Are Algorithms in Charge?
On 19 October, thirty years ago, the stock markets experienced a crash that is still known as the worst trading day ever. The EOE-index, the predecessor of the AEX-index, went down with 12%. That was nothing compared to Wall Street. The Dow Jones lost that day 22,6%.
Stock markets at the moment can’t be compared with the stock markets in 1987. Thirty years ago, there were brokers on the trading floor who wrote down the orders of buyers and sellers. Nowadays, all the trading goes through the Euronext-data center near London. With the knowledge of today, the crash in 1987 looks a lot like a “ flash crash in slow motion”: a very rapid, deep, and volatile fall and recovery in security prices occurring within an extremely short period. Because the trading time was slower in 1987, the ‘flash crash’ was over a period of two days. On may 6th, in 2010, a $4.1 billion trade on the New York Stock Exchange resulted in a loss to the Dow Jones of over 1000 points and then rise to approximately previous value, all over about fifteen minutes.
‘A snowball effect’ set off both flash crashes, in 1987 and 2010. The big differences between the flash crashes are the traders and automated trading systems. When there is bad news on the Internet, algorithms directly sell stocks that are related to the bad news. Eventually, if the bad news turns out to be fake news, algorithms will buy stocks again. With these algorithms it is possible to have a ‘flash crash’ within only a few minutes instead of two days. The flash crash of Ethereum, last year, was only in a period of five minutes. Do you think ‘flash crashes’ will be faster in the future? Could it maybe happen within one minute or seconds?
Erik Van Rijn, Joost Bobber (2017) https://fd.nl/beurs/1223235/nu-steken-handelaren-elkaar-niet-aan-maar-algoritmes
