(Bloomberg) -- The artificial intelligence behind the Horizons Active A.I. Global Equity ETF has apparently never heard the phrase "sell in May and go away."
The Canadian exchange-traded fund, which became the first global equity ETF run by robots when it launched in November, went all in on stocks when it rebalanced on April 30. It reduced its cash position to 1 percent from the 25 percent maximum it can hold, and increased its equity exposure to 99 percent.
That’s the opposite strategy taken by some ETFs run by Horizons ETFs Management Canada Inc., which can almost fully rotate out equities during the typically volatile month of May, said Steve Hawkins, co-chief executive officer the Toronto-based firm.
“It’s going to be interesting to see who’s right: historical technicals or a forward-looking AI system,” Hawkins said. “I’m nervous and excited all at the same time.”
The ETF, which trades on the Toronto Stock Exchange under the ticker MIND, moved from a minimum 10 percent position on Europe to a maximum holding of 32 percent, and increased its North American exposure to 60 percent from 54 percent.
The fund has returned 1.8 percent since it launched on Nov. 1, slightly underperforming the 2.6 percent gain in the MSCI World Index.
Part of its underperformance was because the artificial intelligence behind the ETF, developed by Korea-based Qraft Technologies Inc., didn’t initially include currency hedging and underperformed as the loonie strengthened against the U.S. dollar through January, Hawkins said.
Horizons added a separate currency-hedging AI to the fund at the end of January, which boosted its performance by 1 basis point of alpha in February, 14 basis points in March and an estimated 17 basis points in April.
“You don’t really know what you can expect until you’re running something live, and that is exactly what happened here,” Hawkins said.
Horizons is in the early stages of developing other robot-run ETFs, and sees particular opportunity in the fixed-income space where inefficiencies are rife.
Just don’t expect the robot to explain itself.
“We will never be able to explain why it either outperformed or underperformed,” he said. “We just hope it will consistently outperform so we don’t have to.”
©2018 Bloomberg L.P.