Picture: Picture: 123RF/John Williams
Picture: Picture: 123RF/John Williams

South African investors who love flesh-and-blood stock-pickers should prepare for a step change as the investment industry is disrupted by automation, machine learning and artificial intelligence (AI).

Less human input could improve returns and reduce costs in the long run.

Vladimir Nedeljkovic, chief investment strategist at Alexander Forbes Investments, says the big move internationally to passive, and particularly smart-beta investments, is putting managers' profits under pressure.

Managers will in future earn their fees by selecting and combining investments and asset classes optimally and managing risks rather than managing money directly, Nedeljkovic says.

Local investment managers are lagging behind their overseas counterparts in embracing AI, but a few fledglings are offering AI-managed funds to individual investors.

Elize Botha, managing director of Old Mutual Unit Trusts, and Grant Watson, head of the customised solutions boutique that manages index and quantitative funds at Old Mutual, say the company is already using AI and exploring how algorithms can provide better insight into financial markets.

The group has been using AI for more than a decade in its Managed Alpha Equity Fund, which has delivered an average 11.6% a year over the 10 years to end-2018.

The fund uses AI to put the portfolio together, taking into account expected returns and market risks.

It uses AI to analyse the market and the leading indicators that identify which shares are likely to perform well.

Botha and Watson say AI will only become a mainstream way of managing investments when asset managers have evidence that it can add value in the construction of portfolios.

Lebo Thubisi, head of manager research at Alexander Forbes Investments, says before AI goes mainstream, managers' skills, heavily biased towards accounting, will have to shift towards data science.

Botha and Watson say data scientists, statisticians and programmers will all need to work together to navigate financial markets, but humans, including quants managers, will still have a big role to play in managing money.

In the meantime, AI is proving useful in the analysis of data, company statements, market sentiment and news about shares.

AI-managed funds available to local individual investors may have track records that are too short to be meaningful, but their ability to produce market-beating returns at below-average costs and with the benefits of greater diversification should be keenly watched, their managers say.

Sanlam Investments' Global Investment Solutions has used AI to manage its dollar-denominated UCITS (undertakings for the collective investment in transferable securities) fund, called the Sanlam Managed Risk Fund, since June 2017. The fund tracks the MSCI World Equity Index Fund through the Satrix World Equity Tracker Fund. A rand-denominated feeder fund is coming soon.

Ten percent of the fund is used to short futures to alter the exposure of the fund to the equity market. This exposure can range anywhere from 10% to 90% and is updated weekly based on the AI output, says David Itzkovits, CEO of Sanlam Global Investment Solutions.

Itzkovits says AI is better at predicting short-term market changes, therefore using the weekly AI signal to alter exposure gives a good balance between flexibility and the costs of trading.

The fund has over the past 18 months delivered an average return of 6.25% a year relative to the benchmark Morningstar category average of 0.2% a year and the MSCI world equity index's average of 4.1% a year, Itzkovits says.

He says the fund allows investors to participate in global equity markets while trying to minimise the market draw-downs (the gap between a fund's highest return and its lowest).

The Sanlam Managed Risk Fund drawdown was -8.5%, compared with -11.8% on average for its peers and -18.4% for the MSCI world index.

The second AI-led fund in SA was launched just over a year ago by a new boutique manager Nmrql (pronounced "numerical"), on Sanlam's unit trust licence.

The Nmrql Balanced Fund invests across the asset classes and, though still in its infancy, is using exchange-traded funds to get exposure to some classes such as offshore equities.

Investment manager Thomas Schlebusch says AI is used to determine the fund's strategic asset allocation based on analysis of which asset classes perform best at each stage of the business cycle.

The software, unlike traditional active managers, doesn't have a particular style but uses whatever data will give it an edge in selecting securities, Schlebusch says.

This includes companies' financial statements, news and market reports that indicate market sentiment, and weather data that informs decisions on commodity and food producers, he says.

Recently Glacier launched the third SA AI fund - the Glacier AI Flexible Fund of Funds, invest across the asset classes and uses AI to adjust the international exposure from 0-30%.


Understanding the jargon

Passive investments are managed without human managers actively picking the shares, bonds or other instruments.

Index funds track a market index as closely as possible by investing in the same instruments in the same weighting as, for instance, the all share index.

Smart beta investments select and weight instruments in an index according to rules aimed at capturing "factors" proven to deliver market-beating returns. For example, the value factor or the small-cap factor.

Quantitative or quants investments are driven by numerical methods rather than human judgment.

Artificial intelligence is used to design algorithms to manage investments and to improve algorithms and data analysis in order to achieve a set outcome.