Belvoir Capital’s CEO Steffen Bauke said he expects more investors to start using big data as a way to avoid personal bias, having launched a fund driven by news analysis.
‘It’s all about big data – capturing it and then getting what you need out of it,’ he told Citywire Switzerland.
‘As an analyst, you’re not able to look at everything without bias. You tend to overallocate towards your bias. Sentiment news analysis is a different approach, getting away from the personal bias that might be experienced by a CIO.’
Zurich-based Belvoir Capital launched the Sentiment Sicav Europe 600 fund in February 2017 and believes it to be the first equity fund that is managed based on the sentiment analysis of financial news. It is based on the STOXX Europe 600 index and invests in the 15 largest companies in 15 different sectors and in ETFs of the four remaining sectors.
The tool captures between 20,000 and 40,000 news articles per day, feeding them into an application programming interface (API) that analyses market opportunities and risks based on the media discourse.
Sectors are labeled with green, yellow or red signals - positive, warning or negative.
Bauke said his research team was initially sceptical of the API’s reliability. ‘Our portfolio management team is more bottom-up when it comes to fundamental research. They didn’t initially believe the signals would work, but over time they’ve been watching them more and more.’
Since implementing the tool, Bauke said his team has been able to make better allocations amid the increased volatility.
‘Once a sector turns negative, we sell it. But, if two thirds of the sectors are positive, then we maintain our overall market exposure,’ he explained.
‘In April, after the markets went down, we reduced our exposure to the market from 100% to 40%.’
Bauke added that Belvoir is looking into the possibility of launching more funds in the future, specifically an emerging market fund and a micro-cap fund, which would sit alongside the sentiment news analysis fund and a global allocation fund.