AI, let’s go – ETFs go robotic

by , 29th November 2017

On the global freelancing website Upwork a job was advertised recently that gives us a hint as to the direction that ETFs are heading when it comes to the adoption of technology.

The posting was in the data science section and asked for applicants with a mathematical or statistical career background who were needed to work on an “ETF forecasting algorithm.” The posting continued: “The job involves building an algorithm that uses proven economic indicators to forecast the probability of future return for a range of ETFs.”

The poster attached a document that showed how simple indicators including the price-to-book ration and Shiller CAPE can be used to forecast stock market returns. He then asked applicants to turn their minds to producing similar algorithms that could then be used to help forecast probabilistic outcomes across a range of equity fund types.

Modestly, the author added that he had documented one way of achieving this but that “it is most likely not the best approach.”

David, meet Goliath

Although the poster deserves top marks for effort, in truth the best approach to employing artificial intelligence is not to be seen via a crowdsourcing gig economy post. Instead, we can rely on the world’s biggest asset managers to pitch in with their own ideas on how to take advantage of the AI revolution.

On the face of it the fund is nothing new: according to the filing made in mid-November with the SEC the iShares Evolved fund will be an actively-managed tech fund. Nothing to see here then.

That is except that the filing goes on to say that the classification process will use data analysis capabilities including machine learning, natural language processing and clustered algorithms. The filing suggests that such a system will allow companies to be classified into multiple sector rather than being assigned solely to a single sector, reflecting their multi-dimensional nature. “Sector constituents are expected to evolve dynamically over time to reflect changing business models,” it added.

In achieving this aim iShares Evolved is using the available technology to better identify companies using the same innovations to push their businesses forward. It is not alone in this idea. The AI-Powered Equity ETF (AIEQ), powered by IBM Watson alongside proprietary algorithms built by a company called EquBot, was launched in October.

The fund will invest across the broad market indices, using AI to compare thousands of US companies on a daily basis to optimise the portfolio. To do this, the filing with the SEC said, it uses AI to analyse information from more than 1 million regulatory filings, quarterly results releases, news articles and social media posts, all on a daily basis.

It is a reminder of the commonly repeated assertion that more data has been produced in the past two years than in the whole of human history preceding it.

Skinning the cat

In a report issued in September and looking into the digital challenges facing ETF sponsors and service providers, analysts at PwC suggested that exponential advances the ability to gather data combined with the increase in the amount of sources as well as the compound increases seen in computing power and advances inn predictive analytics and artificial intelligence were sure to have an impact on ETFs, particularly in what they term the “active 2.0” space.

The report provides an answer for those who believes that AI-powered ETFs are in any way a gimmick. Boiling down what eh advances made in AI might mean for the asset management sector, the report suggests that AI is simply complex mathematics tasked with solving while asset management is largely based on prediction.

Hence, the report’s authors point out, with predictive technology getting exponentially cheaper and better, it will lead to three outcomes. More problems will be recast as predictive problems; predictions historically made by humans will be made by computers; and the use of prediction will increase.

It is an exciting technology and one that as well as driving the technology behind ETFs also provides a fruitful feeding ground for other funds looking to profit from the rise of the technology itself.

I Robot

One such example comes from the thematics range of product offered by the Canvas division of ETF Securities that was recently sold to L&G. It is the Robotics and Automation UCITS ETF and to source the index – and the expertise to put it together – ETF Securities turned to advisers Robo Global.

Richard Lightbound, chief executive for EMEA at Robo Global, says this is not a “typical passive index.” Robo Global has a wide network of experts which it can tap into, giving it access to the newest developments and ideas. “You have start from the top down to understand the sector,” he says.

He says the company is very excited about the future economy that AI and robotics generally are set to deliver. “This is about enabling technologies,” he says. “The picks and shovels our AI and robotics. It is about machines providing efficiency and productivity.”

Though the end product of Robo Global’s work is wrapped up in a passive structure “it is almost a hybrid process to seek out and find the potential index members,” Lightbound adds.

“Initially we thought it might be an automated process,” he says. “But our sense is that it is too early to pick the winners. We have got better, but when we began we would struggle to pick what would be the top-performing sub-sectors, let alone the top performing companies within that.”

Howie Li, chief executive of Canvas at ETF Securities, points out that this is not a standard tech equities investment. “Access to such deep understanding of the robotics industry and how artificial intelligence is driving the growth of selected companies forward is why the portfolio is so different to other products in the market,” he says.

As for whether AI processes will eventually mean that we come to recognise robo-fund manager as much as we do robo-advisors today, Li remains to be convinced. He points out that artificial intelligence, machine learning and quantitative investment strategies are based on data and codes and knowing what to do with those drives fintech and is one of the primary reasons why

investing is so much quicker today. But he adds the caveat: “That being said, what is clear is that end investors generally will still want to engage in human interaction even if there is financial technology that automates investing behind it.”

ETF Issuers Featured in this Article