2 Comments

Summary:

SEC filings don’t have to be dense PDFs filled with numbers. A former SEC analyst turned the data into an interactive database so companies and their filings could be easily searched.

A Circos diagram shows the relationship Google's leadership has with other companies.
photo: RankandFiled.Com

At their best, SEC filings are troves of information that businessmen, reporters and retailers drool over. But in their current state as a long winded PDFs of financial and legal jargon, it’s hard to find the right file, let alone a nugget of information.

Rankandfiled.com wants to change that.

Maris Jensen, a former SEC analyst, spent the last eight months working to make the filings not only more accessible, but also easy to understand. The problem though wasn’t getting the data, it was doing something with it.

RankandFiled.Com shows the SEC filing counts for Yahoo. For those who want to dig deeper, an option to switch to a table mode lets users see a complete list of documents filed by the company.

Rankandfiled shows the SEC filing counts for Yahoo in comparison with their stock price. For those who want to dig deeper, an option to switch to a table mode lets users see a complete list of documents filed by the company or you can click on any point to be taken to the filing itself.

The SEC EDGAR database is notoriously hard to use, and the agency has dragged its heels in finding ways to better utilize its data and comply with the government’s mandates for interactive is stagnant, according to an official letter from the Congressional Committee on Oversight and Government reform. (For more on making data accessible to the public, check out our Structure Data conference in New York next month).

Maris Jensen, creator of Rankandfiled

Maris Jensen, creator of Rankandfiled

While still at the SEC, Jensen originally pitched the idea to upgrade the system into an interactive idea in March 2013, but was shot down. After she was then fired — for what she said were unrelated reasons, including “profanity unbecoming of a federal employee” — she returned to her idea and started searching how to make a website with her limited coding skills. “Nothing an actual programmer would call programming, but I had used some of the data analysis packages before,” Jensen said.

She spent a month writing a system to parse and index the filings, to the sum of more than 25 million files. But then came the question of what to do with it. Jensen started by searching freelance sites for what data people were looking for and then read lots of message boards about the problems people had with the database to narrow down what she needed to display.

A view of Facebook's market activity by day, including trades and cancellations, on Rankandfiled.Com. The chart uses data released quarterly by the SEC's Division of Trading and Markets.

A view of Facebook’s market activity by day, including trades and cancellations, on Rankandfiled.com. The chart uses data released quarterly by the SEC’s Division of Trading and Markets.

Then it was a matter of visualizing the normal black and white documents. Jensen reached out to open-sourced code from D3 developers to create the interactive graphics.

“Pretty much everything you see there is me smashing together examples written by all of those people,” Jensen said. “I wrote to a lot of those people, and they would write me back and help with things. Honestly it was a collaborative effort.”

The result is a site that turns the numbers and names hidden in the 25 million files into colorful Circos diagrams and area charts. Instead of searching for a company and returning a text list of all of the filings, Rankandfiled.com places them on a line chart, overlaid against the stock price. A click on the point opens up the original filing.

However, its real strength is its ability to demonstrate relationships between companies and people that would normally take opening up hundreds of computer tabs and speed reading PDFs to connect the dots.

The most interesting visualization on RankandFiled.Com shows the influence between companies in a Circos diagram. Here, the diagram shows ownership among Microsofts officers, directors and owners since 2003. The width of each link is proportional to the number of filers who have owned both companies.

The most interesting visualization on Rankandfiled.com shows the influence between companies in a Circos diagram. Here, the diagram shows ownership among Microsofts officers, directors and owners since 2003. The width of each link is proportional to the number of filers who have owned both companies.

While some of the tabs show the basics like the financials, the influence and related funding sections are where Jensen’s data processing sets it apart from just another SEC database. The influence Circos diagrams, shown above, link a company’s officers, directors and beneficial owners to other companies they have owned. The tapered end shows which company was owned first, and if you hover the line, it shows you who was the person to own it.

The program also crawls through select exhibits in SEC filings to find information on things like subsidiaries.

“To do that [in the EDGAR database], you’d have to know that companies file subsidiary information in a certain exhibit. So it’s Exhibit 21 on the form 10K, so you’d have to go and find a form 10K and click through and find Exhibit 21,” Jensen said. “Most people don’t know that. I didn’t know that before starting.”

On RankandFiled.com, the list of a companies subsidiaries if they have them are quickly listed under the tab.

This related funding chart from RankandFiled.com shows the relationship between Google officers and other companies they fund.

This related funding chart from Rankandfiled.com shows the relationship between Google officers and other companies they fund.

There are some limitations with the site though, many of which Jensen acknowledges.

Its homepage is plain and can be a bit hard to figure out what you’re looking at at first glance. There’s a five to ten minute lag time from the SEC’s feed, Jensen said. Plus, there’s room for error when it comes to parsing the files if company’s input the data incorrectly, like using paragraphs to describe something instead of listing companies. The same alternative or misspellings of words are present in the EDGAR database too though.

This heat map shows the distribution of word frequencies in Facebook's Management's Discussion and Analysis sections. Words like compensation, administrative and advertising are used more often.

This heat map shows the distribution of word frequencies in Facebook’s Management’s Discussion and Analysis sections. Words like compensation, administrative and advertising are used more often.

While Jensen said she would love to keep working on turning public data into visualizations, her concern at the moment is getting the SEC to pick up her website since she can’t host it forever. That’s why she wants the SEC to step in and continue to build on her site.

“I’m not an SEC filing expert at all, so there’s probably a lot more that the website should and could be doing,” she said. “I’m not an expert in these filings so if you go to the financials tab, I didn’t try to organize the data in any certain way. It’s just listed how it’s listed in the filings. There should be some effort there, but somebody who knows how to read SEC financials should do that, not me.”

  1. Just been browsing this and it is quite an excellent effort to make sense of the regulatory filings that underpin our financial system. Quick and insightful access will become a lot more important with the ramp up for the JOBS Act and the change in the startup funding rules that are just now coming into effect.

    Share
  2. Shahbaz Chaudhary Thursday, February 27, 2014

    Ten or so years ago I tried to make sense of EDGAR , but couldn’t parse the data. There was an open source project on sourceforge which tried to parse the reports. Unfortunately, IBM held a patent on similar technology and whoever created that open source project had to shut it down. Very glad Maris is doing this.

    Share

Comments have been disabled for this post