Machines worried about global markets

MACHINES, apparently, are concerned about the state of global investment markets.

South Africa’s machine-learning-powered asset manager, NMRQL Research, said this week that the Uncertainty Metric of its investment algorithms was at levels seen only a handful of times historically – for example, just before the 2008 stock market crash. According to Stuart Reid, the chief scientist at NMRQL Research, this uncertainty has resulted in the NMRQL Balanced Fund reducing its exposure to equities significantly and increasing its exposure to cash.

He says that examples of such periods of uncertainty include:

• August 2006 – after the United States Federal Reserve changed its interest rate policy;

• February 2008 – just before the collapse of Bear Stearns;

• November and December 2014 – when the Federal Reserve was again debating rates;

• August 2015 – just before the 2015 market sell-off; and

• January to April 2016 – the sell-off in Asian markets.

Reid says that during their simulations and tests, this aggregate measurement has proved to be a leading indicator of turmoil in global financial markets. “So, the statistically high values we are seeing are cause for concern,” he says.

“When the algorithms learn how to invest in this ‘new normal’, they will redeploy the cash. Whether that occurs after a further correction, crash, or rally is still to be seen,” Reid says. “We cannot be sure of how these metrics relate to future returns of the market or what has caused them to spike, but what we can say with certainty is that the models acknowledge this data and are currently very uncertain about the future direction of the market.” – Staff Reporter