• Ishaan Gvalani

Analysts v. Machines


For years now, Artificial Intelligence (AI) has been a buzzword in the finance industry. It has caused massive disruption in the sector, leading to several much-needed reforms. Along with this, it has enabled people to profit from financial activities in previously unimaginable ways. However, one question has remained constant: does AI have the ability to eventually supersede the human race? This article will explore how AI, through a process known as Machine Learning (ML) is beating humans in an industry of crucial importance: the stock market.

In light of new research by experts, it is now believed that analysts have traditionally been overly optimistic of firm’s earnings, and this bias has constantly changed over time with different stocks. The experts Jules H.van Binsbergen, Xiao Han, and Alejandro Lopez-Lira, have successfully developed a:

"machine-learning model to generate a statistically optimal and unbiased benchmark for earnings expectations”

By successfully predicting the behaviour of different stocks in response to positive and negative analyst reports, the researchers claim that their model can successfully achieve profitable trading strategies. Hence, when investors notice that analysts’ expectations are extremely pessimistic, they can sell their investments and vice versa.



Findings of the Paper: New Opportunities


To publish their paper, the researchers gathered information on firms’ balance sheets, macroeconomic variables, and analysts’ predictions. The paper found certain behavioural patterns of analysts’ that remained consistent over time: their biases usually increase in the forecast horizon (when earnings announcements are far away), but analysts tend to “revise their expectations downwards” as the date of these announcements approaches. This discovery enabled the researchers to profit from a unique situation of arbitrage: by measuring the difference in the earnings expectations of analysts’ with the rational benchmarks provided by the machine learning algorithm, there was an opportunity for investors to profit by short-selling overly optimistic stocks. On the other hand, investors could buy pessimistic stocks at reasonably low prices and sell them after better than expected earnings for a healthy profit margin. Bisenberg identified two main findings in the paper.


Firstly, the ML model enables investors to “track over time the stocks for which analysts are too optimistic or too pessimistic”, thus furthering their opportunities to profit. Second, it is impossible to make a general statement saying that analysts are pessimistic about all stocks. Different stocks face varying levels of optimism and pessimism, and although these are often biased, they cannot be aggregated to make a general statement.


The researchers also found a window of opportunity for corporations. On several occasions, markets do not correct the price of an over- pessimistic or over-optimistic stock. By using the ML model, a manager can make more profitable decisions with respect to issuing or buying back shares: either issue new stocks when the companies’ shares are overpriced or buyback stocks when their shares are underpriced.



Concluding Insights


It is clear that AI’s dominance is seeping into almost all important aspects of our daily lives. With major technological leaps in recent years, it is likely that AI’s power and influence will only increase. With respect to the stock market, is AI posed to become the next “Intelligent Investor”?



 

References:


1. van Binsbergen, Jules H. and Han, Xiao and Lopez Lira, Alejandro, Man vs. Machine Learning: The Term Structure of Earnings Expectations and Conditional Biases (September 14, 2020). Jacobs Levy Equity Management Center for Quantitative Financial Research Paper, Available at SSRN: https://ssrn.com/abstract=3625279


2. "How to Beat Analysts and the Stock Market with Machine Learning." Knowledge@Wharton. The Wharton School, University of Pennsylvania, 13 October, 2020. Web. 18 December, 2020 https://knowledge.wharton.upenn.edu/article/beat-analysts-stock-market-machine-learning

Related Posts

See All