Robots are being let loose to write investment reports for Morningstar, the research house that helps investors choose among thousands of mutual and exchange traded funds for saving and retirement.
At a time when the finance industry is debating the workloads of its staff during a boom in activity and a disruptive pandemic, Morningstar has found a way to increase its written research without further taxing its army of human analysts.
The machine-generated reports that began rolling out this week set out the rationale behind Morningstar’s so-called analyst rating on a fund, which run through gold, silver and bronze to neutral or negative. The ratings, similar to Wall Street buy or sell recommendations, are separate from Morningstar’s more famous star ratings system, which just measure funds’ past performance.
Robots were already being used to generate the analyst rating itself on thousands of smaller funds. Morningstar said this week that the robot ratings have performed as well as the recommendations generated by human analysts, based on three years of data.
The company’s 130 global analysts will stay focused on the most popular funds, said Lee Davidson, head of manager research and quantitative research at Morningstar. They currently provide written reports on 4,284 funds globally, representing $20.5tn of assets under management.
Human or robot: can you spot the difference?
The following is a selection of quotes from human and computer-generated Morningstar research reports, but which are which?
1 — “One of three managers has left, but Pimco Total Return still offers a manager juggernaut and a battle-tested process.”
2 — “Management turnover at Janus Henderson Balanced reduces conviction in the execution of its process, warranting a downgrade of its Morningstar analyst rating to neutral from bronze across all share classes.”
3 — “Templeton Growth has a new lead manager, but there remain more compelling options for global equity exposure.”
4 — “Increased confidence in Strategic Advisers Large Cap’s management team is the primary driver of this share class’ rating upgrade to a Morningstar quantitative rating of silver from bronze.”
5 — “American Funds New Perspective’s stake in the polarising electric car maker Tesla grew from 3.65 per cent in June to 7.3 per cent in August, 2.5 percentage points larger than its next-biggest holding.”
6 — “T Rowe Price US Equity Research’s notable investment process and strong portfolio-management team underpin this share class’ Morningstar quantitative rating of gold.”
“You need boots on the ground to interview a fund manager and talk to the team about their investment process,” said Davidson. “That kind of colour is not captured by the quantitative process.”
Computer-generated ratings apply to a further 37,962 funds that account for $14tn of assets under management, and it is these funds that will now start to get written reports penned by robots.
“Providing a written report helps explain the justification for a fund’s rating, its risks and rewards in the future,” Davidson said.
Morningstar has spent the past 18 months working on the machine writing process and Davidson said “we will need to keep improving them”, incorporating suggestions from asset managers and investors. The machine-written analysis will automatically update each month.
Both the human analyst and machine-derived ratings are meant to tell investors whether a fund is likely to outperform or lag an index or benchmark over a market cycle.
The algorithmic ratings, which Morningstar has been generating since June 2017, are meant to mimic the firm’s human analysts by considering the same factors, such as a fund’s fees, management turnover, track record and portfolio risks.
Human or robot? The answers
Quotes 1, 2, 3 and 5 were written by Morningstar’s human analysts, while 4 and 6 are computer-generated.