Opening Cross: At the End of the Day, It’s All Data

Data isn't just prices; it's unstructured text, charts, sentiment scores, and even analysis of photographs.


If you haven’t heard, this Wednesday, May 20 is the date of Inside Market Data’s flagship North American Financial Information Summit in New York. But in the evening, after the main conference agenda, we also host the annual Inside Market Data and Inside Reference Data Awards, where we highlight the best and brightest individuals, companies and products of the past year. One of the categories is Best New Data Product, and casting my eyes over the stories in this week’s IMD, it occurred to me that the winners of this category in future may be very different from what we’ve seen in the past.

What I’m getting at is the changing nature of what we classify as market data. No longer is market data necessarily data from a market, like an exchange or a broker price. Now it includes synthetic prices created to serve the function of a reference price, such as Interactive Data’s Continuous Evaluated Pricing service, which the vendor has just made available via its Consolidated Feed.

Market data also incorporates all kinds of technical signals and analytics, such as those provided by technical analysis software vendor Updata, which has begun offering data from Toronto-based economic data aggregator Quandl as a source for its analytics.

And in today’s era of social media analysis and crowdsourcing of data, market data also encompasses signals derived from social media, such as those created by Market Prophit, which—with S&P Dow Jones Indices as calculation agent and Bloomberg distributing the data—has created an index based on analysis of social media chatter. Like many new datasets, this is really a case of revamping an existing data source like an index by introducing a novel new methodology, just as how Estimize has revamped old-school estimates with a new crowdsourced approach.

And this is far from the least traditional datasets that we now recognize as market data: in the same way that latency figures became an important input to trade routing decisions, network security information such as that provided by Redscan—which has partnered with data consultancy Cordatum to help increase its sales penetration of financial markets—is likely to become an equally important signal about the health of a company’s network, or perhaps a counterparty’s or supplier’s infrastructure.

Meanwhile, TickerTags, a startup provider of tags covering mentions of companies or keywords on social media that just closed a round of seed funding, is preparing to launch its online platform for monitoring social media chatter for recognized keyword “tags” that could be indicative of a market-moving change relating to a company or its products.

Having moved from structured price data to unstructured text and sentiment, there’s still another level of data that’s even harder to derive, but potentially just as useful—visual data that needs to be interpreted by the human eye, then requires computers to be trained to recognize trends or changes on photographs. For example, companies like Genscape rely on a combination of physical monitoring with visual analysis of power stations or oil fields, while others use photographs of how low freight tankers sit in the water to draw conclusions about their cargo. And similarly, Orbital Insight, which recently hired Eze Software sales veteran AJ DeRosa to boost its presence among financial firms, uses satellite imagery to estimate data about store sales, crop yields and building projects. A picture may be worth a thousand words, but reverse-engineering that level of value out of a picture is a complex and no doubt expensive process.

But ultimately, these vendors, like Ticker­Tags, are “creating new forms of financial data,” says TickerTags co-founder and chief executive Chris Camillo, who adds that the kinds of data scientists required to perform these tasks don’t come cheap. Of course, no one expects data to come cheap; they just want it to deliver value, which is what all these new types of data aim to achieve.

  • LinkedIn  
  • Save this article
  • Print this page