This type of article is one of the most fun for me to write because it’s really just a romp through the imagination. Since the 1990’s, I have made a hobby out of exploring new and varied ideas for analyzing the markets, frontiernews and this is a great opportunity to dust off some of my old notes, publish some of those ideas and perhaps get some feedback on them. I’m also looking forward to using some of the following concepts in my ongoing research work on FOREX price behavior. So put on your “what if…” hats and let’s get started!
Market Models – Old & New
Most traders are familiar with the two basic schools of market analysis that we call Fundamental Analysis and Technical Analysis. In the 1970’s, members of the academic community proposed a new model of the market known as the “Efficient Market Hypothesis”. localletter This is more commonly known as the “Random Walk Theory” and basically said that the first two schools of thought were both wasting their time. In response to the Random Walk Model, other academics put forth an even newer theory of how markets work called “Behavioral Finance”. These are all examples of comprehensive explanations of what factors drive market prices. Here’s a brief summary of market models, some of which are only in their infancy:
Fundamental: Market prices are driven by tangible events and conditions in the real world, such as earnings, sales, management, natural disasters, weather, newspoke economic conditions, geopolitical tensions and so forth.
Technical: Market prices are driven by what prices have done in the past. As traders observe these past and present price movements, their expectations about future prices lead to feelings of greed and fear which in turn create buying and selling pressures. topicals
Random Walk: Current market prices are efficient reflections of all known fundamental and technical information, so we can discern nothing about future price movements. The factors that cause future price movement will be so varied that such movements can only be random in nature. tbadaily
Behavioral Finance: Prices are driven by human psychology which is not always rational. Traders may base expectations about price movements, risk and reward on erroneous reasoning, thus causing prices to behave in non-random ways. Bubbles and crashes are classic examples of this. kulfiy
Chaos Theory: Market prices are part of a non-linear dynamic system in which outputs are re-introduced back into the system as inputs, causing complex behavioral loops and very sensitive dependence on slight variations in conditions. pressmagazines
Fractal Geometry: Price patterns are recursively nested, meaning that a large pattern may be composed of several smaller similar or even identical patterns and so on through all time scales. Elliot Wave Theory is a classic example of this idea.
Scott’s Emergent Property Model: I’ve discussed this one in more detail in other articles, but the idea is basically that identifiable properties of price behavior emerge from the combination of unique individual trading styles of the current market participants. An analogy would be how a person’s personality emerges from the combination of individual neurons in their brain. This price behavior changes gradually over time in an evolutionary way in the same way that the behavior of an organism changes over time due to both internal changes in its makeup and external pressures from its environment.
My apologies if I have neglected or grossly mis-represented any of the various ways of explaining what makes the market tick.
Mechanical Trading Systems
Another subject that often grabs my interest is the design of mechanical trading systems based on money management rules. Some examples of such systems are “buy and hold”, “dollar cost averaging”, Robert Lichello’s “Automatic Investment Management” (AIM) and any other systems that try to take the emotion out of trading through the application of a rule-based system for buying and selling. Systems like this differ from other mechanical systems in that the rules are based entirely on money management variables such as cash on hand, average cost per unit, total portfolio value and current position value.
These kinds of mechanical systems were generally designed for the securities markets however, not for the futures or FOREX markets where the cash management situation is much different. In FOREX, unlike the securities markets, we are not taking some currency out of an account that we own and exchanging that currency for some security (i.e. dollars for stocks). FOREX involves simply putting down a margin deposit and then using that as the basis for borrowing some larger amount of a currency and exchanging it for another currency, making us long one currency and short another at all times. This entirely different structure of the FOREX market presents a new challenge to the design, use and understanding of the classic money-management based mechanical systems.
Artificial Intelligence & Artificial Life
What if we could design a neural net which could learn over time how to make consistent profitable buy and sell decisions based on FOREX chart data? Or if you prefer cellular automata (CA) we could create one of those in a multi-dimensional format in which symbols “swirl around”, colliding and combining in new ways creating emergent behaviors. If we rewarded profitable behaviors and punished unprofitable behaviors would this CA eventually learn to behave like a super FOREX trader? What’s that you say? Why not harness the power of evolution by using Genetic Programming? Ok, let’s create an environment full of trading programs that have to compete with each other to survive and reproduce offspring programs. After many generations of “nature – red in tooth and claw” we may end up with a group of very robust FOREX trading programs. They will have earned their place in this virtual world where the prime law is “survival of the fittest.”
These are all examples of how the ideas popularly known as AI and A-Life might be applied to FOREX trading. Using programming languages such as LISP, I think it would be interesting to use neural nets, cellular automata, genetic programming environments and other techniques to create rudimentary trading programs. These programs would be exposed to many sets of market data (probably intraday charts) and over time would “learn” or “evolve” into expert trading systems. This isn’t science fiction, but it’s at the frontiers of cognitive science.
Computer Modeling of Dynamic Systems
What if the markets are deterministic in some ways? In other words, I’m wondering if there is a kind of “physics” behind price movements that is ultimately subject to complex cause and effect relationships. After all, we know that cause and effect relationships certainly exist. I decide to buy some FOREX currency pair, which causes me to place an order, which causes several offers to be hit in order to fill my order, which causes the bid and ask of the FOREX pair to rise, which causes some stop orders to be triggered, which causes more buying, which causes another price rise, which causes several news services to take notice, which causes several other people to buy, which causes the price to become so high that people start to take profits, which causes me to sell. Whew!
Keeping track of all these relationships where each event may be caused by and in turn causes several other events is a job for computer modeling. In fact, computer modeling or simulation is what we use to try to understand the behavior of any complex system that we can describe through a few simple rules.