Artificial Intelligence: Is It Good for the Markets?

Technology is changing many things that impact our lives and has certainly made us more productive.  As to the stock market, we have seen a deluge of artificial intelligence in the form of computerized trading and “robo” advisers over the past few years.  This is not, however, the first time machines have influenced the stock market.

The stock market crash in 1987 was blamed partially on computerized “portfolio insurance” trading where computers were automatically selling as the market was falling.  In 1998, the Federal Reserve bailed out Long-Term Capital Management as it was using quantitative models to invest in securities that deviated from fair value.  Long-Term Capital Management had some major fire power as it had two Nobel Prize winners developing these strategies, including Myron Scholes.  Myron Scholes is the co-creator of the Black-Scholes model for pricing options.  Both of these models were good strategies that would typically work in a “normal” market, but markets are not always “normal”.

Today, we not only have more computerized portfolio management and trading power at our fingertips, we have a growing liquidity problem.  Many investors – including robo advisers – have moved money to ETFs and index funds.  If there is a market correction which causes a run on these funds, the funds will be forced to sell positions quickly.  This creates the liquidity issue – there is no cash buffer in these passively managed funds that you would normally find in an actively managed fund.

What does this mean for your portfolio?

Even though THOR uses models to help us make our investment decisions, the final decision always rests with our investment committee.  As we like to say, our investment approach is 50% science and 50% art.  Why is “art” important?  Even though every market is different, the one constant is human emotion.  At the extremes, markets typically are driven either higher or lower than “fair value” by emotions and speculation.  Computer models do a great job of cranking out numbers; they don’t do a great job of measuring emotion.  Given that they have only been around for a short time, robo advisers and ETFs have not been tested when emotions run high and they are forced to sell.  Such an event could cause cascading selling like portfolio insurance and Long-Term Capital Management did in the past.  It is usually at those times (2002 and 2008) that investors need a human to help them avoid making bad emotional investment decisions.  Computers can’t do that.