We look at the rise of ‘robostocks’ and algorithmic trading, and consider the repercussions on financial markets.

An old investment adage mockingly states that “a failed trade becomes a long term investment”. The idea behind it is that if a security is bought and underperforms, investors tend to keep it until it eventually becomes profitable, in what Nobel laureate Daniel Kahneman described as the “endowment effect”.

Nowadays, however, the “trade” and the “investment” are mostly executed by different entities. Modern day trading is less dominated by human biases and more by algorithmic trading, computers trading on parameters set by mathematicians and CERN-educated physicists rather than fund managers driven by fundamentals.

The theme of a robot-dominated dystopia is prominent in fiction and our culture, with series such as “Westworld” and novels like Philip K. Dick’s wonderful “Do Androids Dream of Electric Sheep?”,  and has been gaining prominence in the past few years. There’s good reason for this. 20 Years after a program called Deep Blue defeated World Chess Champion Garry Kasparov, another AI program called “Libratus” recently beat four of the world’s best poker players in a gruelling 20-day tournament.

Trading robots

Speed, hedging needs and best execution directives in multiple exchanges have taken precedence over cash flow and capital expenditure analysis. This is partly reflected in the huge growth in ETF assets, coupled with a decline in long-only fund assets over the past few years. Studies estimate the amount of algorithmic trading to be anywhere between 50% and 85% in the US, and a bit lower for Europe, rendering “Algos” a dominant market force. In a recent report, the Bank of International Settlements took a good look at the Sterling “Flash Crash” on October 7th 2016, during which the UK’s currency depreciated versus the Dollar 9% before retracing the move. The report cites, among other factors, the presence of staff “with less expertise in the suitability of particular algorithms for the prevailing market conditions”. What is particularly worrying is that currency markets are very deep and difficult to manipulate, which means that the fault probably lies not with one but with a large number of algorithms, set to react roughly the same way, a sort of computer “groupthink”.

Gavekal’s Anatole Kaletsky recently also cited the algorithms as key to explain buoyant market reactions after major events, like Brexit or the US election, the outcomes of which had been predicted as catastrophic by experts. He even went a step further, juxtaposing active management (like client portfolios and mutual funds) to passive management, which essentially means computers deciding instead of humans.

Insights from the GFC

It is, by now known that the 2008-2009 Global Financial Crisis did not begin with the demise of Lehman Brothers but with large size quantitative hedge funds going bust one year earlier, in August 2007, as automatic trading failed them. By their nature, algorithms are set to respond to a set of predictable circumstances, as they draw on past experience, and then use probabilistic models to assess future conditions. Also by design, they are short term. The longer one projects into the future, the more uncertain that future becomes. Robots can deal with probabilities of short term events but would find it harder to deal with the consequences of important market shifts, including the selloff by other robots. Thus, while “algos” may benefit from the Dow dropping 800 points in the hours after the US election, they could find it harder to respond to something more unpredictable and less binary, like for example China dumping US Treasuries in the open market, as a result of an escalating trade war. Policy changes, a shift in earnings trends, new regulations, anything “new” in nature is very difficult to program in advance. So, less policy guidance from central banks, for example, could level the playing field between fund managers and mathematical formulas.

However, unless Janet Yellen wakes up one morning to the belief that she doesn’t want to talk to the markets any more, or remove the Fed’s implicit protection, the interest rate environment is more predictable, allowing robots the edge.

What next for fund managers?

Fund managers now find their comfort zone in the “longer game”, where assessing the improbable is more than just about mathematical assumptions. In virtually 95% of the meetings we attend, we hear about the importance of security selection and the implementation of long term strategies. When we assess absolute return funds, the closest thing to a hedge fund with daily liquidity and within the UCITS frame, we are well aware that they best operate within predictable levels of volatility and could underperform if the market shifts completely, either to the upside or the downside.

A long term perspective

Could the next crisis then come from computers, as “Big Short” writer Michael Lewis hinted in his “Flash Boys”? Maybe. But, quite like other humans, we play the long game in our portfolios. We think of macro trends and how they may drive earnings. We ignore short-termism, not because we have no choice, but because the client’s objectives are set over the long term, and that is when they, and we, feel more comfortable. Thus portfolio owners should maintain their long-term perspective. As for investment managers and economists? We are driven to focus less on the repercussions of “known unknowns”, such as a surprise number in US payrolls, and are driven more towards focusing on the repercussions of “new events”, confident that humans can identify cyclical and structural breakpoints in the markets, even if computer trades persist in a different direction.