What has been happening in the financial markets? And what exactly is algorithmic trading?
Stock markets around the world have been having a turbulent time recently, the cause of which has been attributed by some to the use of automatic ‘algorithmic trading’.
On Monday 5th at around 3 p.m. ET the Dow Jones did something weird. No big headlines announced what was going on and, although it had been steadily sinking throughout the day on the back of quite genuine inflationary fears, all of a sudden, it took an almighty nosedive and started to hemorrhage red ink.
Amidst the confusion, Walter “Bucky” Hellwig, the senior vice president of BB&T Wealth Management in Alabama offered an explanation to Bloomberg:
“The drop in the morning was caused by humans, but the free-fall in the afternoon was caused by the machines.”
Essentially, “algo-trading” is when you program software to buy or sell stocks automatically. This removes pesky human traits such as emotion or conscientiousness, and instead can make rapid-fire trades based on a number of mathematical factors.
The keyword here is programming and we should never lose sight of the fact that all programs have bugs. It is merely a question of how many.
Price, quantity and timing are all taken into consideration when buying or selling stock, and an algo-trader can make these decisions faster than our monkey brains ever could. They also allows for institutional investors to buy or sell huge quantities of stock in many smaller blocks, so as not to affect the stock price in the process.
However, seeing as algo-trading tends to magnify upward or downward trends, occasionally you’ll get what happened on Monday. Thankfully the Dow largely recovered after the machines were done selling everything off, even if, at the time of writing, it is floundering some 2,000 points off its late-January highs.
So we know algo-trading can be volatile, but where has this trend come from?
A brief history of machine traders
Stock exchanges began moving from traditional auction to computerized transactions back in the 1970s, and in the late 80s and early 90s, Electronic Communication Networks (ECNs) were opened for traders looking for more efficient access to the markets.
Algorithms offered a clever solution as they can be programmed to follow a simple set of rules to execute in sequence until a desired end point is met.
And this was the beginning of humans slowly being taken out of the equation.
In 2001, algorithmic trading gained even more traction when IBM researchers published Agent-Human Interactions in the Continuous Double Auction, a paper that found that simple automated strategies outperformed humans by a clear margin, setting the stage for the high-frequency trading that is widely used today.
Faster networks and low latency proximity hosting on the global exchange allowed for faster and faster trades, and algorithmic traders can now act on information instantaneously. But what happens when it all goes horribly wrong?
Over-keen robo-brokers just can’t help losing billions
Algo-trading has often been a problem; this is due to the fact that trying to automate an understanding of something as chaotic as the stock exchange creates a lot of instability. In short, some of them are actually not very good at their job.
On May 6th 2010, there was a ‘flash crash’ that sent the Dow Jones industrial average plummeting 700 points in just minutes. The cause, you guessed it, was automated selling algorithms that had “no regard to price or timing”.
And again, on Aug 1st 2012, market-making firm Knight Capital Group lost $440 million in 30 minutes due to a “large bug” in its trading software.
Whilst these techniques are being constantly refined and will soon be morphed in to full-blown AI architectures, the last two weeks’ news is a salutary lesson in not over-relying on the algorithms that alarmingly underpin a lot of our world.
At this point, it is probably worth bearing in mind the old maxim; BS in, BS out.
Here’s an interesting take on algo-trading from the BBC’s flagship Newsnight TV program.