Understanding the Effect of Artificial Intelligence and Machine Learning on Investing
9 min read

Understanding the Effect of Artificial Intelligence and Machine Learning on Investing

Understanding the Effect of Artificial Intelligence and Machine Learning on Investing

From the birth of statistics with the publishing of Bayes theorem in 1763 to Alan Turing’s learning machine in 1950 to the self-driving cars of today, machine learning and artificial intelligence have been integral parts of the human experience.  Although artificial intelligence has been in the works for decades, it was not until the late 90s that machine learning started to become a part of our everyday lives. With the advent of Google, machine learning and algorithms became an integral and irremovable part of day-to-day activities in the internet-age. The investing community has not been untouched by the influence of machine learning and artificial intelligence either. While the totality of the effect of machine learning and artificial intelligence on investing is near impossible to accurately measure, I believe that that the current impact of artificial intelligence on investing has already begun to take effect in small (and quickly growing) sophisticated investing circles and that the long-term effects – as this technology continues to develop – will change the entire way that we think about investing and markets.

First, it helps to define the difference between artificial intelligence and machine learning. Although I use artificial intelligence (AI) and machine learning (ML) somewhat interchangeably, there is a difference between the two terms. Artificial intelligence can loosely be broken up into three categories: Artificial Narrow Intelligence (Weak AI), Artificial General Intelligence (Strong AI) and Artificial Superintelligence (The villains in action movies).  In the simplest form, machine learning is a division of artificial intelligence that is currently the most developed and the most commercialized branch of AI. Machine learning falls into the lowest tier of AI, as a weak form of artificial intelligence. Weak AI can be used to do only one function (identify your face on Facebook photos, self-driving cars, trading algorithms, etc.). Strong (general) AI, can perform several broad functions, can pass for a human (pass a turing test), and can potentially display human-level consciousness.

It's also useful to make a distinction between how this new AI/ML enabled trading existing quantitative trading methods. Motoyuki Sato of Man Group Japan, Ltd. sums it up best,

“While they both rely on computers to make investment decisions; AI programs go a step further than quant software by attempting to improve themselves over time – mimicking the human brain’s capacity for learning.”

While, high-frequency trading firms (HFT) utilize automated and algorithmic trading strategies, my general impression of current HFT techniques is that they follow more of a tracking of large block trades in which they can capitalize on price swings, rather than cross-examining data to develop their own ideas of correlations between seemingly unrelated factors.

An example of “develop[ing] their own ideas” comes from Renaissance Technologies where, “By studying cloud cover data, they [Medallion Fund traders] found a correlation between sunny days and rising markets from New York to Tokyo. “It turns out that when it’s cloudy in Paris, the French market is less likely to go up than when it’s sunny in Paris,” he [Peter Brown] said.”

In the current investing ecosystem, it is tough to get an accurate read of the effect of machine learning and artificial intelligence on the investing world. Being that the most sophisticated algorithms are employed by hedge funds, who pride themselves on secrecy, most of the commentary is educated speculation from journalists and industry insiders.

While limited details are available, Bloomberg published a lengthy piece on “The Medallion Fund”, which is an employee only fund for those at Renaissance Technologies that utilizes several artificial intelligence technologies. Per Bloomberg, Renaissance’s Medallion Fund has, “generated average annual returns of 71.8 percent, before fees, from 1994 through mid-2014. That’s more than seven times the average annual gain for the S&P 500.” Renaissance is interested in hiring scientists, a belief rooted in the very founding of the firm. Much of the firm’s original hires, and still a strong base of RenTech employees, hail from the computing giant, IBM. Much of the original Renaissance team was working on problems including “speech recognition and machine translation […] The group’s work eventually paved the way for Google Translate and Apple’s Siri.”  While the exact extent of Renaissance’s use of AI/ML cannot be pinpointed, it is evident that they employ more AI technology than the average firm and in turn, book returns as high as 7x above the S&P 500.

Another example of an established hedge fund turning to machine learning and artificial intelligence technologies comes from Bridgewater Associates (largest hedge fund in terms of assets under management). In February of 2015, Bridgewater shared:

“will start a new, artificial intelligence unit next month with about half a dozen people, according to a person with knowledge of the matter. The team will report to David Ferrucci, who joined Bridgewater at the end of 2012 after leading the International Business Machines (IBM) Corp. engineers that developed Watson, the computer that beat human players on the television quiz show Jeopardy! […] The programs will learn as markets change and adapt to new information, as opposed to those that follow static instructions.”

Ray Dalio has cited that a lot of the success of Bridgewater comes from (among many other things) the diverse history of its associates and in his mind, their extensive experience in studying markets. Ray Dalio felt that his knowledge on the effect of interconnected macroeconomic events, allowed him to make trades from information that others might ignore. With the rise of machine learning, artificial intelligence, and the mass amounts of data available for consumption, he is now able to run that process on a machine/algorithm that can learn at rates exponentially faster than he could have ever dreamed for himself. As noted above when looking at the effect of the weather in Paris and stock trading, machines can process and find correlation between events that humans would never have had the intuition to connect or resources to study. What may have taken Ray Dalio years of studying history, markets, and two degrees from Harvard, could be learned without human bias by a super computer in a mere fraction of the time. I won’t attempt to put an exact estimate on the speed at which an artificially intelligent computer program could become more intelligent than a human, but as writer, Tim Urban, points out in a hypothetical (and potentially exaggerated) situation is:

“It takes decades for the first AI system to reach low-level general intelligence, but it finally happens. A computer is able to understand the world around it as well as a human four-year-old. Suddenly, within an hour of hitting that milestone, the system pumps out the grand theory of physics that unifies general relativity and quantum mechanics, something no human has been able to definitively do. 90 minutes after that, the AI has become an ASI [Artificial Superintelligence], 170,000 times more intelligent than a human.”

Returning to the topic of investments, I find it hard for any individual to compete with a machine of that caliber. As a testament to how effective and profitable this method of trading is, there are also several other prominent hedge funds getting involved in the science or whose founders have roots in artificial intelligence.

Other examples of large hedge funds include firms like:

  • Two Sigma Investments. One of the Two Sigma founders, David Siegel, earned a PhD in computer science from MIT where he studied in its artificial intelligence lab. David Siegel was also recently hailed algorithms saying, “that common biases against algorithms are frequently irrational, given humans’ own susceptibility to making mistakes. And that the sooner we come to terms with the power and potential benefits of algorithms, he writes, the better off we’ll be.”
  • D. E. Shaw & Company – Founded by David E. Shaw in 1988, after he completed a PhD in computer science at Stanford, joined the Columbia University computer science department and then eventually left to pursue interests in computational finance.
  • Simplex Asset Management – One fund run by Yoshinori Nomura, who has a master’s degree in Physics, also uses AI technology to trade. Yoshinori has “three patents to his name, including an algorithm designed to predict hit songs in pop-obsessed Japan.”
  • Point72 Asset Management – Steven Cohen’s fund, is also "building out their big data investing division with several hires that will increase the firms quantitative reach." While the article does not directly cite artificial intelligence/machine learning, it can be reasonably assumed. Steven Cohen has also funded a crowdsourced algorithmic trading fund, Quantopian.

Start-ups enabling artificial intelligence/machine learning investing:

  • Kensho Technologies – “Kensho is out to shake-up the financial data industry (now dominated by Bloomberg and Thomson Reuters) by giving the masses the type of complex, quantitative computer power currently used by a few top hedge funds like Bridgewater Associates, D. E. Shaw and Renaissance Technologies”. Goldman Sachs, Google, and the CIA are all investors (among others).  
  • Rebellion Research – “We use our proprietary artificial intelligence based Machine Learning powered system to operate multiple strategies for our clients. Our technology was once available only to very high net worth investors through our hedge fund, but is now offered to you through managed brokerage accounts at Interactive Brokers.”
  • Sentient – “Led by senior executives from both Wall Street and Silicon Valley, Sentient Investment Management is developing and applying proprietary quantitative trading and investment strategies built using the Sentient Technologies distributed artificial intelligence system, the most powerful system of its kind.”  Per Bloomberg, Sentient is also working with Highbridge Capital Management to “develop investing strategies using artificial intelligence.”
  • Aidyia – “Founded in Hong Kong in 2011 by computer scientists and financial market veterans. We deploy cutting edge artificial general intelligence (AGI) technology to identify patterns and predict price movements.”
  • Other notable AI-enabled algorithmic trading startups include: Clone Algo, Alpaca, Walnut Algorithms, and Binatix.

All the research above shows that large and successful hedge funds are hiring the individuals needed to capitalize on AI-enabled trading strategies. Additionally, as Goldman Sachs has shown with their significant investment in the MIT/Harvard start-up, Kensho, there is no shortage of startups with strong AI/ML capabilities, that are ripe for investment or whose proprietary software can be acquired by large financial institutions.  

In an opinion piece from a Spanish professor, Enrique Dans, he makes the point that as more and more companies adopt AI-enabled technology, there will be a new “Darwinian event” that puts Ai-enabled companies at a level of competitiveness significantly above companies that do not utilize AI.   I believe that this era is already upon us. The hedge funds cited above who use or are seriously developing their AI-capabilities are all behemoths in the field who have provided their investors with above average returns for many years. In line with Professor Dans’s point above, I believe that in the short term or until a critical mass of investors (hedge funds, trading desks, some level of individual investor) have access to AI-technology for trading, there will be a tough to reconcile advantage for firms who utilize this new technology.

In the long-term, this poses even graver consequences for traders and investors. The inevitable consequence of Ai technology on investing is purely (in the sense that 6% unemployment = full employment) efficient markets. As more and more trading is done by institutional investors (investors with the capabilities to afford advanced AI-technology), it will become near impossible for a non-institutional or fundamental trader to make any money in the markets (assuming he could now to begin with).

One of the first hypothesis I had as to how the widespread implementation of artificial intelligence would affect trading, was that it would force us to redefine the efficient market hypothesis and, a level deeper, how we think about efficient market. Tshilidzi Marwala is a PhD out of South Africa who, is one of the only individuals who have delved into this topic. His specific research interests include, “the theory and application of computational intelligence to engineering, computer science, finance, economics, social science and medicine.”  

While, I have not yet been able to read his entire book on “Impact of Artificial Intelligence on Economic Theory”, I read the abstract summary of some of his work. In short, Dr. Marwala, along with several Quora writers  have agreed that the rise of artificial intelligence technologies will have a profound impact on economic theory as we know it today.  As information is more readily and accurately shared through computers, and as intelligent machines are able to initiate trades free of human delay and bias, markets will become more efficient. More of a theory (and ironic consequence) I think would come hand in hand with this increase in efficiency, would be a decreased in the need for active money managers as it becomes evident there is very little alpha to be made in the markets.

Another affect this will have on the investment process is that it will continue to change the type of individuals that money managers look to hire. As the cost and ability to achieve accurate predictions deceases while the accuracy increases, investment firms will have less and less of a desire to hire professionals from the finance realm (as Renaissance Technologies has already demonstrated). Instead, with accurate predictions costing less than ever before, firms should be more interested in bringing in individuals who are specialists in diverse fields to help investors identify large macro trends to look for or to be able to provide commentary on various industries beyond the role of an internal bank analyst.  For example, investors could enter into short-term consulting contracts with geologists, climate experts, and other obscure academic studies to pick their brains on freak occurrences in their industries and then back test their theories against the history of the markets to search for correlations.

All in all, continuous advancements in computing power, sophistication, and the development of computer intelligence, will have a paramount impact on economic theory and how we think about investing. As we look towards the future of the financial industry, it is a good time to study computer science. My prediction is that the start-up companies involved in developing artificially intelligent trading algorithms will receive a hotbed of funding over the next few years and will upset institutional money managers in profound ways until they can either develop algorithms of their own or make significant acquisitions. Additionally, beyond the efficient market hypothesis, we believe there are several economic theories that we will be forced to rethink. As information becomes more readily available and less flawed, we will find ourselves in a very objective and efficient world of investing with limited returns for professional investors.