With Disney+ a month away from its US launch, Disney on Monday revealed the streaming platform’s entire line-up of movies and series. In a…
A Financial Times article published on Tuesday (20 March) describes interest in quantitative investing (quant investing) — the use of powerful computers and artificial intelligence (AI) to search markets and big data sets for patterns that can be used by trading algorithms — as one of the biggest trends in the money management industry.
The article states that last year Larry Fink, the CEO of US asset manager BlackRock fired seven under-performing fund managers and moved the money they were managing to a “computer-powered” investment division called Systematic Active Equities (SAE).
This division now reportedly outperforms the other more conventional stock pickers.
A blog post on Old Mutual Wealth’s Knowmore website estimates that this global trend will pick up in South Africa over the next three to five years.
Enter NMRQL Research, a Stellenbosch-based asset management startup that was founded in 2015 by CEO and director Thomas Schlebusch along with venture capitalist Michael Jordaan as an investor and chairman.
Although the startup is an independent company, it has a partnership with Sanlam. Schlebusch explains that because of regulations, when running a collective investment scheme “you have to do it through a regulated platform or a management company”.
NMRQL Research was founded in 2015 and has a partnership with Sanlam
Last year the startup launched the NMRQL SCI Balanced Fund — a unit trust fund that uses machine learning to research and analyse stock picks.
The startup claims that this is the first machine-learning powered asset manager in the country.
“There is very little use of artificial intelligence in the South African financial services industry,” says Schlebusch.
‘Crowding in the market’
“The (general) mindset is stuck with models people are used to, which they learnt in university,” he says, adding that “most investors are fundamentally driven, they use financial statement data”.
Traditionally, investment managers look at metrics like profit, debt and earnings as well as qualitative factors like strategic plans.
This, he says, has resulted in what he describes as “crowding in the market”.
He adds that globally, people are waking up to the opportunities of data science and big data. Around the world, other asset managers like Goldman Sachs also have quant divisions.
“From my previous experience, I have seen smaller hedge funds using AI and big data, ” he says, adding that traditional hubs are still “locked in a philosophy that is fundamentals driven”.
‘So much data around’
Schlebusch says when it comes to data sources, there’s “so much data around”.
“The challenge is that the data is not coming from Bloomberg screens or the Reuters application. You need different tools to connect that data into usable information,” he says.
He says although the tools NMRQL Research uses are open-source platforms and open APIs “our industry is slow to pick up and use them”.
How it works
In an earlier statement, NMRQL Research chief engineer and partner Stuart Reid (pictured below) explained that an algorithm tunes parameters until it is able to produce future returns.
Both Schlebusch and Reid believe machine learning algorithms are better equipped to process information for investment analysis than humans who have an innate sense of bias and subjectivity.
Reid also explains that traditional investment methods are not only flawed due to this inherent human bias, but because they only analyse structured data — like indices, company financials and currencies.
“The real power of machine learning is that it can make sense of unstructured data — like tweets, blog posts, news sentiment — and use this to inform stock picks,” he adds.
Once NMRQL Research’s algorithms uncover the hidden patterns, says Schlebusch, it then exploits them to forecast returns across all asset classes and markets, resulting in steady, long-term growth of capital and income
What makes this approach stand out?
By employing machine learning, Schlebusch says the startup is able to “process more data and glean more relationships”.
He adds that NMRQL Research is able to process things like sentiment and photos and in doing so get “better monitoring”.
He also says that NMRQL Research’s approach is “extremely scalable” and does not require a lot of investment managers.
‘In the pack’
So how is it working out for the firm? When asked by Ventureburn how the NMRQL SCI Balanced Fund is performing and how it compares to more conventional unit trusts, Schlebusch says “it is still early days”.
“It’s less than five months since the company started using the process,” he says, adding that the company is “in the pack and not shooting lights out”.
“We need a little more time to judge if it is luck or skill,” he adds.
Featured image: NMRQL Research founder, CEO and director Thomas (Tom) Schlebusch (Supplied)