After AI update, Bloomberg looks to boost terminals with more alternative data
Bloomberg will expand its alternative data offering on its terminals this summer, the company’s chief technology officer told City PM, after the firm announced that it plans to integrate an ChatGPT-like AI model into its software.
Shawn Edwards said the financial data vendor aims to bring new sets of data to its terminals alongside the company reports and market estimates that Bloomberg already compiles.
The Bloomberg Terminal is a popular computer software used by financial services professionals to monitor financial data and make trades.
Bloomberg already offers some forms of alternative data, but much of it is extremely expensive to collect and difficult to use, meaning only the most sophisticated hedge funds use it routinely.
Edwards said he wants that to change and “bring it to the masses.”
“What we’re doing, I think that’s unique and special, is we’re taking all this information and we’re aligning it to all the Bloomberg data we already have,” he said.
Edwards said he was hoping to add more up-to-the minute data points such as live footfall data, app usage and information about web traffic.
“Often companies don’t even report this level of detail, but I can look at it right now and see how a company is performing well before earnings,” Edwards said.
With this extra data, Edwards said “you can look at store sales of Lululemon, geographically broken down in the United States. You can look at the current number of subscriptions for something like Blue Apron.”
Edwards highlighted the case of exercise equipment company Peloton, where using alternative data, like credit card analytics, some hedge funds were able to see Peloton sales were falling post-pandemic well before the company published its earnings.
News of the expanded data offering comes after Bloomberg released BloombergGPT, its large language AI model built for finance.
As a large language model, it is likely to replace many of the other AI tools that Bloomberg already has. “Its understanding of language is so much better than our prior models”, Edwards said.
BloombergGPT could also be used to help Bloomberg’s chat systems run more effectively by helping traders to focus on the most urgent conversations with their clients. Basic tasks can be automated, while the largest trades are flagged for human intervention. “All of a sudden, they become superhuman,” Edwards said.