MARKET IN CHAOS

Economics, in contrast with the ‘pure’ sciences, is a fairly modern subject with its first major ideas emerging in the late 18th century. Economists since then have for over 2 centuries tried to come up with a basic set of laws that could predict the complex system of exchange of goods and services mathematically, producing results that are fairly accurate. But then, the question arose: is it possible to model human behavior the way it is possible to model other systems in science, say, the motion of objects?
Using Newton’s Laws, it is possible to predict the motion of any macroscopic object under the influence of a force. But what of man? Could we accurately predict how he would react to certain laws or policies? And if we could would it then be possible to model the ideal deterministic economic system? One where everyone was well off and one where everything including stock markets could be predicted with reasonable precision?
As of today, there is no system yet discovered that can guarantee a good return, or for that matter, even a zero return (no profit, no loss) in the stock market. And this, despite petabytes of data from stock markets around the world with records of every transaction that has ever taken place over centuries. Given enough data to analyze, software today can very accurately determine spam emails, guess the next word you’re going to use while texting, or even recommend which video or song you’d like to watch or listen to next. So, why doesn’t this machine learning and artificial intelligence and data sciences find itself humbled and ridiculously useless in front of stock market data? The answer lies in chaos.
Ever since Adam Smith’s publication of The Wealth of Nations in 1776, economists have tried to come up with laws that closely resemble the ones in Physics. And this trend has continued even to the present day. Most economic theories proposed are linear in nature: every action has a reaction and every economic event is causal in nature, that is, we can link the beginning to the end with a series of intermediate events each being caused by its previous.
Of late, this approach has met with considerable criticism. And it was this futile search for a possible marriage of economics to mathematically modelable laws that contributed majorly to what is now known today as chaos theory.
Before we try to understand what chaos theory states, let us ask a symmetrical question: why is it that physical laws have proven to be so successful? For a theory to be successful, it must be able to give a reasonably accurate output for a given input. But gaining an absolutely accurate input is impossible. For instance, Heisenberg’s Uncertainty Principle tells us that it is impossible to know the position and momentum of an object simultaneously. So, if a problem requires both the above values to calculate a third, one cannot provide the ideal inputs. But this is where the strength of physical laws lie. It can tolerate errors in calculations. The magnitude of error in output is directly proportional to the inputs’.
This, however, isn’t the case with some systems, such as the weather. Like stock markets, it is impossible to predict the weather. Even if we assumed the weather to be completely deterministic (i.e given all the initial conditions, it is possible to predict the future), it wouldn’t be possible to predict the weather accurately. A small error in the input values yields an extremely huge change in the final result. The magnitude of error of output is completely out of proportion to that of the input’s. In the words of mathematician Edward Lorenz, a pioneer of chaos theory, “Chaos is when the present determines the future but the approximate present does not approximately determine the future.”
Almost every economic theory in existence assumes man to be rational. Every theory assumes that man will take actions that will produce the maximum benefit to him, his family and everyone that he cares about. But this is often not the case. People, contrary to what we might think initially, are not completely rational beings. We often are prone to take decisions based on the kind of emotions that are triggered by the introduction of certain laws. Most other times, we simply act based on already existing ‘rules of thumb’. For instance, it was discovered that people face greater distress losing a certain amount of money than they derive happiness from gaining the same amount. These observations and discoveries led to creation of an entirely new branch of study known as Behavioral Economics.
So, even if we assumed that man was reasonably rational, our predictions of his behavior and choices after the introduction of a law might be fairly accurate. But it is impossible to be 100% correct with our predictions. And since economic systems have the ‘butterfly syndrome’, anything less than a 100% simply isn’t good enough. Even minute errors will magnify multiple times and give us results that are totally out of sync with that with the predicted ones.
For argument’s sake, let’s say that we did end up building a computer that could predict stock markets with 100% accuracy. And let’s say that the computer was bought by every person who deals in the markets. What would happen then?
As you may have guessed, the situation presents itself with a paradox. Stock markets are adversarial in nature. If there’s one person who’s gained millions in some part of the world, it’s extremely possible that someone else may have gone bankrupt in some other part. The net gain or loss of a stock market is zero. If someone gains in the market, someone loses an equal amount. It is impossible to build the ideal AI system in adversarial conditions. It is something like chess. If the rival happens to be more intelligent, it is more likely to lose.
So the next time your financial advisor asks you to invest in a mutual fund because it would give you ‘guaranteed returns’, you’d know better than to take his word for it. Quite like the uncertainty principle, we can never speak of 100% returns; only of probability of success from a transaction. And this probability cannot be a very encouraging number either.
It can be safely said that it is practically impossible to build the ideal stock market predictor. It is beyond our capabilities, at least for the time being, to predict the behavior of emotional beings such as humans. So, even though we can expect and wait for a Grand Unified Theory to explain everything there is to know in Physics, it might not be the case for economics. If there’s one thing that man might not be able to completely understand, it is man himself.

IIT News