Machine Learning and AI Impacts on the Financial Markets

How do investors find and capitalize on alpha generating investment ideas in the stock market when competing with machines that are trading on data in microseconds?

A Barclay’s Hedge survey found that the majority of hedge fund managers use Machine learning in their investment strategies to “formulate investment ideas and build portfolios informed by data analysis that the human brain could never hope to accomplish.” [1] It is pivotal to understand how machine learning is impacting the financial markets as this megatrend is controlling large movements in security prices and making alpha more difficult to find than ever before.

Almost a year after completing my undergraduate studies, I started working as a long short equity investor at Citadel, a Chicago based hedge fund. Citadel’s aspired competitive advantage comes from being able to use technology, machine learning, and human capital to amass all available public data, and then to “extract the most relevant pieces of information and then be able to make the right interpretation of that data.”[2] Machine learning is used in many ways at Citadel, predominantly to assist in predicting metrics like same store sales, identifying risks in the portfolio, and generating ideas.[3]

However, machine learning at Citadel is used to assist human judgment, rather than to substitute it. CEO Ken Griffin discusses this idea at the GS Systematic Investing Conference where he states that balancing the influx of new technology with the “power of the human mind” is key for better decision-making in the financial markets.[4]

Other funds like the $96 billion AUM Man Group have close to half of their investments in technology vehicles that find patterns and execute buy and sell orders independently.[5] Sandy Rattray, CIO of Man Group, stated in an interview with CNBC that Man Group will steadily increase its investments in models that are not directed by humans but learn from data on its own.[6] One example is “instructing computers to listen to multiple earnings calls at a time, whereas the individual portfolio manager can only listen to one.”[7]  However, Man Group concedes that humans remain better at data interpretation at this point in time. [8]

As machine learning improves, Citadel’s reliance on human capital will be challenged. Machines will be able to analyze data, interpret data, and trade on ideas quicker than humans. Citadel already has senior machine learning employees working to stay at the forefront of technology. The company has been hiring specialists like Pradeep Natarajan, a senior quantitative researcher in machine learning, joined from Amazon, where he was a senior research scientist.[9]

However, Griffin believes that for now machine learning has its limitation and investing free of human judgment is a long time out.[10] Citadel prefers to invest in ways to make investors more productive. The company also invests in ways to identify why stocks are moving the way they are moving to enable investors to capitalize on the crowding that comes about from machine learning insights.[11]

I would recommend Citadel to test out strategies that are solely machine learning for industries that are more cyclical, more mature, and less on high growth stocks where valuations are more susceptible to changes in perception.

Along with the amazing insights that machine learning allows, there are negative considerations as well. More reliance on technology and engineering enables bigger funds to have an advantage over the average investor thus making the markets perhaps less fair.

Another consideration is that machine learning creates interesting insights and patterns but will also lead to crowding, (when many people are buying the same stocks). This can potentially have exacerbating effects and large swings for stocks causing large amounts of volatility. The degree to which these machine learning algorithms exacerbate selloffs has been “hotly contested, with some managers arguing they are too small to spur such an impact.”[12]

 

(Word Count 795)

 

 

 

 

 

[1] BarclayHedge. (2018). Majority of Hedge Fund Pros Use AI/Machine Learning in Investment Strategies – BarclayHedge. [online] Available at: https://www.barclayhedge.com/majority-of-hedge-fund-pros-use-ai-machine-learning-in-investment-strategies/ [Accessed 13 Nov. 2018].

 

[2] Citadel. (2018). Fundamental Analysis at Citadel – Inside the Job with Tio Charbaghi – Citadel. [online] Available at: https://www.citadel.com/news/fundamental-analysis-citadel-inside-job-tio-charbaghi/ [Accessed 13 Nov. 2018].

[3] YouTube. (2018). Billionaire Kenneth Griffin: Investing, A.I and Career Advice (2017). [online] Available at: https://www.youtube.com/watch?v=nO0wQEGDF7s [Accessed 13 Nov. 2018].

[4] Goldman Sachs. (2018). Goldman Sachs | Talks at GS – Ken Griffin: How Talent and Technology Are Shaping the Markets. [online] Available at: http://click.gs.com/l8md [Accessed 13 Nov. 2018].

[5] CNBC. (2018). One of the world’s largest hedge funds is now letting computers trade completely on their own. [online] Available at: https://www.cnbc.com/2017/09/28/man-group-one-of-worlds-largest-funds-moves-into-machine-learning.html [Accessed 13 Nov. 2018].

[6] Ibid.

[7] Ibid.

[8] Ibid.

[9] Microsoft, C., Clarke, P., Clarke, P., Butcher, S. and Butcher, S. (2018). Citadel has just hired a new head of artificial intelligence from Microsoft. [online] eFinancialCareers. Available at: https://news.efinancialcareers.com/us-en/288118/citadel-has-just-hired-a-new-head-of-artificial-intelligence-from-microsoft [Accessed 13 Nov. 2018].

 

[10] YouTube. (2018). Billionaire Kenneth Griffin: Investing, A.I and Career Advice (2017). [online] Available at: https://www.youtube.com/watch?v=nO0wQEGDF7s [Accessed 13 Nov. 2018].

[11] Ibid.

[12] Bloomberg.com. (2018). Bloomberg – Are you a robot?. [online] Available at: https://www.bloomberg.com/news/articles/2018-03-12/robot-takeover-stalls-in-worst-slump-for-ai-funds-on-record [Accessed 13 Nov. 2018].

Previous:

“News”, “News”, “Fake News”: Can Machine Learning Help Identify Fake News on Facebook?

Next:

Invisalign: A Pioneer of Mass Customization through 3-D Printing

6 thoughts on “Machine Learning and AI Impacts on the Financial Markets

  1. M – Very nice article. I wonder whether there is a ceiling to how much machine learning can control decision making in financial markets independent of human judgement. It reminds me of a recent talk I attended at HBS in which Jamie Dimon and Seth Klarman were asked about the rise of passive investing. Both acknowledged that the asset class has grown tremendously and has its advantages, but each stated that there is a limit to how large this passive segment can be. The reason is that if no humans are monitoring the market and everything is passive or machine based, eventually alpha-generating opportunities will arise, in which case it is worthwhile for a human to actively do his / her homework analyzing securities. I believe this same idea will hold with machine learning in which there is ultimately a limit to how many funds can run algorithms that trade independent of human judgement (outside of writing the algorithm). It will be interesting to see how this trend unfolds.

  2. I’m worried about crowding too! Machines are dumb, and even worse, they’re consistently dumb in the same ways. I’m pretty sure they’ll all figure out the same strategies at the same time. If machines one day become super good at detecting alpha, maybe one day they should run our businesses too. I wouldn’t mind working for an AI if it’s superior to me.

  3. Quantitative hedge funds have been one of the fastest growing strategies in recent times; 56% of hedge fund respondents to a BarclayHedge poll say they use artificial intelligence or machine learning in their investment process [2]. I agree with you that they add volatility to markets, and despite their ability to process more data rapidly they are not always right. In March 2018, the AI Index tracking performance of quant funds fell 7.3%, vs a 2.4% drop in the overall Hedge Fund Research Index aggregating all strategies [1]. Established hedge fund funds are increasingly looking to hire investment professionals with quantitative PhD backgrounds, but enforcing more stringent risk limits that limit portfolio managers’ ability to generate alpha, resulting in these portfolio managers being fired and hired within a pool of quant funds.

    [1] Burger, Dani. 2018. “Bloomberg – Are You A Robot?”. Bloomberg.Com. https://www.bloomberg.com/news/articles/2018-03-12/robot-takeover-stalls-in-worst-slump-for-ai-funds-on-record.
    [2] Whyte, Amy. 2018. “More Hedge Funds Using AI, Machine Learning”. Institutional Investor. https://www.institutionalinvestor.com/article/b194hm1kjbvd37/More-Hedge-Funds-Using-AI-Machine-Learning.

  4. This was a thought provoking read, thank you! I wonder if crowding will really become an issue. Even if all trading would be done by machines, wouldn’t the markets be self-correcting? If everyone/all machines want to buy the same stock, prices of that stock go up, and the related alpha will disappear. If I follow this logic the markets would become even less volatile than with human investors, because all investors are now fully rational

    1. I would agree with this sentiment. But what happens when the computers are wrong or when the macrobackdrop is wrong? Assuming all the machines are finding the same insights and buying the same stocks, when they are wrong the stock crashes. It inherently is creating bubbles for certain stocks.

  5. Very interesting essay on machine learning in the financial sector and definitely a great example of how AI complements human judgement vs. fully replacing it – thank you!

    I am especially interested in the question you are raising whether AI and machine learning will make the markets less fair for the average investor. While I can fully see your perspective, I am wondering whether there are also arguments for the opposite view. In the long-term, machine enabled investing might be more affordable than human-based services, therefore making the markets more fair. We see many examples where technological advancements, while at first expensive and only affordable for few, become more affordable with scale and give new opportunities to a population that was previously excluded from a certain service or at least disadvantaged (e.g. open education through the internet).

    Definitely a very interesting read, that provokes the consideration of many opportunities but also risks.

Leave a comment