Wells Fargo’s AI Computer Program Predicts Where Amazon’s HQ2 Will End Up
In January, Amazon winnowed the search from the hundreds of cities and regions that applied to just 20 finalists, including Atlanta, Austin and Toronto. But the Aiera computer program run by Wells Fargo Securities says Boston is the most likely landing spot for the $5 billion new Amazon outpost and its 50,000 jobs, according to a report in Barron’s Tech Trader Daily blog. Chicago is second-most likely, Aiera concluded.
The program crunches hundreds of thousands of pieces of information, including news articles and social media posts on Twitter and Facebook, and then uses machine learning to discern trends and make predictions. That process seems better suited to making stock calls, where the buy and sell decisions of millions of investors every day figure into the price—as opposed to the HQ2 hunt, in which decision-making authority ultimately resides with just a couple of people. (Jeff Bezos’s diary probably has a lot more relevant info about where HQ2 is going than every public Twitter feed in existence.)
Analyst Ken Sena, who oversees Aiera, acknowledged as much in his report revealing the picks. “Because the basis for Aiera’s prediction is really a statistical measure of significance across public commentary and market reactions, it is fair to characterize the output (i.e., Boston) as what the broader public (and the market) believes will happen when Amazon announces their decision later this year,” Sena noted. In an additional portion of the report, Sena reviewed factors like workforce talent, locations of existing Amazon offices and government relations benefits that Aiera likely skipped.
Aiera is an acronym that stands for “Artificially Intelligent Equity Research Analyst.” As to stock calls, the system works pretty well. Aiera has made almost 8,000 day, week or month-long stock predictions and was correct 82% of the time, according to Wells Fargo.
Other sources on the HQ2 hunt have picked other favorites. At the betting site Paddy Power, the wisdom of crowds also favored Boston, followed by Austin and Atlanta.
Get Data Sheet, Fortune’s technology newsletter.
Experts have crunched their own factors to come up with possible winners. Seattle tech writer D.E. Wilson made a persuasive case that HQ2 will end up in the northern Virginia/Washington, D.C. area because of its proximity to a key Internet exchange and suitability to house the chiefs of the company’s Amazon Web Services unit. John Cook, who founded the Seattle tech news site GeekWire, also ranked the D.C. area most likely, trailed by Austin and Boston, based on his lifetime coverage of the company.
And then there are the breadcrumbs and clues from Amazon itself. The company’s move to lease to 1 million square feet of office space in Boston in January sounded a lot like the publicly announced “Phase 1” of HQ2 as outlined in Amazon’s original RFP. Amazon execs were spotted visiting sites in the Washington areas this month, fueling further speculation.
Amazon (AMZN) hasn’t said when it will announce the winner beyond specifying this year. Until then, the guessing games will continue.