Considering the Human Element in Online Trading
Earlier this year I participated in the Computational Science Ventures lecture series hosted by Harvard’s Institute for Applied Computational Science. It was an opportunity to engage with students and faculty on a topic we follow closely at Ascent – the merging of the physical and digital worlds and the resulting opportunities for innovative new products and services.
One of the subject areas I spoke about was the increasing role of computer-driven decisions in stock trading. Today algorithmic trading comprises an estimated two-thirds of the trades made in the U.S. markets. In many cases these systems make trading decisions in a fraction of a second, something humans cannot do – our brains simply don’t work that way.
During the talk, one point I highlighted was that the human element of trading should not be marginalized. Only two years ago a “flash crash” likely caused by algorithmic trading resulted in almost a trillion dollars momentarily vanishing. We haven’t had an event of that magnitude since, but the lesson was clear: if not properly monitored, managed and regulated, computer-driven trading operations have the potential to cause financial chaos.
Last week the Wall Street Journal covered a new firm that aims to put more of the human element back into algorithmic trading. System Two Advisors L.P. – founded by Wall Street veteran Robert C. Jones – operates under a model whereby computer-driven trading incorporates research from a large team of analysts. The goal is to capture the best qualities of each – the nuanced perception of the human analyst and the unemotional, data driven decision making of the computer.
Approaches like the one pursued by System Two are cropping up in areas well beyond Wall Street. One of the most mature of these areas is the market for online ads. Our portfolio company Bizo, for example, employs advanced algorithms to crunch massive amounts of audience demographic data and deliver thousands of ad buying decisions per second.
Across the board, businesses are increasingly relying on data – which continues to grow exponentially – to make decisions. Currently these decisions are primarily driven by human analysts, but over time we will surely see a migration towards computer-driven decision making across our economy. As we’ve seen from Wall Street, there is risk in becoming too heavily dependent upon computer-driven decisions, and it will be interesting to see how we ultimately balance the benefits of algorithmic approaches with the value of human insight and oversight.