Last Updated: 19 May 2021
Well, last week’s joint SEC/CFTC flash crash report (which followed the preliminary findings released in May) makes it clear that Heather had it spot on.
HAL, of course, was an early testament to the fact that humans and computers don’t always mix. For those of you who haven’t seen Stanley Kubrick’s 1968 film, “2001, A Space Odyssey”, HAL, a “Heuristically programmed ALgorithmic computer” discovers (by lip-reading) that two astronauts on his spaceship want to shut him down after an apparent malfunction. In revenge, he tries to kill them both: he succeeds in eliminating one, while narrowly failing to confine the other to permanent, spaceship-free orbit. Even if he fails in the end to gain control, HAL has since had some kind of moral victory by making it into the top 20 “greatest film villains of all time.”
And in fairness to HAL, he seems to have operated with much greater intelligence than some of the computer algorithms that control stock market execution.
The SEC/CFTC report states that the event that precipitated the dramatic May 6 sell-off was a single, very large order in the S&P 500 e-minis futures contract.
What is clear from the report, though, is that it wasn’t so much the size of the order as the way it was executed that did the real damage.
Using a so-called “percentage of volume” algorithm, the trader entered an order to “target an execution rate set to 9% of the trading volume calculated over the previous minute, but without regard to price or time,” according to the report.
The initial selling pressure caused by the order was absorbed by high-frequency traders (HFTs), fundamental buyers and cross-market arbitrageurs, the authors point out. However, as you’d expect, these market intermediaries then sought to offset their own exposures, greatly pushing up overall volumes in the e-minis. In particular, HFTs executed trades in 140,000 e-mini contracts in just three minutes, between 2.41pm and 2.44pm.
“This is consistent with the HFTs’ typical practice of trading a very large number of contracts, but not accumulating an aggregate inventory beyond three to four thousand contracts in either direction,” say the SEC and CFTC.
Unfortunately, it was just this pick-up in volume that caused things to spiral rapidly out of control.
“The sell algorithm used by the large trader responded to the increased volume by increasing the rate at which it was feeding the orders into the market, even though orders that it already sent to the market were arguably not yet fully absorbed by fundamental buyers or cross-market arbitrageurs,” the report points out.
In other words, the greater the market turnover, the faster the algo tried to sell, since it was programmed to try to sell 9% of the trading volume as calculated over the previous minute.
The rest is history: uncoordinated breaks between markets, a disappearance of liquidity, stocks and ETFs falling to a fraction of a cent, busted trades, a market recovery and, so far, an effect on investor confidence that is still difficult to assess.
But the real story of the flash crash is not malevolent high-frequency traders, a Wall Street conspiracy or even poor regulation (though some of the aftereffects of 2005’s regulation NMS in fragmenting market liquidity were not foreseen). Instead, it was simply a case of a badly designed, poorly implemented computer programme.