How the Algo Stock Markets Imploded

Photo credit: iStockphoto/Bigmouse108

Share markets around the world have plunged in recent weeks amid the COVID-19 pandemic in movements not seen since the global financial crisis of 2008, or even the crash of 1987.

Contrast the 2020 market plunge with that of 1987, in particular, and there are some remarkable differences.

Where in 1987 traders were on the end of landline telephones shouting sell orders to “chalkies” at stock market blackboards, this year’s crash saw a vast majority of trades automatically carried out by programmed algorithms.

In the U.S. market, for example, a fund manager told CNBC’s “Squawk Box Europe” that as much as 80% of all trades on both the New York Stock Exchange and the NASDAQ markets is algo trading, done through pre-programmed mathematical modeling.

Even three years ago, J.P. Morgan estimated that “fundamental discretionary traders” might account for as little as 10% of stock trading volume.

In a crisis such as today, the high percentage of algo trading raises significant questions to which there are no immediate answers. This is partly because the basis of many of the algorithms come out of a “black box” and their rules lack transparency.

However, it is legitimate to ask how the market volatility we have seen in recent weeks would have differed if it was humans deciding.

Disrupting human thought

In the “old days”, there was such a thing as market sentiment, a human herd mentality which could create panic, or adversely create a bull run on a market or a particular stock. It may not have been rational, but it was on full display in all of its craziness.

Target levels for indices and stocks falling through predetermined price points trigger today’s algo trading. Speed is of the essence and can make a significant financial difference.

This is another reason to cut the human out of the equation.

In a falling market, getting out quickly means losses averted. After the bull run for most of the year leading up to the pandemic, getting out quickly also means profits kept.

There are some out there in the market who believe that the algo trading, en masse, delivers even greater volatility to the markets, both on the way up and on the way down.

Breaching a price point or an index level triggers the algo for an automatic response, creating the digital equivalent of the human herd. In the past, a human trader would have made the call.

The question is whether the algo and the human would make the same choice, or if the human would have had some additional information or just the experience of a long-time working in the markets to make a different call.

Black box vs. explainability

One point to make about algo trading is that it is still a relatively recent phenomenon. Its lack of transparency reflects this.

The AI applications are based on prices and volume prediction, spreads and parameter selection. There is an argument that the greater the volume of trades the AI deals with, the more it learns and the more effective it becomes in “understanding” and executing the trades.

Another view is that well-constructed algorithms deliver better risk management, with preset sell orders saving millions.

Developers have been so excited about this technology and so convinced of its absolute benefits that they have created the black box algo.

As the technology matures and the markets get more used to it, the move is likely to be towards “explainable AI,” where explanations of the decisions behind the algorithms are made available through a user interface.

This can explain the reasons behind decision making to clients, but there are also regulatory, security and governance issues at play.

In the wake of the recent volatility, how many individual traders or trading firms could explain the parameters behind the algorithms which determined their trades?

Realistically, the numbers would be low.

A recession or an algo-based distortion?

So, as it stands, we are in the worst share market slump of the decade which could get even worse, and algorithms are the ones deciding.

We have moved to implement this technology so quickly that we don’t know if this is an excellent idea, or a terrible one.

The hope is that when all this is over and the panic recedes, then we can do some detailed analysis to understand if the algos are distorting the market or saving us billions.

Right now, we don’t know either way.

Photo credit: iStockphoto/Bigmouse108