YCharts Preps Unstructured Company IR Datasets for Screening, Research

YCharts will collect data from press releases, presentations in Microsoft PowerPoint slide decks, and videos on companies' investor relations websites, and make it searchable alongside the vendor's existing database of economic indicators.

Ara Anjargolian, New York-based co-founder and chief technology officer of YCharts, says the new datasets will help traders identify companies related to other companies, sectors or topics by capturing information that might not be readily apparent from the company's fundamentals.

"For example, you can search for terms in text... such as companies besides Apple talking about [their exposure to] the iPhone. Or, if you're a hedge fund manager, you might be looking for a certain type of company that runs a software-as-a-service business. Well, there is no industry classification or sector for SaaS, and there's no way to get that just from a company's financials," Anjargolian says. Alternatively, traders could use this feature to find more in-depth information on companies they already follow. For example, solar power companies typically don't report important ratios such as cost per watt in their financial statements, he says, but would be likely to mention that figure in their conference call slides.

"Collecting financials is commoditized-it's the other information that's valuable," he adds. "All Big Data projects are essentially trying to turn unstructured data into structured data."

To obtain the data, YCharts used Amazon's Mechanical Turk crowd-sourced jobber platform to identify the relevant IR website URLs for around 8,000 US companies, then used an automated system to recognize each new type of announcement or document

"From collecting economic indicators, we've learned a lot about how to collect this data," says YCharts chief executive Shawn Carpenter, adding that the vendor decided to go "straight to the source" rather than collecting "regurgitated" news that goes through several hops before being published, and which might only appear hours or days after an IR document is first posted on a company's website.

Anjargolian says YCharts has mostly completed the first step of initial data collection and building an effective search tool that can search and present links to the original source documents, and now just needs a process to ensure it does not miss current data if a company changes its IR website URL or its IR provider, before the next step of parsing data out of the documents and aggregating all relevant information in one place. "That's more like a two year project... [though] some people will be happy with just having access to the data," he says.