Artificial intelligence (AI) has begun to infiltrate the world of Canadian ETFs with the launch of Toronto-based Horizons ETFs Management (Canada) Inc.'s Horizons Active A.I. Global Equity ETF on the Toronto Stock Exchange on Wednesday.
The ETF employs an investment strategy run by a proprietary and adaptive AI system that makes investment decisions based on patterns discerned from split-second analysis of millions of pieces of data. The system is designed by humans, but is then free to make its own decisions within preset parameters.
"AI is an automated learning machine," says Steven Hawkins, president and chief investment officer of Horizons. "It continuously evaluates data and will make decisions as to where to allocate from a geographical perspective, but without the emotion or biases that apply to an individual portfolio manager who is making investment decisions."
The ETF, which is subadvised by Horizons' parent, South Korea-based Mirae Asset Global Investments Co. Ltd., will choose its holdings from a universe of 32 North American-listed ETFs offering a variety of global exposures, of which four are sponsored by Horizons.
Essentially, the AI ETF is an ETF fund-of-funds, with AI responsible for changing asset allocations and for overweighting regions, countries or indices that are expected to outperform and underweighting those expected to underperform. The maximum number of ETFs held in the portfolio will be 20 at any one time; the minimum is five.
"The AI system applies machine learning which constructs artificial neural networks, essentially training a computer to recognize patterns and make decisions, much like how the human brain operates but at hyper fast speeds," says Horizons in its background material. "The AI system is completely autonomous in that it is a learning computer that effectively teaches itself how to invest and independently makes investment decisions based on past experience."
The four main regional categories covered by the ETF include North America, Europe, Emerging Markets (including China) and Asia Pacific. The ETF has a management fee of 55 basis points.
The AI "neural network" system was designed by Qraft Technologies of Korea. Although the AI analyzes historical data points, the allocations will be based on AI's predictions of where the best potential for future returns lie. The portfolio will be rebalanced monthly, taking into account the regional, country and individual index exposures of its underlying selection of ETFs.
The ETF must adhere to some risk controls. For example, the maximum allowable exposure to North America is 65% and minimum is 30%. The maximum emerging markets exposure is 15% and the minimum is zero.
"We are the first AI portfolio manager of a global equity asset allocation strategy," says Hawkins. "It's on the cusp of innovation. Although some of the larger ETF providers stick to the large, index-based strategies, Horizons has been an innovator and leader in introducing new ETF strategies. We are not the Wal-Mart of the ETF world."
The Horizons AI-based ETF is not the first in the world, though. For example, an AI-based ETF was launched last month in the U.S The AI Powered Equity ETF is sponsored by EquBot of San Francisco, a startup from the IBM Global Entrepreneur program. The ETF uses IBM's Watson technology to analyze data for approximately 6,000 individual U.S.-listed equities, and then selects up to 70 companies with the best upside potential.
Photo copyright: sakkmesterke/123RF