From veganism to #BLM: Big Data promises to transform niche ETFs

Ethan Powell says he has a vision. The founder of Impact Shares wants to get to the day when someone “likes” a pro-Black Lives Matter item on social media and then finds themselves presented with an advert for the non-profit’s NAACP Minority Empowerment exchange traded fund.

Mr Powell is not alone in thinking that day is not far off. Advances in big data and artificial intelligence look set to expand the reach of so-called thematic ETFs, which allow investors to focus on anything from Christian values, to water shortages or particular social issues.

“For thematics, I think it’s their time in the sun,” said Debbie Fuhr, founder of ETFGI, a consultancy, adding that the availability of alternative big data was transforming the research behind them, as well as providers’ ability to target these products.

These advances might partly explain recent improvements in their survival statistics. They still suffer a higher failure rate than the broader industry, but the differential is narrowing, perhaps because improved research and more focused marketing is improving their long-term outlook.

“Thematic investing through ETFs has found a large audience this year, breaking usual sector and country allocations,” said Anne-Valère Amo, head of ETF selection at TrackInsight. 

That sense that the products have caught the attention of their intended markets is also felt by Ms Fuhr who said that they were gaining resonance with a growing number of investors who were looking for new ideas.

Thematic ETFs are particularly prey to reputational risk if news breaks that reveals a constituent security is not aligned with the ETF’s broadly stated purpose. That could lead to divestment, which for a small ETF could be fatal.

“Size does matter,” said Ms Fuhr, who said most ETFs needed about $100m in assets under management to break even. Without building that scale even thematic ETFs built on potentially good ideas can find it hard to stay afloat. SLIM, the obesity ETF that tracked companies that offered health services to sufferers from obesity is a recent victim. A decision was made to close the fund before coronavirus, which is more risky for the overweight, took hold. Since then its index has outperformed the broader S&P Global Healthcare index. 

The need for good data is increasing the appeal of companies such as Truvalue Labs. Truvalue relies on alternative data and artificial intelligence to provide insights to fund managers and asset owners.

The company uses unstructured data from a wide variety of sources, including legal documents, media and government announcements. It follows trusted individuals on social media and scours their posts for references to studies or reports that could affect the value of the companies it is researching.

“You have to really mine the nuggets,” said Susan Lundquist, chief marketing officer at Truvalue, adding that the team had rapidly discovered that using uncurated social media had generated “too much noise”.

Claire Smith, founder of Beyond Investing and the VEGN ETF, is similarly cautious. The “Twitter fire hose” and other social media are monitored by human researchers for clues, she said. “They then go and find data to support what we’re hearing.”

The thematic ETF, which tracks an index of US large-capitalisation stocks screened according to vegan and environmental principles, also uses social media for marketing. However Ms Smith said they had decided to use it in a traditional way, posting on Beyond Investing’s channels, from where the information is then shared.

“We did try social media advertising in September last year, but found that the targeting on the social media platforms was not very good,” she said, adding that the algorithm had selected people on mentions of the word “vegan” but had failed to detect that some were talking about it from a negative standpoint. The marketing posts ended up attracting a lot of unsavoury posts, she said.

Also waiting for improvements to social media tools is Mr Powell who said Impact Shares was thinking of using artificial intelligence and natural language processing to scrape social media for mood statements. “It’s definitely something we’re thinking of doing,” he said, though he added that the researchers Impact Shares had been talking to had not yet developed sophisticated enough models.

Periscope Capital, the hedge fund behind one of the most audacious uses of big data for a thematic ETF, believes it has cracked one way of using social media as a very effective tool. It uses machine learning and AI analysis techniques to scour social media — specifically Stocktwits and Twitter — as well as other online sources for positive mood statements about large-cap US stocks and selects…