Waters Rankings 2020: Best Low-Latency Data Feed Provider—S&P Global Market Intelligence

When people think of low-latency feeds, they inevitably picture raw exchange feeds, microwave transmission, and ultra-fast switches and feed handlers. But as the cost of achieving meaningful latency gains increases, and the return on investment from pursuing outright latency gains gets smaller, data consumers are looking at other areas where speed can make a difference.

As a result, the winner of this year’s best low-latency data feed provider category, S&P Global Market Intelligence, represents a departure from previous winners, capturing readers’ votes for its Event-Driven Alerts (EDA) feed. EDA isn’t a pricing feed, but is rather a low-latency XML feed of transcript content from events such as earnings calls, with millisecond-level timestamps, and embedded with metadata that identifies the company, what kind of call it is, and even which company executive is speaking, using natural-language processing and entity recognition tools developed by recent acquisition, Kensho.

“Hedge funds and investment managers are all coming up with new and different ways of testing this—using it alongside traditional data, or other data they have in-house,” says David Coluccio, managing director of data management solutions at S&P Global Market Intelligence. “There are so many ways they can look at this contextual data. For example, they can use it to spot red flags and to identify similar companies without those red flags, or capture data points that are not reported as financial items in a document.”

S&P has a historical dataset of millisecond-timestamped transcript data for back-testing that goes back to 2016, when it launched EDA, and 10 years’ total of non-timestamped transcripts. As clients test the service, they are suggesting other ways to develop the dataset. For example, some clients are asking for the original audio of earnings calls alongside the transcript so they can analyze the tone of a speaker’s voice, Coluccio says.

The vendor is considering adding its proprietary sentiment scores to the text of call transcripts. But its main focus for the future of EDA is to seek out and integrate unique datasets from data partners that can add insights to EDA, and incorporate them into the feed, ensuring that all datasets are linked and structured to make them easy to use alongside other content.