Method for strategists and analysts to break the barrier of bounded rationality in equity research
In less than one month since the current earnings season began, more than 3000 earnings conferences were conducted by business executives of public companies. It is impossible for any strategist or analyst to read every transcript. Although quantitative chart depicts trends, it does not tell the story, i.e., facts, outlooks, markets, customers, competitions, strategies, or business models. It is also humanly impossible for anyone to read and digest voluminous earnings transcripts, 10-Q and 10-K. This is limitation of bounded rationality where decision makers are making decisions without the benefit of all available information.
Using a method called “abstraction”, data-science analyzes these documents to produce abstraction of textual documents. This technique has been used among software developers and data scientists to depict complex systems using different level of abstractions. Level of abstraction is defined in PC Magazine’s encyclopedia as:
“The amount of complexity by which a system is viewed or programmed. The higher the level, the less detail. The lower the level, the more detail. The highest level of abstraction is the entire system. The next level would be a handful of components, and so on, while the lowest level could be millions of objects.”
When the highest level of abstraction presents a table of content or index to a report, users get the benefit of a summary assessment with the option to drill down into specific details.
Before there is data science, analysts and strategists have been using annotations, high-lighters, and tabs to index key discussions. With the advent of data science, it is possible to automate these tedious tasks. Thus, eliminating the limitation of bounded rationality. The following is an example showing partial multi-level of abstraction of a risk model:
