Computers can win at chess, control driverless cars and predict the weather. But can they pick great dividend stocks?
Srikanth Iyer is convinced they can.
Mr. Iyer is the lead portfolio manager for the Horizons Active Canadian Dividend ETF (HAL), whose five-year annualized total return of 6.8 per cent (through Feb. 28) was tops in The Globe and Mail’s recent survey of Canadian dividend exchange-traded funds. It also beat the annualized total return of 5.5 per cent for the S&P/TSX Composite Index over the same period.
What makes HAL different? While most ETFs passively track an index, HAL employs an active approach that relies on sophisticated computer models to choose stocks to buy and sell.
John Heinzl’s secrets for building a powerful dividend growth portfolio
“Any stock that pays a dividend participates in our overall portfolio universe, and once it does it goes through the wringer and gets cross-compared against myriad components like growth, margins, earnings, value, credit, dividend yield and dividend growth, momentum, sentiment analysis, volatility, the whole gamut,” Mr. Iyer said in an interview.
A key component of HAL’s methodology, he said, relies on machine learning to predict – based on reams of historical data – the future dividend growth of a company and the probability of a dividend cut. The higher the expected dividend growth rate, and the lower the odds of a cut, the more attractive a stock is.
“Without a doubt dividend growth is one of the most significant factors for long-term returns, period,” said Mr. Iyer, managing director at Guardian Capital, the sub-adviser to the $34.9-million fund.
Staying on top of the news is important when assessing stocks, particularly when trying to determine investor sentiment toward a company. But no human eyeballs or ink-stained fingers are required in this case.
“We don’t read papers, we just have machines reading the papers for us. They read thousands of news items a day, and we actually have our own news sentiment aggregate for each of the stocks,” he said.
Another difference between HAL and index-tracking ETFs is that HAL’s turnover can at times be quite high, depending on market conditions. Over the past five years, annualized turnover has ranged from the low teens to more than 40 per cent. But that’s not because Mr. Iyer is a compulsive trader.
“Turnover is not dictated by me, it’s dictated by our models. If we see volatility … and a disconnect between fundamentals and the stock market, and our artificial intelligence model is predicting higher [probability of] dividend cuts or a lack of dividend growth, then those stocks will get exited out of the portfolio,” he said.
For all of HAL’s sophisticated stock-picking techniques, the 44-stock portfolio still holds many of the same companies you would see in most Canadian dividend ETFs and mutual funds. The top 10 holdings include familiar names such as Royal Bank of Canada (RY), Toronto-Dominion Bank (TD), Enbridge Inc. (ENB), TransCanada Corp. (TRP) and Brookfield Infrastructure Partners LP (BIP.UN).
But HAL yields just 3 per cent, which is low for a dividend ETF and reflects the presence of companies with modest payouts but high dividend growth rates. One example is Canadian National Railway Co. (CNR), which yields 1.7 per cent but has a record of annual dividend hikes stretching back more than two decades – including an 18.1-per-cent increase announced in January. “It is by far the best-in-class dividend growth stock within the industrial sector,” Mr. Iyer said.
HAL also includes some lesser-known names such as Boyd Group Income Fund (BYD.UN), which operates auto collision and glass repair shops in Canada and the United States, and is growing rapidly through acquisitions. “Historical [earnings per share] growth has been just absolutely stunning,” he said of Boyd. But the stock, which yields 0.4 per cent, “had a pretty big run-up … it is expensive, not cheap.”
Other low-yielding stocks with high earnings growth in the portfolio include convenience store operator Alimentation Couche-Tard Inc. (ATD.B), label and packaging maker CCL Industries Inc. (CCL.B) and software company Open Text Corp. (OTEX).
HAL’s management expense ratio of 0.78 per cent is higher than most ETFs, so the fund’s active management comes at a cost. Will the computer keep winning the dividend investing game? That’s impossible to say, but HAL’s five-year return suggests that, in this case, letting a machine pick stocks has added value.