There is a constant flow — you could even call it a torrent — of academic research about the investment world. Most of it is quantitative in nature, describing conditions of the past in the hopes of finding a key to unlock the future.
For investors employing systematic strategies, such studies offer ideas that could be incorporated into their models. They also can be of value to others, providing questions to consider about investment process, decision making, organizational design, and the functioning of markets. This posting references a few recent papers in that light and includes some of those questions (in italics).
As you might expect and as the selection below illustrates, the percentage of studies that involve artificial intelligence in some way has expanded significantly of late. (An asterisk next to the link indicates that AI was used in the analysis.)
Analysts and analysis
“Large Language Models as Financial Analysts,” Miquel Noguer i Alonso and Hanane Dupouy (link*).
If, when answering “extrapolation questions that are the core of valuation and stock picking, the level of analysis provided by these LLMs is similar to that of skilled humans,” how should investment processes be structured? Should we be hiring prompt engineers rather than experienced analysts?
“Re(Visiting) Large Language Models in Finance,” Eghbal Rahimikia and Felix Drinkall (link*).
Are you using general purpose LLMs or (much smaller) domain-specific ones? How do you mitigate the look-ahead bias that results from LLMs being trained over multiple time periods?
“Can news predict firm bankruptcy?” Siyu Bie, et al. (link*).
Since the evidence shows that “ChatGPT-generated news-based variables significantly improve bankruptcy prediction,” how should that change equity and bond analysis?
“(Deep) Learning Analyst Memory,” Laurenz De Rosa (link*).
How are analyst beliefs about future earnings formed? Does recalling salient events and periods from the past make analyst forecasts better or worse?
“Learning to Be Overprecise,” Christoph Merkle and Philipp Schreiber (link).
Why do the confidence intervals of forecasters remain unrealistic in the face of evidence that they are too narrow? Why do they “adjust their beliefs too little, which results in a persistence of overprecision”?
Financial reporting
“The Impact of Tone and Readability on Understanding Earnings Releases,” Yoshitaka Hirose and Takeaki Ito (link).
Can investors be misled about a company’s prospects by the readability and tone of its earnings announcements?
“Generative AI in Financial Reporting,” Elizabeth Blankespoor, et al. (link*).
Do you care if company financial reports are being written to some degree by generative AI systems? If so, are you using tools to detect such usage?
Gender
“Gender Differences in Sell-Side Analysts’ Corporate Site Visits,” Guangyu Li, et al. (link).
Why do female analysts visit companies less frequently and have higher levels of relational visits (those accompanied by buy-side investors) rather than analyst-only ones?
“Beyond the Ticker: Female Brands and Fund Manager Investment Decisions,” Emanuele Bajo, et al. (link*).
Since brands often develop distinct gender identities, are the stocks of companies evaluated and used differently by male and female portfolio managers?
Big questions
“Measuring Multi-Period Returns,” Raman Kumar, et al. (link).
Is “Cash-Flow Weighted Return” a better way “to measure average multi-period returns for investments with intermediate cash flows and varying periodic returns”? Could it displace the deeply-entrenched IRR as a metric?
“Limits to Diversification: Passive Investing and Market Risk,” Lily Fang, et al. (link).
Do the benefits of diversification diminish as more and more investors invest in index-based portfolios?
What interesting research have you read lately? Please send it along.
Published: January 4, 2025
Subscribe
To comment, please send an email to editor@investmentecosystem.com. Comments are for the editor and are not viewable by readers.