Algorithms are everywhere

They can be useful in our daily lives but do they know too much about us? Many companies may not want you to know the answer.

They have names such as HardBacon, Ambo, Wealthica and Covera. These are among the dozens of startups across the country that are sparking a revolution in the financial technology sector (a.k.a. fintech). Their secret weapon? Algorithms, paired with financial analysis. Their goal? To enable individuals to build stock portfolios on their own or find the insurance policy that best suits their needs. But algorithms are by no means exclusive to the world of finance. They shape vast areas of modern life, influencing our habits and telling us what to read, watch or buy. One thing is for sure, however: while you probably don’t know much about them, they sure know a lot about you.


“A financial adviser can evaluate a portfolio to determine the degree of diversification and exposure to risk,” says Julien Brault, CEO of Quebec startup Hardbacon. “What we can do is transform that know-how into algorithmic code.” Founded in 2016, the company has launched an application that “makes it easy for anyone to analyze investment portfolios and invest in the stock market.”

Ambo Technology is part of the same new generation of algorithmic trading software enterprises. The firm’s president and cofounder, Pascal Leblanc, combined mathematical models with investment strategies to develop artificial intelligence-based software that optimizes these strategies. The funds invested through the software generated returns of 87% in 2016 and roughly 125% in 2017. “There is no human intervention involved,” says Leblanc. “Better still, the software can carry out transactions 24/7.” It’s a similar story at Hardbacon, which is geared toward independent investors. “We cut out financial middlemen to give information directly to investors,” says Brault.

More and more, portfolio management decisions are made using algorithms developed by programmers, who are growing in ranks. Today, a computer science graduate is just as comfortable in the financial industry as a financial expert. Croesus, which created a software platform for portfolio managers at large financial institutions, is no different: more than a quarter of its 175 employees are computer engineers or programmers. “They are teamed up with our financial analysts,” says Patrick Chassé, the company’s director of software development. “One group creates the financial algorithm rules, while the other integrates them into our software.”

Launched in 1987, in the early days of microcomputing, Croesus is a fintech pioneer. The Quebec SME’s founder, Rémy Therrien, was an engineering physics student at Université Laval at the time. He was working at a computer store for the summer when he got a call from brokerage firm Lévesque Beaubien Geoffrion. The brokerage, which would later become National Bank Financial, wanted a software tool to manage its clients’ accounts. For Therrien, that call was the springboard of his business. Thirty years later, the company’s software manages assets valued at $1 trillion.

Today, even traditional financial institutions are upgrading to algorithms and new financial technologies. Three years ago, National Bank launched InvestCube, a platform it promotes as a smart, low-cost investment management system. “More and more clients want to take control of their portfolios,” says Robert Girard, senior manager of business development at National Bank Direct Brokerage. InvestCube allows them to do just that. Based on their profile (conservative, growth or balanced), investors can access a ready-made portfolio composed of exchange-traded funds, which InvestCube will automatically rebalance up to 12 times a year if necessary (a portfolio is rebalanced whenever a security deviates by 10% or more from its target allocation, based on the investor’s profile). Since its inception, InvestCube’s six portfolios have generated returns ranging from 3.36% to 15.99%.

Algorithms have taken over other industries as well, including the insurance sector. Since 2013, Desjardins Insurance has been offering the Ajusto telematics program to its Quebec and Ontario clients to help them monitor their driving habits. Provided as a voluntary and free smartphone app since 2015, the software measures driving smoothness (fast acceleration, hard braking and hard cornering), speed, the time of day a driver is normally at the wheel and distance travelled. In return, program members can save up to 25% on their insurance premiums; since the program’s inception, they have seen a 10% to 12% reduction on average. Desjardins says that the data collected by Ajusto cannot be used to penalize insurance clients. “There are very strict privacy policies in place stipulating that the data will never be used to cancel an insurance policy, decline renewals or raise premiums,” says Desjardins spokesperson Valérie Lamarre.

However, Alain Tapp, associate member of the Montreal Institute for Learning Algorithms, warns that there is always a risk that the information could be used against people. “Those who don’t take part in these kinds of programs could pay the price,” he says. He isn’t the only one with concerns. Telematics has an increasingly large presence in the North American insurance industry. In fact, the Société de l’assurance automobile du Québec (SAAQ), Quebec’s driver licensing agency, once hoped to track driver habits. However, faced with opposition from wary drivers who feared their data would be shared with a government body, SAAQ reversed its plans. “We concluded there wouldn’t be enough participants,” says SAAQ spokesperson Gino Desrosiers.


Finance and insurance are not the only industries capitalizing on the exponential growth of algorithmic power. PricewaterhouseCoopers developed a data analysis tool that uses algorithms and statistical software to monitor struggling businesses over a period of 20 years or more. The data, combined with sector-specific information, can be used to identify trends and prepare forecasts.

“The growth — and even survival — of many businesses depends on their ability to optimize their equipment and infrastructure,” says Ramy Sedra, partner and data and analytics consulting leader at PwC Canada. “Algorithms can play a key role in improving their business processes.” In the era of big data, companies now have a considerable volume of real-time data at their disposal. Sedra points out that Amazon and Alibaba, two e-commerce giants, revolutionized the retail industry by using algorithms to determine their customers’ purchasing habits and recommend more products to them.

With industry 4.0 — also referred to as the fourth industrial revolution — now upon us, algorithms have even made their way into the manufacturing sector. These days, digital technologies allow manufacturers to create smart factories by bringing together the internet, wireless sensors, software and other cutting-edge technologies to optimize production or even manage it remotely. According to a 2017 BDC study, companies that were “early adopters of digital technologies have increased their productivity, reduced costs and improved the quality of their products.” The study also reports that adopters of digital technology are almost twice as likely as non-adopters to forecast annual revenue growth of 10% or more over the next three years.

So is industry 4.0 and its use of algorithms threatening jobs? “Historically, industrial jobs have always been automated and replaced, mainly to address labour shortages,” says Tapp, who points out that some office jobs are just as much at risk. “In the past, you would go to a travel agent to plan a vacation. Now, algorithms will book your flight and accommodations,” he says. The financial industry is experiencing the same phenomenon, although Girard thinks financial advisers and portfolio managers will always have a role to play. “The added value of an adviser is exactly that: advice,” he says. “But since clients now have access to the same information, advisers who merely buy and sell stocks will become few and far between.”


Algorithms came out of the shadows thanks to Google and other internet giants such as Facebook, Amazon and Netflix, which founded their business models on these calculation formulas and built massive databases along the way.

“Product discoverability has been a game changer. Before, consumers had to find the cultural products they wanted to read, watch or listen to on their own,” says Jonathan Roberge, Canada research chair in new digital environments and cultural intermediation at the National Institute of Scientific Research in Quebec City. “Today, it’s the technological platforms that find consumers.” According to Roberge, we are clearly living in an algorithmic culture. Netflix, with its 109 million users in more than 190 countries and with more than 140 million hours of movies, TV shows and documentaries watched daily, is the world’s leader in online entertainment. It’s also the ultimate algorithmic machine, says Pierre Bélanger, a professor at the University of Ottawa’s department of communication. Bélanger is especially interested in how technology is being integrated into the media landscape.

“Netflix is powered entirely by algorithms. It’s fuelled by the billions of footprints we leave all over the web,” he says. “Once it gets to know us better, it recommends content that reflects our preferences.” Bélanger uses “netamorphosis” to refer to these powerful algorithms that influence our habits. “I worry that machines have begun to think for me,” he adds.

The issue is all too real: should we be concerned about the rapid proliferation of algorithms or their growing impact on our daily lives? “Through information categorizing, personalized advertising, product recommendations, behavioural targeting and location tracking, supercomputers are becoming more involved in our lives,” says Dominique Cardon, director of the MédiaLab research centre at the Institut d’études politiques de Paris. In his 2015 book, À quoi rêvent les algorithmes (what do algorithms dream about?), he contends that very few everyday habits, purchases, trips, or personal or professional decisions are not guided by a computing infrastructure.

“Today, it’s so much simpler and easier to consume cultural products, people don’t even realize that algorithms are influencing them anymore,” says Roberge. “But this is no reason to become paranoid and see it as a systematic intrusion into our private lives.”

Tapp agrees. “We shouldnt worry, but we do need to be vigilant,” he says. And with good reason: although algorithms have made it much easier to access information, we should be wary of misuse. This is especially true for internet users for whom social mediaespecially Facebook and its two billion usersis their main source of information. Some of the shortcomings of algorithms came to light during the 2016 US election, when a multitude of fake news stories circulated on social media. It is even alleged that Russian-backed agents posted content to help Donald Trump get elected. “The danger of social media only showing us content based on our interests is that it reinforces our opinions rather than exposing us to different points of view,” says Bélanger. “The predictive functions of algorithms are their biggest downside. The more people like something, the more of it theyll get.”

Another risk: in the US and elsewhere, police departments use algorithms to prevent crime. The best-known crime prevention program software, PredPol, which was created by a US university-based startup, alerts police patrols of possible break-ins, thefts, assaults and homicides.

But while these programs can help thwart terrorist attacks, Roberge thinks they can also lead to arbitrary arrests based on racial profiling. “We should always be mindful of possible abuses,” he says. “We cant pretend that algorithms are neutral. Theyre developed by humans and can therefore be biased.” The researcher worries that the technology will evolve faster than our ability to understand it.

Facial recognition is a clear example of this risk. Used by Apple to unlock the new iPhone X and currently being implemented in Canadas major airports (including Montreal, Toronto and Vancouver), the technology also raises questions about privacy and the protection of personal information. “If we were able to identify anyone at any time, and provide access to biometric data, then there would be no more anonymity,” Roberge says.

But lets be clear: we are by no means living in the age of Big Brother. And although algorithmic systems are not malevolent in and of themselves, safeguards need to be in place. “Im not saying that algorithm giants have evil intentions,” says Roberge. “Still, these platforms contain an enormous amount of personal information that the great dictators of the 20th century would have loved to have had at their disposal.”


Algorithms are nothing new. They date back to antiquity, and were initially developed using variables in mathematics. The Euclidean algorithm, conceived about 300 BC, helped mathematicians determine the greatest common divisor of two integers. The word “algorithm” even got its name from a ninth-century Persian mathematician, Muhammad ibn Musa al-Khwarizmi, who is considered the father of algebra. His name was later Latinized as Algoritmi, which was itself influenced by the Greek word arithmos (number).

Initially used to solve arithmetic problems through a succession of operations, today algorithms are written in code to be understood primarily by computers, whose power has resulted in an increase in their use, in recent years, to compile and quickly analyze billions of pieces of data.

Invented about 20 years ago by Google cofounders Larry Page (right) and Sergey Brin, the PageRank algorithm is the source of the powerful search engine used by billions each day.

Facebook’s creation can also be traced back to an algorithm, which was originally called EdgeRank and later renamed Newsfeed Ranking Algorithm. It sorts through billions of pieces of content and chooses which ones to post on its users’ news feeds, sparking discussions and debates on fake news along the way. US internet guru Eli Pariser wonders why social media platforms such as Facebook and Twitter do not provide users with effective tools to filter their own feeds. “Right now, algorithms control what we see but we can’t control them,” he says. “Users are therefore trapped in a filter bubble.”

Algorithms are also said to have been behind the Flash Crash that shook Wall Street in May 2010, wiping out 10%, or US$1 trillion, of stock market value from the New York Stock Exchange. A British trader allegedly used an automated trading program to place a high volume of fake sell orders that pushed down prices. He then quickly cancelled the trades to create the illusion of abundant supply.