DWS, the manager formerly known as Deutsche Asset Management, is tapping into the latest research that has, inspired by work at some big universities, used the ‘wisdom of crowds’ phenomenon to identify above-average researchers. Artificial intelligence and other new technologies are enhancing the results.
In an interesting speech to the Investment Strategy Forum, produced by the Investment Innovation Institute, in Torquay, Vic, last week, Christian Hille, the DWS global head of multi-asset strategies, spoke of the pioneering work by IARPA (the Intelligence Advance Research Projects Activity) which commenced one four-year project, known as the “Good Judgment Project’, in 2011.
That project involved using a series of competitions for academics and practitioners which asked them to see whether people in a 3,000 universe (the crowd) did any better or worse than others, such as experts in their field or people with above-average intelligence, in predicting the future.
They found that over time none were better, although they discovered that an identifiable number in the crowd did consistently better than all of the rest, including the experts and people of above-average intelligence.
In an interview prior to his presentation at the conference, Hille said that the characteristics of a ‘super forecaster’ included having broad interests, being open minded and prepared to revise his or her own forecasts over time and that they tended to listen carefully to others in their own groups.
“We have been implementing some of the results of this work with our key investment teams,” Hille said. “We find that, with super forecasters, a team of about six or seven should be the maximum size and it should contain a diversity of personalities and capabilities. You should also have a ‘red team’, which critiques the others’ work and views.”
Super forecasters, and Hille’s colleagues at DWS, are aware of behavioural biases, such as anchoring, loss aversion and regret aversion. He believes that this new work in the research field allows portfolio managers to have a better understanding of how decision makers come to their conclusions. “At DWS we are implementing some of the result of this work in a careful way,” he said.
There is a system of assessing investment forecasts, known as ‘BRIER’, whereby the lower the better. It gives a random forecast a score of 0.50. It finds an ‘expert’ will deliver an average score of 0.39. With super forecasters, Hille says, the average score is 0.14.
Another interesting aspect of this is that the super forecaster will tend to do better at the extremes, where there may be either a high or a low probability of an investment invent occurring. They are also good where the odds on a particular event – from an interest rate rise in a particular country or a major geopolitical event – are less well-known.
But, notwithstanding the recognition and adoption of this pioneering work in research, DWS is basically a “fundamental investment house”, Hille says. “We use all the usual risk indicators and apply traditional research techniques. But we are also always looking to enhance those techniques.”
The DWS multi-asset strategy currently has a preference for emerging markets, both for equities – versus the US, say – and fixed income – emerging markets in hard currency sovereign debt versus high yield.
“This is a medium-term view based on a number of fundamental, valuation and technical arguments,” he says. But, even with the benefit of the very latest thinking on forecasting, Hille says that if the rest of the market does not see or believe it, you may be ‘wrong’ for some time. In a five to 10-year view, you will usually be ‘right’.
Hille timed his visit to Australia well. DWS was able to announce last week that it been awarded, by Colonial First State, a $170 million mandate to invest in Australian REITs. The local DWS listed property team of three in Sydney is led by Chris Robinson.