The age of AI or the third industrial revolution many fear will disrupt the industries and displace humans. Autonomous cars will put drivers out of jobs and drones will displace your mailman. Does it seem evident that AI will eventually displace mankind at investing?

It’s Not About Duplicating Here

Google’s AlphaGo and IBM’s Deep Blue defeated the respective world champions to showcase the power of AI. Both Go and Chess contain a set of rules. The set of actions and the possible rewards are well defined. Calculators once an impressive technology performs faster calculations than any human mind and without any errors. Basic mathematics is also a well-defined function. Then, should it be too surprising for us to witness machines outperforming humans at a task that humans have perfected but machines can just do it better?

A simple set of rules and some practice. That’s what it takes for a 16-year-old to learn driving. Humans don’t find it complicated to follow a simple set of rules — follow the traffic rules and don’t hit anybody. Then, it doesn’t seem too difficult to teach a machine to drive by itself. Plus, they only drink petrol. They are even quitting that nowadays. Automation is evident in the tasks humans have the capability to fully understand as machines will just bring in efficiency.

Making investment decisions, however, doesn’t come with a rulebook. If forecasting was that straightforward, risk would inherently become zero. The market dynamics might be easy to understand but well convoluted for humans to predict the movements. Past decade or so we have been borrowing tools from other sciences to get more systematic in making investment decisions. Portfolio Managers still haven’t found a magic potion to get returns with full certainty. The concept of certainty doesn’t exist in this science. Often more than not it’s about acting on correct information from a plethora of noise. The convoluted nature of market dynamics make it hard for a human brain to understand it completely. AI brings in the power of faster analysis and would, therefore, be a tool to deepen our understanding and not just duplicating our existing methods. There’s no money in information that everybody knows. The uncertainty in this business is what will keep humans in-charge.

Not the Last Step

The rule is that you make money on something that others haven’t recognized yet. Firms have long been using algorithms and statistical arbitrage to predict market movements. It works at first but eventually faces crowding effect and that strategy doesn’t provide a competitive advantage anymore. The market works in a way to eventually eliminate any risk-free advantage from mispricing and if investment products are correctly priced can you beat the market even with AI? The efficient market hypothesis argument could be applied here to ask how long would an automated investment strategy outperform the market? Isn’t it too prone to crowding?

Further, Machine learning techniques left to their own tend to overfit i.e. it finds nonsense patterns that will not hold up in a broader context. How often has the market repeated itself? Are the cause and effect relationships that machine learning depends upon reliable to understand and predict the market? AI too will require upgrades and further innovations. This is especially true for financial data as information in price movements a decade ago might not be relevant to today’s situation.

Does it mean there’s no potential for AI in the investment industry? The question could be answered with the image below. If the technology has its limitations, it doesn’t mean there’s no value to it. It will work until it doesn’t.

There is Going to be Some Shake-up

Wall Street today competes with Silicon Valley for technical talent. With the rising technical prowess of quantitative hedge funds its only reasonable to assume that knowing the Markovitz portfolio construction taught at business schools wouldn’t be sufficient for a portfolio manager to get hired. Hedge funds might not require analysts for basic decision making. Instead, the demand will be for engineers to automate basic tasks and for scientists and researchers to bring innovation in decision making and making it more systematic.

Moreover, technical prowess isn’t sufficient to run a business. With complex processes being used and basic functions being automated, human involvement would still be required in maintaining client relationships and developing business strategy. Data being the fuel for AI would be another challenge for the industry. With increasing use of strategies requiring quality data, human involvement will be required to acquire and attain competitive advantage in data. With prevailing issues on data and privacy regulations, human leadership and management will never take a back seat.

The views expressed above are not necessarily the views of Thalēs Trading Solutions or any of its affiliates (collectively, “Thalēs”). The information presented above is only for informational and educational purposes and is not an offer to sell or the solicitation of an offer to buy any securities or other instruments. Additionally, the above information is not intended to provide, and should not be relied upon for investment, accounting, legal or tax advice. Thalēs makes no representations, express or implied, regarding the accuracy or completeness of this information, and the reader accepts all risks in relying on the above information for any purpose whatsoever.

Print Article