Canadian Investment Review

The role of data in equities management

Written by Yaelle Gang on Tuesday, January 29th, 2019 at 8:22 am

Big data business scientist presenting the conceptWhile the role of data in equities management is important, human expertise is still key, according to speakers at the CFA Society Toronto’s 2019 equities symposium on Jan. 22.

“If you want to beat the market as an active manager you have to believe you know something that the market hasn’t incorporated into prices,” said Sam Shapiro, vice-president of quantitative investment strategies at Goldman Sachs Asset Management. “Now let me be clear: We think there are lots of ways to do that. You can get on a plane, you can go visit some far-flung factory in China and try to figure out something no one else knows. But actually, if you look at the world today, the growth of data creates an opportunity to know something that other people don’t know.”

Shapiro highlighted that there are massive amounts of data that goes unanalyzed. “There is a massive opportunity to know something that other people don’t know, but it requires specialized skill,” said Shapiro. “It requires technology infrastructure investment.”

It’s important to understand that people make decisions based on information, not data, he added, noting technology and computing power is critical to translating data into information. This can be particularly challenging with unstructured data, which is experiencing large growth, said Shapiro, and people need to leverage new technologies to make sense of it.

One way to make inference from data that’s unfamiliar is machine learning, he said. “Machine learning is really the combination of computer science and statistics that allows computers to learn without explicitly being programmed.”

And in machine learning, there’s a lot of nuance, he added. “In our view, it’s really critically important that . . . the portfolio managers are the data scientists. We are one team. Data science isn’t a support centre, but intuition and technology should be embedded, intertwined, indistinguishable in some sense.”

Shapiro provided examples, including how data can be used to look at sentiment through natural language processing, noting there can be big differences when looking at one word versus multiple word phrases to understand the meaning. “Behind the layers and the buzzwords, there is a lot of skill and technique that’s necessary to translate data into useful information. The old adage ‘garbage in, garbage out,’ it’s true. But if you can translate data into useful actionable information then you have a competitive advantage.”

Also speaking about how data is shaping the investment industry, John Chisholm, client portfolio manager of global and international equities at Schroders, said with the rise of data, machine learning and artificial intelligence, the reports that people’s jobs will become mundane or redundant are a bit stretched. “It’s been our experience that AI in its current form is really only good for things that require about one second of thought,” he said. “We like to think of it more in the context of what we call IA, which is intelligence augmented.”

Chisholm noted the risk of attributing too much certainty to data that may not be fully understood. “It’s really about knowing that degree of uncertainty that enables you to have the right degree of conviction — and that’s really what it’s all about.”

Copyright 2019. Canadian Investment Review. All Rights Reserved.