I recently read a book on stock picking. The book on value investing, exhorted the readers to be aware that good stock is not the same as a good company. Over the years, stock-picking books have become a genre in themselves.
Investors trust passive equity funds
I always like to see how a well-formulated argument fares against the data. In this case, the data shows that despite the accumulated knowledge of more than a century and availability of information, the art of stock picking is not doing that well after all.
In the last 20 years, investors have increasingly put their faith in fund managers that just mimic the index instead of applying their mind.
How did we end up here? The problem lies in over-reliance on financial results and subject matter experts.
Alternate Data + Artificial Intelligence = Future of Asset Management
The financial reports come only at the end of the quarter, but equity transactions happen all through that period. Secondly, it can be quite a task to compare two companies based purely on audited financials. There are many companies which are profitable but have negative free cash flows. You would need to be an expert accountant to get to the bottom of all the assumptions.
Is there a way to get the pulse of a company independent of its financial reports? Many analyst firms have started looking at alternate data such as:
- Satellite pictures of cars in the parking lot of a retail store
- Satellite pictures of agricultural fields
- Pictures of the number of ships in ports
- Scanning emails for invoices
- Scanning social media for adverse comments
If you read the last paragraph cursorily, I suggest you read it again.
You would notice that while the ideas listed above seem innovative, the task of extracting insights from abstract data can be overwhelming. For example, how do we count the number of cars in front of a retail store? Does it change by the time of the day or day of the week? And let us not forget that the company can have many stores. Welcome to the world of large and unstructured data!
It would be expensive and error-prone if we ask human beings to do this. This is where AI comes into the picture. Using computer vision models, we can count the objects in an image at a rapid scale in a matter of seconds. Using natural language processing(NLP), we can scan the emails of volunteer shoppers and find out the value of retail invoices for various brands. Similarly, we can monitor social media to identify adverse comments by employees at an industrial scale. Another set of AI models can use these insights and build a correlation with the company’s revenue.
The use of Alternate Data and AI has already started in the world of asset management. It is only a matter of time that it becomes mainstream.
Read the original article on Centelon site.