It was a bluebird day in Midtown Manhattan on May 6th, 2010. At 2.40pm in the afternoon, I can imagine that most Wall Street traders were almost ready to start packing up and heading home for the day, or at least had grabbed another coffee to get them through the afternoon slump. Then something happened that woke them the hell up.
At 2.42pm, the Dow Jones started dropping. It dropped 600 points in five long, terrifying, and confusing minutes. For five minutes, everyone panicked. By 2.47pm, the Dow had dove rapidly to an almost 1,000-point loss on the day – the second largest point swing in Dow Jones history – until someone literally pulled the plug on the market and trading stopped.
When trading opened again a few minutes later at 3.07pm, the market had regained most of that 600-point drop.
This was the 2010 Flash Crash. In order to understand the Flash Crash, the first thing I needed to understand was just how outdated my idea of the stock market actually was; I pictured Wall Street, v.1 – lots of white guys in suits shouting BUY and SELL and cursing on the phone to other brokers all around the world. These days, and for the last twenty years or so, over 70% of all stock market trades are run by super computers who trade tens of thousands of stocks in milliseconds – we’ve gotten rid of sluggish human beings completely. I needed to picture a gigantic room full of computers making a high-pitched whine instead.
During that five-minute period, the stock market – and in turn, the economy – lost billions of real $$ money. No one knew what had actually happened. The SEC tasked an unlucky committee to immediately figure it out. That report, which took five months to research and compile, came to the conclusion that it was one bad computer algorithm that sent the market into a spiral. More importantly, however, that report documented:
The joint report “portrayed a market so fragmented and fragile that a single large trade could send stocks into a sudden spiral,” and detailed how a large mutual fund firm selling an unusually large number of E-Mini S&P 500 contracts first exhausted available buyers, and then how high-frequency traders started aggressively selling, accelerating the effect of the mutual fund’s selling and contributing to the sharp price declines that day.
Critics of the SEC’s report were many, and included much deserved criticism around how, despite the fact that the SEC employed the highest-tech IT museum in their research, which included five PCs, a Bloomberg, a printer, a fax, and three TVs – it still took nearly five months to analyze the Flash Crash. Specifically:
A better measure of the inadequacy of the current mélange of IT antiquities is that the SEC/CFTC report on the May 6 crash was released on September 30, 2010. Taking nearly five months to analyze the wildest ever five minutes of market data is unacceptable. CFTC Chair Gensler specifically blamed the delay on the “enormous” effort to collect and analyze data. What an enormous mess it is.
So: What does it mean when our machines make a split-second mistake that costs us real billions, but takes humans months to understand what actually happened?
PS: Radiolab does a spectacular piece on the Flash Crash called Million Dollar Microsecond – I would highly recommend a listen.