Barclays leans on alt data for the nuance it provides
The devil is in the details.
That’s how Ryan Preclaw, who co-leads Barclays’ data-science research group, explained his approach to alternative data. The group, which is made up of an investment sciences team and data science team, publishing reports and analysis for the banks’ clients.
“The most exciting thing alternative data offers over traditional data is its granularity,” Preclaw told Business Insider via email. “We can provide our clients with actionable insights and answer very specific research questions that would have been impossible in the past.”
From investment to retail banking, their is value across the firm in using obscure datasets, Preclaw said. But there is no specific type of alt data that the UK bank targets. Barclays uses everything from geolocation data, which has proved particularly useful in recent months, to open data, which is usually published by some kind of government entity.
However, when it comes to the format the data is in, Preclaw said Barclays made a strategic choice to have data be as raw as possible.
“We want to be maximally flexible in what we can do with it, even though that means a lot more upfront investment in infrastructure, storage, and algorithm development to turn it into useful information,” he said. “Raw data has two advantages for us: first, we can get answers to questions that might be impossible if we received data in a pre-processed form; second, it puts us into a position to talk to our clients about the nuances of how to work with alternative data.”
While Preclaw declined to disclose the bank’s budget for alt data, he did have some advice for prospective providers. First, it’s important to provide written material about the dataset before even trying to set up a conversation. When a chat does occur, Preclaw said the bank is eager to understand what the data set is offering “in a very specific way.”
“We want to see what a row of your data looks like, what all the features are that you have in it in as plain language as possible,” he added.
Transparency is also a top concern, as is simplicity, Preclaw said. Be prepared to explain the process in a concise way. Tech and salespeople are also encouraged to attend the first meeting to provide a 360-degree view of the data, he said.
Preclaw also had a piece of advice for something to avoid.
“One thing we are NOT interested in seeing is a back test against returns,” he added. “Those are so easy to do incorrectly that we can’t put any dependence on it unless we’ve done it ourselves.”