07.21Stock Market Pseudoscience
by Jon Blumenfeld
Do big fund managers and Wall Street experts use pseudoscience to make trading decisions? If you have any mutual funds, insurance policies, or any other managed investment, chances are that the managers use a pseudoscientific principle known as technical analysis to decide when to buy and sell your investments.
Technical analysis is the principle of applying statistical analysis to the historical chart of the price of some security in order to predict future prices – without reference to any other information. It is not necessary to know what is happening in the world, what the supply and demand situation is, or even what security is being traded. All the data that is needed is in the chart. It is used by almost every Wall Street trader from time to time, even though many profess to know that it is pseudoscientific. It is recognized by the US regulatory authorities as one of the two basics types of trading – the other being fundamental analysis, which tries to predict future prices by placing an actual value on a security.
Investment literature is rife with advertisements from companies trying to sell technical systems. Many of them read like get-rich-quick schemes, promising guaranteed returns with very little risk. Ads in a recent issue of Futures Magazine1 have copy like “$2500 a Day. Quit your job w/my new S&P Trading System… 6 mo@$1995,” and “The big moves generate the big profits. Risk $500 to make $5000 and win on 8 out of 10 trades.” KeyPoint, a video workshop, promises that you will trade “profitably 88% of the time,” for only $95. Recurrence V, a computer program put out by Avco Financial Corp., of Greenwich, CT, promises to win 70% of the time, and Avco will refund your purchase price (several thousand dollars) if you follow the system for a year and fail to show a profit.
The question is, if these systems are so good, why are they for sale? The fact is that in commodities trading, a highly leveraged activity in which most small participants are using technical systems like the ones advertised in Futures, between 70% and 90% of individuals lose money2.
What is Technical Analysis?
There are basically three different types of market analysis – Fundamental Analysis, which uses information like supply and demand or economic figures, Technical Analysis, which uses only information found in historical prices, and Arbitrage (or relative value) which compares the value of different securities against each other.
The first technician was Richard Donchian, who started his Wall Street career in 1930 and started the first commodity fund in 1949. In 1957 he introduced his first trend following system. Trend following systems are supposed to tell traders when a market is going into a prolonged period of movement in one direction or the other. Donchian’s first system used moving averages to identify trends, and to this day many technical systems (including some of the most successful) are “trend followers.”
Today, technical systems are developed using complicated statistical analyses, and such fashionable “new” concepts as fractal geometry and chaos theory. Some systems attempt to model the markets on a known physical system, or on the Fibbonacci sequence (1,1,2,3,5,8, etc… where f(n) = f(n-1) + f(n-2)). Some systems try to capture market psychology by identifying so-called “high volume” prices where lots of trading has taken place.
Why is it pseudoscience?
You might, at this point, be saying to yourself “Well, okay, it all sounds a little funny, but why is it pseudoscience? Maybe these guys are right.” Maybe. Maybe not.
First of all, is there anything wrong with trying to formalize a system based only on data and not on any theoretical principles? This is, after all, what we’re saying: some technical systems are – a formalization of prices based only on data gathered in the market, with no external theorizing. Are there any examples of scientific principles being established this way? Yes, there are, and one of the most famous is the development of the laws of planetary motion by Johannes Kepler (1571 – 1630). Using only data on planetary positions, Kepler developed his three laws of planetary motion. Only later did Sir Isaac Newton show that the theory of gravitation led directly to Kepler’s laws.
If Kepler can do it, why not today’s technical trader? There are some important reasons. The system Kepler was looking at was very simple and was periodic, repeating itself over and over, in a predictable way. Kepler also had very few variables – really only one independent variable, the diameter of the planet’s orbit. Despite this, it still took a leap of imagination for him, because he had to give up the Copernican idea of circular orbits in favor of elliptical orbits. This reduced his error in predicting the positions of the planets from 5 degrees (which many might have accepted, considering the difficulty in making observations without telescopes) down to 10 minutes, an error of just 0.05%.
Imagine if a modern-day Kepler tried to analyze a market. Could he write an equation to describe the observed data? Yes, he could. There are several mathematical tools for creating equations out of a set of points, such as Fourier analysis. Unfortunately, for these tools to be useful in predicting the future, the functions they describe must be periodic, like the orbit of a planet around the sun. Establishing this periodicity for markets has never been done. In stark contrast to the regular motion of the planets, the stock market is a chaotic system. In addition, while Kepler knew what his variables were, the technical system designer doesn’t even know how to quantify the variables underlying the market. Those who have tried to apply Fourier analysis to markets have found this out to their cost.
This type of analysis, commonly called curve-fitting, leads us to one of the main reasons why technical analysis is pseudoscience. One can always describe a curve through a set of points, even random points, to any arbitrary degree of accuracy. This, however, will not predict the future behavior of the curve unless it actually describes a physical system. All evidence points to the fact that the stock market is a truly chaotic and random system, with minor trends and cycles obscured in the chaos. Using curves to predict the future of the market is like trying to predict coin tosses from a previous sequence of heads and tails.
Some analysts claim to have found the physical system underlying the market, but if this were true they could make millions in trading, and by selling the secret they make the knowledge worthless. Furthermore, if there are fifty different analysts claiming that the market behaves like fifty different physical systems, can they all be right? Can more than one of them be right?
Why does it work sometimes?
Why, then, do these systems seem to work sometimes, some of them so much so that just about every market professional uses them at least sporadically? How do fund managers who use these systems to the exclusion of all other forms of analysis stay in business?
The most important reason why these systems seem to succeed is that the best “technical” traders are really very good “fundamental” traders. There are at least two areas where fundamental decisions are hidden in technical trading systems. First, most technicians believe that no system works in all markets. The decision of when to switch from one system to another is “highly subjective,” as famous technician Richard Dennis said at a recent conference (MFA Forum ‘97). Second, decisions about how much of an investor’s resources should be committed to a trade, and when to cut losses are often subjective as well. These subjective decisions are made by feel, but are really based on the fundamental external situation surrounding the market.
Another important reason is the “misinterpretation of statistical significance” – also known as luck. It’s like the old ESP test where a thousand researchers are sent out into the world, each to test the ability of one subject to control the flip of a coin. Subjects are told to concentrate on heads. After the first flip, five hundred potential subjects are eliminated. After the second flip, another two hundred fifty fail. After nine flips there are one or two subjects who have successfully tossed heads every time and are considered to have psychokinetic powers. There are thousands of technical systems out there, and they only have to be right a little better than 50% of the time to be successful. How many of them will succeed for years just by luck?
In addition, statistics on the success or failure of technical systems are biased toward winners. It may be true that nearly all technical traders with a three-year track record are winners, but this only means that systems that fail are abandoned early, leaving only winning systems for the statisticians.
Widely available technicals tend to become self-fulfilling prophecies, with thousands of traders at least aware of the same set of levels. Traders don’t care whether the system makes realistic predictions or whether their own behavior validates pseudoscience. They just want the best price every time.
Finally, it must be grudgingly admitted that there are some cycles in the markets. Some commodities have “seasonals,” which are cycles that reflect different patterns of consumption in different seasons of the year. These seasonals are so well understood, though, that they can no longer be of use to technical systems. The market simply prices that information in too quickly.
So how well do these technical systems work? We’ve already seen that on a daily basis, with thousands of traders watching the same levels, they can be too powerful to ignore. The general rule of thumb seems to be, the shorter the time frame and the more widely disseminated the system, the better chance it has of being successful.
In the case of a technical system which truly captures something that no one else understands, wide dissemination tends to destroy the system. For instance, if there was a seasonal effect in soybeans that only I knew about, I could make money out of it, but if the great mass of bean traders found out about it, the price would quickly move to reflect the previously unseen effect.
Wall Street is full of traders who call technical analysis “tea-leaf reading” and “astrology.” That doesn’t stop them from checking the technicals every morning. There are plenty of traders who have made careers on technical analysis, but unfortunately many of them have made those careers by selling their worthless systems to individuals only too willing to take a chance on hitting the big jackpot. Some have succeeded through luck, many have failed, and the best of them have brought hidden fundamental analysis into their supposedly rigidly technical trading regimen. Maybe someday someone will discover the great system underlying the markets, but until then, keep a careful eye on your money.
1 Futures Magazine, June 1997
2 ‘Managed Futures Accounts’, Chicago Mercantile Exchange, 1996
3 ‘How to Spot Profitable Timing Signals’, Commodity Price Charts, 1990
The following is from the Letters To the Editor concerning this article
Stock Market Pseudoscience
The only thing that this article proved to be pseudoscientific was the article itself. Statistically how many errors are needed in an article before it is classified as being unworthy of publication. I will not pick on the errors regarding the technical analysis as it would just be claimed to be bias in favor of TA and would frankly take up most of the day. So let’s pick on Mr. Blumenfeld’s use of “misinterpretation of statistical significance.” Having taken 1000 subjects and tossed a coin with the object of obtaining heads by telekinetic power, we eliminate those who achieve tails. “After nine flips there are one or two subjects who have successfully tossed heads every time and are considered to have psychokinetic powers.”
Who has considered them to have these powers, the author? No rational scientist would. The fact that mathematically it fits does not have any relevance to the conclusion. This was typical of the approach :
1. Have a theory
2. Find some scientific or statistical commentary which gives the impression of research regardless of its worth.
3. Apply that commentary as if to show that it proved the theory.
Pseudoscience at its best. Virtually all the conclusions drawn in the article are incorrect.
All that I will say about myself is that I do not sell systems or books – I trade using technical analysis and have done so for 25 years. The profits prove it works and they have been checked for randomness.
I congratulate the author on his 25 year track record of using technical analysis. It should be noted that the article never said that this could not be done, but that there are alternative explanations to the technical analysis hypothesis, and while randomness and luck is one of them, it is by no means the most important. In my opinion, it is the use of other, non-technical information, either consciously or unconsciously, that is responsible for most successful, long-term technical trading.
The article is called pseudoscientific, and we are told that it would take up most of the day to enumerate the errors contained in it, but the only example cited is completely misinterpreted and turned on its head. No rational scientist would consider my coin flipper to have psychokinetic powers? This is exactly my point – the story was intended to show how pseudoscientists often misinterpret lucky events.
Where is the three-step process described in the letter used in the article? If “virtually all the conclusions in the article are incorrect,” then name one and refute it. Accusations fly in this letter, such as that we will dismiss it by claiming bias, but we are not given the chance because not a single fact (other than the author’s biographical information) is provided. Who’s being pseudoscientific here?
– Jon Blumenfeld
The above (Stock Market Pseudoscience by Jon Blumenfeld) was well written, and I enjoyed reading it. It was true what was written, but let me add one caveat. The writer has ignored one fact that has been too compelling for those of us that have delved into the matter, and that is the trading of W.D. Gann. I would encourage you to take an objective look at what I have written on my personal website, and entertain the idea that there really is an order that can be useful in trading, but that no man has been able to find and implement other than Gann, and no man has been able to determine what Gann did enough to be useful, except for the ways that you describe in your writing. When you look at the website, I encourage you to take what I have written objectively. Don’t let my poor writing style hamper my point that I am trying to make. Don’t dismiss what I am trying to say because I come across as being a kook. My website address is http://pages.prodigy.net/davebarry
Thank you very much for your response to my article. I have spent a little time on your website, but I must say that nothing you have said there convinces me that Gann was anything other than the standard technical analyst. I don’t know how he traded, or how to explain his high rate of success, but I believe that if a close investigation were possible, we’d find the same sorts of things I wrote in my article. I have seen many traders with high rates of success, but further investigation usually tells us that there was something more than pure technical trading going on. Of course, the possibility exists that Gann really found THE system to describe the market, but now you say that we don’t know exactly what he did or how he did it. It seems almost like a mystical Holy Grail, with Gann providing us just not quite enough information to reach it.
If you have more concrete information about Gann, or can direct me to something, I’d appreciate it. I certainly won’t dismiss you out of hand, and I’m always interested in this fascinating subject.