Can an algorithm predict market movement?
Can algorithms help us correctly forecast future market conditions and movement? According to a fascinating talk by Kevin Slavin, it certainly is possible although the solution may be one of huge complexity.In today’s trading markets, algorithms are pervasive. Now, before we proceed any further, let’s clarify what an algorithm for those who do not already use them. An algorithm is simply a software program or part of a software program that acts in a certain way depending on a set of preconditions.
In computerized trading, software algorithms automatically execute trades in certain instances. However, much more common are algorithms that send trading “signals” to investors. For example, a forex trader might receive a signal to buy a certain currency depending on market conditions.
Algorithms monitor market conditions 24/7
Software algorithms resting on computers and devices around the world are constantly watching for “signs” from all corners of the globe in order to detect possible future market movement.
In most cases, these algorithms connect to the Internet and scan particular websites or they may have direct feeds from online data sources. In our electronic world, algorithms have come to dominate certain spheres of human activity and market trading may be near the top of the heap in this regard.
Because these algorithms automatically influence the markets to such a high degree, it is natural to wonder whether one could analyze their behavior to predict future market movement. Indeed, there are already algorithms out there that watch what other algorithms do in order to calculate their own response.
Algorithm upon algorithm
One of the tough obstacles to tackle in figuring a sort of all-encompassing algorithm that takes into account all the others is the hopelessly massive number of software programs working out there in the current marketplace.
We are not only referring here to the popular commercially available software programs, but also to myriad other types of software, some private and highly proprietary, positioned around the globe. Is it possible to analyze how these algorithms are working even if you are not able to examine personally the underlying code?
Slavin notes that it is possible to chart the way the market reacts under various conditions that can have an impact on the market’s direction. However, algorithms are so profuse that it can be daunting at times to even understand how they all work as a type of entity unto themselves.
Even the experts have a difficult time unraveling the impact that all the algorithms are having on the Wall Street ecosystem. Rather than humans as the driving force, to some extent, the algorithms are driving the humans at this point in history.
Of course, all these algorithms are the result of some human software designer or design team, but in some cases, their history is so complex that even the creators do not truly understand them anymore. For example, at the higher end, design teams of the most sophisticated software may go through many phases of development. The current version of the algorithm may have seen many core team members that have come and gone.
Then, there are the arcane algorithms that may be working for some secretive but high-powered investment firm. Taking into account all this automated trading or trading advice involves so many variables that it would seem impossible to untangle. However, one then wonders whether it would be possible to write an algorithm for that specific purpose!
Preparing for the future
Current trends indicate that that use of algorithms will only continue to grow. The younger generations grew up in a wired world unfamiliar to the older traders that currently dominate the marketplace. They are much more accustomed to trusting their high tech devices than most baby boomers.
For the texting generation, “old-fashioned” approaches will be less important because they will have a very different worldview. With the importance of speed in the current market, it will become increasingly important to react faster than is possible with human judgment. The traders of the future will depend on algorithms to make decisions in tiny fractions of a second.
For those enlightened traders who want to stay ahead of the crowd, the idea of an algorithm that can analyze all the other algorithms of the market may seem very appealing. The possibility of being able to predict within mere microseconds how billions or even trillions of dollars may move in just short periods of time based on algorithmic responses is overwhelming to think about.
Whether such a thing is possible though is beyond saying at this point. As Kevin Slavin notes, even most experts scratch their heads over how algorithms currently work in the trading world. How much more difficult will it be to gauge such activity in the future when algorithms will certainly be even more prominent than they are today.
Possibly an algorithm itself will be able to one day analyze all the variant streams of data to create a new algorithm capable of making forecasts across the market.
By the way, Jim Simons’ Medallion Fund’s returns are so spectacular that Jim is one of the richest people on the planet. Medallion Fund employs high frequency trading and exploits inefficiencies in the stock market. One strategy they use takes advantage of the inefficiencies in the execution of large transactions. One of their algorithms determines whether a very large order is executed and front runs it. Now, Jim, you might want to give (part of) your fortune, estimated at $12 billion, for your own happiness.