (Chaos theory). The winds on the ocean give rise to waves of regularity, not one big wave, or billions of tiny confusing wavelets.
If systems of natural oscillation were not periodic and would not decompose to underlying repeated and monotonic patterns, then nature's chaos would be un-selfsustaining. Fourier's theory would not operate. In essence, Fourier, in saying that any function could be described as a composition of periodic sine functions, says the truth. Underlying all change, no matter what the discontinuity inherent, is a regular pattern of cycles. And these cycles, one could suppose are related to the observed output in a routine way. The caveat is that the discontinuous patterns are not overwhelming the changes of interest. I guess one has to trade outside the noise level.
All noise is, is a change of seeminly monotonic or complex discontinuities that are not predictable with conventional analysis. This does not mean at all that sudden change is noise. Or that high harmonics kicking in on a blue moon basis are noise either. Noise is strictly defined as effects outside the normalcy of the complex system as it can be routinely but expertly divined.
We must reflect that Fourier analysis is able to predict the supply of wheat without knowing the weather to a fairly good degree. I am not saying that anyone has the Chicago prices aced, as that has a few more factors than the raw supply of wheat, but they have managed to place cars in western Canada to save a few bucks here and there, with some regularity. It was also used, commbined with factor analysis to predict the 1974 recession with good precision, when no other causative factors seemed to indicate it.
Why then, if these methods are uncanny, do traders not use them with greater effect? One, it takes lots of data going back many years. Two, it takes some interpretation, once the spectrums are known, to do the prediction. Three, the math routines are not actually one slick package but require a lot of massaging and thought to feed into one another. And four, the method of factor analysis, that is the place it is highly recommended to start, requires lots of computer, and is an art before a science. Want to predict a pet stock in a pet sector? Prepare for weeks of reasearch, paying for data, upgrading your computer, learning a mountain of math, and doing some deep thinking too. In 1958 you just couldn't have done it. In 1978 your computer would have cost you 25,000, and the data would have taken you a month to load, never mind what you would have had to pay for it. In 1988 the data would be hard to load too, and find. Jump to 1998 and we are getting there, as the data was on the net, or for sale there. So now you are up to date.
IN 1955 a mathematician named Claude Shannon, whose theory of information allowed the electronic age to advance its communications to a high degree said, in defiance of the theory of the random walk, that the information inherent to predict the variations in any natural system were extant in the system's variations themselves. He set out to prove that one grand "unpredictable" system could thus vindicate this theory. He decided to predict the stock market. For six months he used his information theory to trade. His take was half a million dollars. His methodology has never been revealed. He did indicate that not everyone could do it at a once.
I don't think we need fear that.
EC<:-} |
