Last year I reported on an interesting firm, Thinknum, and its innovative new business model. The company makes sophisticated financial models available to everyone on an open, distributed computing platform. It is one of a handful of firms at the cutting edge, and I felt it was time for a check in.

For a start-up, Thinknum has undergone a surprising number of changes in a short period of time. In this case, what has changed can be distilled down to one phrase: If it is on the internet, it can be analyzed.

Below is the unedited transcript of an interview I conducted with Thinknum’s co-founders, Justin Zhen and Gregory Ugwi, that brings us up to speed on the current state of the art in the research analyst world.

CFA Institute: Walk us through an update about eXtensible business reporting language (XBRL) at the regulatory level. Are regulators mandating that more financial data and business data be XBRL tagged?

Justin Zhen and Gregory Ugwi: When we started Thinknum, we were big believers in XBRL. In fact, I joined the FASB [Financial Accounting Standards Board] taxonomy board to help with their XBRL implementation. Eight months after marketing XBRL technology to financial analysts, it became clear that there was weak market demand for that technology. We learned that financial analysts were much more interested in accessing alternative data that is now available on the web, a problem not being addressed by XBRL. They want to see data on how weather is affecting retailers in various regions or go through state filings to see which drilling companies are producing the most oil. Analysts currently spend a lot of time pulling this data where the infrastructure for a computer program to do so already exists. In short, XBRL has no traction, while alternative data is revolutionizing financial analysis.

What could be done to improve the quality of information in XBRL filings?

The most important lesson we have learned in creating a start-up is to build something that users want. XBRL proponents can also take this lesson to heart. The current XBRL ecosystem is heavily weighted towards the companies that are publishing this data and not towards the analysts who are end users.

XBRL proponents need to start with the end user in mind and work backwards. There is a need to scale back and develop a minimum viable product that some analysts are excited about and then constantly iterate based on user feedback. It is tempting to get excited about a certain technology and keep building until the product has become large and bloated. If XBRL were an actual start-up, it would have failed a long time ago.

Earlier you mentioned alternative data. Tell us a little bit more about what you mean by that term?

Sure. Traditionally, investors valued companies by analyzing earnings reports and market data. With the onset of the networked world, savvy analysts have identified ways to scour the web for all kinds of information that provide insights into the operations of publicly traded companies in amazingly granular detail. Analysts track how a company is doing in real time by monitoring product pricing changes for retailers, mobile app adoption for e-commerce companies, or updates to clinical trials for individual drugs in the case of pharmaceuticals. They study how route changes are affecting various airlines or track how unemployment trends are affecting companies across all sectors. It’s been very interesting to see all the various ways creative hedge fund analysts are using our software.

So these sound like some of the things that Thinknum now provides its users, yes? Tell us some of the other new developments taking place at Thinknum.

Most of the new developments at Thinknum have involved building software to solve the unique problems that arise from dealing with alternative data. Alternative data is new and this is what makes it such a rich source of alpha. It also presents challenges where the traditional techniques relied on for analyzing financial data are not effective.

By listening to the hundreds of hedge fund clients that we have and solving the pain points they experience, we have developed a platform that is uniquely suited for extracting insights from alternative data. For example, a few months ago we realized that users are not interested in data, but rather changes in data. We built a system for users to track various data points and understand when these data points are deleted, updated, or added to the system for the first time. This cuts hours from the work flows of our clients.

Fast forward five years into the future. What does the e-landscape look like for analysts? What will we be able to do then that we cannot do now? What about 15 years from now?

The first trend is the spread of high finance to the small business sector. The most important problem any post-subsistence society has to solve is how to most efficiently allocate its resources. As more business activity comes online, financial analysts would be able to deploy their skills to help a larger set of companies operate more efficiently. In the past cycles, we have seen investment banks gain scale by aggressively going after non-Fortune 500 clients. Private equity firms have also been able to extract more value from companies in a wide array of industries. This trend will accelerate as information about the majority of human commercial activity comes online in the next five years. A farmer in China will determine whether he wants to allocate crop space to plant tomatoes or potatoes using the most sophisticated tools available via an internet connection. We are already seeing trends where the web is leading the way in alternative lending. The biggest investment banks and asset managers have become focused on building consumer facing businesses to open the door to a wider array of clients and capture this opportunity.

The second trend is the rise of automation. As the amount of data to be analyzed increases exponentially, we are seeing an increase in returns on automation. Engineers will train machines to take on the manual tasks that analysts now spend many hours on. Over time, the software for automating these tasks will become more sophisticated and get to an inflection point where artificial intelligence will not only answer questions better than any human could, but ask more relevant questions as well. Humans would still be needed to set economic policy and figure out new avenues to deploy this analytical technology.

Image credit: ©iStockphoto.com/Meriel Jane Waissman

Originally published on CFA Institute’s  Enterprising Investor.