On August 6th, a fire started in a 2,900-acre oil refinery that Chevron owned in Richmond, CA, evacuating the facility and crippling operations for much of that day. As is usual following events like this, the fire triggered a price hike in California-blended gasoline (or Carbob, as it is called) of 23 cents--the highest spike in a single day since November 2007, Bloomberg reported at the time.
A data analytics company called DataMinr claims to have known about the fire before big news organizations did, by using Twitter as an alert service for the breaking news. That same information was available, then, to their clients--financial firms who make rapid trading decisions worth millions of dollars. Those that saw this information early could have made a profitable trade based on the Chevron news if they knew a price hike was coming before anyone else did.
In a nutshell, this is what DataMinr does. It delivers useful breaking news to companies that bet high and bet quick. By spotting emerging trends in Twitter’s noisy stream, the company is taking on the work of newswires, with the goal of beating them at their game.
DataMinr works its magic by augmenting its Twitter tips with an individual trading firm’s database of companies and sectors. But it’s more than just an alert service--the news is presented with relevant analytical and statistical context dictated by the company's core investment interests.
“Essentially what we’re doing is predicting the present,” founder Ted Bailey tells Fast Company, quoting Google’s Hal Varian. “[We’re] giving people just a better sense of what’s happening right now, and a greater confidence in the understanding of a particular situation or breaking information." To be clear, DataMinr stops short of telling its customers what to do. The company’s goal is to provide enough information for a client to make a quick decision.
DataMinr just wrapped up a $13 million funding round and is growing its client list of financial firms and government agencies, the kinds of places that place a high premium on early information. It’s even caught the attention of Twitter, with Dick Costolo and other execs counting the company among their favorite Twitter ecosystem startups. And in April this year, Twitter granted DataMinr real-time access to the Twitter firehose. “Twitter is an incredibly rapid-fire ticker tape of emerging events in the market,” Bailey says. “Our big focus is on real-time analytics and event detection… Being able to pinpoint specific hot spots of activity in real time, and categorize those and describe them.”
While that’s a big responsibility, it also comes with big payoffs. “They’re industries where time advantage is an enormously valuable edge, where warnings and metrics that describe the present are really valued,” Bailey says. “I think the monetary value of earliness and immediacy that we provide our clients is enormous.”
Granted, some regions of the world have higher numbers of more prolific tweeters than others. Also, their warning system for news events is limited by the fact that someone needs to be present. But what Bailey says he’s found in the three years DataMinr has been in business is this: “There really aren’t many events these days that occur where there isn’t a local reaction. It just speaks to Twitter’s pervasiveness across the world.”
Which brings us to the DataMinr algorithm. In the early days, building and refining this secret recipe was Bailey’s highest priority. He’s a little reluctant to describe how it does what it does, but in the past he's said that among other things, it picks up not just key words, but patterns in relationships between key words and how frequently words are paired together. Bailey has said that DataMinr was a full 20 minutes ahead of the news in picking up early reports of Bin Laden’s death off Twitter. And it didn’t need more than about 19 tweets to get the alerts ringing.
Ultimately, the sophistication and reliability of DataMinr's algorithm does and will determine the company's success. “I think that for data companies, that is the critical component: To what degree are they focusing on innovation in order to push the algorithms forward, to develop enough nuance to deliver real edge.” A good chunk of the latest funding round that DataMinr won will be used to further bulk up its algorithm even though, in Bailey’s opinion, the company is in a “fantastic position” and is seeing “very high demand.”
[Image: Flickr user K.G.Hawes]