Skip to main content
big data

Emrah Turudu/Getty Images/iStockphoto

Google and other Internet search engines sort through reams of data and bring it all to your screen in an instant. It's hard to imagine life without them. So why couldn't they be used as tools in the investing game?

At least one study has indicated that Google Trends – a tool that compiles search data – can offer an edge in managing a portfolio.

Conducted in 2013 and published in Scientific Reports, the study examined the frequency of Google searches using dozens of finance-related words and their relationship to movement of the stock market.

One search term, "debt," was particularly accurate in predicting the Dow Jones Industrial Average. More searches for the term were a bellwether for a faltering economy; fewer searches hinted at more positive outcomes.

The hypothesis was that Google Trends could offer insights into investor herd behaviour. It turned out the strategy worked. Critics, however, say that investors who try to capitalize on this may end up being disappointed.

That's because while the study found a correlation, causality was not proven, says Graham Westmacott, a portfolio manager with PWL Capital in Waterloo, Ont. "More data analysis doesn't help if the data does not have a causal relationship with what you are hoping to predict."

Although many attempts have been made to find predictors of future market returns, almost none do so consistently over time, says Mr. Westmacott.

He suspects the Google Trends strategy would eventually fall short, too. To prove his point, he cites two studies. One in particular examined the findings of the Google Trends research, asking whether search terms unrelated to finance might also correlate to higher returns.

"It found that if you take 'bone cancer' or 'moon patrol' you would have ended up with a better strategy than using 'debt,'" Mr. Westmacott says. "And this just points to the random nature of this kind of approach."

In general, any strategy that uses past data to predict future performance is problematic, he says. Results that seem to indicate causality wilt under close statistical analysis.

Here Mr. Westmacott points to another study, this one by Vanguard Research, which found only one investment indicator – the Shiller price-to-earnings ratio – has predictive value.

"But even that only explained 40 per cent of future price returns," he says of the Shiller notion, a 10-year average that measures how much investors are willing to pay to own a company for every dollar of earnings. "So that leaves 60 per cent totally unexplained by other factors."

Still, many investors swear by strategies that use indicators to predict price movement. The largest group are those using technical analysis, which seeks to predict prices based on past pricing data.

"Traders of this discipline are often broken down into different categories depending on the time frame they focus on," says Toronto-based blogger Daniel Santos of TrendVesting, which examines strategies using Google Trends.

The Google Trends strategy is just another permutation, he adds, and its premise of identifying behavioural trends is not novel, either. "Investment legend Peter Lynch employed this same type of information at a much smaller scale," Mr. Santos says.

Mr. Lynch used "local knowledge," in other words keeping tabs on popular products and services. The more people who use them the more likely the company producing them is a good investment.

"Google Trends data is like Lynch's local knowledge but at an infinitely greater scale," Mr. Santos says.

He also concedes the limitations of such a strategy. Significant spikes in Web searches for a company you're following could be the result of something totally unrelated to its future value, he says.

Moreover, Mr. Santos says investors should not use such data to replace an investment strategy. It is meant to be supplemental.

A holistic approach – a blending of strategies – is common among investors who use technical analysis along with other methods, says Keith Richards, a portfolio manager based in Barrie, Ont., with ValueTrend Wealth Management and Worldsource Securities. His firm combines technical analysis with fundamental analysis of a company's revenue, cash flow and debt.

While retail investors may be tempted to toss "buy and hold" aside in favour of trend-following alternatives, Mr. Richards warns that can be recipe for losses. Many inexperienced investors often glean one or two ideas from very complex strategies and use them as their entire strategy.

"And even if it works initially, it sets you up for getting slaughtered eventually."

Mr. Westmacott also sees danger in putting too much stake in these concepts, adding that people tend to look only for information confirming our biases.

It's an issue summed up well by a quote from economist Ronald Coase that statisticians know well, he says. "If you torture the data long enough, it will confess."