The Wall Street Journal runs a story today that purports to trace last Thursday’s wild afternoon market gyrations back to an options trade made at 2:15 p.m. ET in the Chicago pits by Nassim Nicholas Taleb’s Universa Investments LP hedge fund.
“On any other day, this $7.5 million trade for 50,000 options contracts might have briefly hurt stock prices, though not caused much of a ripple. But coming on a day when all varieties of financial markets were deeply unsettled, the trade may have played a key role in the stock-market collapse just 20 minutes later.”
In the story, written by Scott Patterson and Tom Lauricella with help from ex-HedgeWorld-er Jacob Bunge, the Journal cites unnamed traders who said the Universa tradeâ€”an out-of-the-money bet that the Standard & Poor’s 500 stock index would fall to 800 in Juneâ€”caused counterparties to sell stock to hedge their exposure. The trade was conducted through Barclays Capital.
“The more the market fell, the more the traders at places like Barclays had to sell to protect their own positions,” Patterson and Lauricella wrote. “This, along with likely dozens of other trades across the market led to a cascade of selling in the futures markets.”
As volume rose, data feeds at brokerages and trading systems at the exchanges couldn’t keep up. Heavy trading begat more trading. At 2:37 p.m., according to the Journal, the Nasdaq said it would no longer route quotes to the NYSE’s Arca electronic trading system because of problems with the data it was receiving from Arca. BATS also stopped routing quotes to Arca.
“For a crucial set of playersâ€”high-frequency-trading hedge fundsâ€”all this turmoil was becoming too risky to handleâ€¦. With the high-frequency funds either selling or pulling out of the market, Wall Street brokerage firms pulling back and the NYSE stock exchange temporarily halting trades on some stocks, offers to buy stocks vanished from underneath the market.”
Reading the story, I understand what the Journal was trying to do: find the origins of a financial event that has everybody spooked. It is the natural way of news outlets to seek identifiable agentsâ€”peopleâ€”to which an otherwise inexplicable series of events can be traced. But does anyone seriously believe a 50,000-contract out-of-the-money options bet on the S&P 500 initiated last Thursday’s market decline? That seems harder to believe to me than the “fat finger trade” error story.
In working the Taleb/Universa angle so hard in this story the Journal, in my view anyway, undermines its broaderâ€”correctâ€”theory that electronic trading and exchange systems failed spectacularly. What happened on May 6 was definitely a computer problem, but attempting to tie it to a single trade by a single fund seems overly simplistic and kind of sensationalistic.
Let’s say, for the sake of argument, that Universa’s trade did spark the trading that crashed the electronic systems that exacerbated the sell-off. Where do we go from there? Do we restrict traders through regulation? Do we somehow apply “speed limits” to electronic trading networks and electronic exchanges during times of heavy volume? Neither really seems to get at what I think is the root cause: too much trading of stuff that by itself doesn’t have any real value.
At least one person in the comments section of the Journal story agrees. “David Ziegelheim” writes, “The problem really is nothing to do with electronic trading. The problem is that the trades do not represents anything to do with the value of the underlying companies. The trades are part of a gaming strategy against other investors as the market value far exceeds the underlying value of the company.
“The stock market has just become a combination of a gaming table in Las Vegas and the Ponzi scheme. The crash was inevitable as the hedge funds and their gaming strategy of trying to detect the collapse of the Ponzi scheme.
“This all comes from a lack of investment opportunities. That in turn is caused by tax and regulatory policies that make new investment unattractive. Investment shows up at the expense on the bottom line whether or not it is depreciated. Buying another company just moves one set of assets from one row to another. With huge amounts of money looking for a place to go and numerous money managers looking for ways to keep their jobs, this is just become a shell game that eventually will collapse. Maybe not last Thursday, but the day will come in the not so distant future where the “rule of sixes” will again be a valid investment strategy.”
All of this makes me think that much of what we see in the way of “market updates” on CNBC and elsewhere really fail to capture the true reasons behind what we’re seeing. Sometimes the reporting seems rooted in the old-fashioned notion that “investors’” views on securities are being represented in the prices. Investors like Home Depot, so the stock price is up. Investors don’t like Proctor and Gamble, so the stock price is down. Investors think the euro is overvalued or undervalued versus the dollar or the yen.
But maybe that’s not entirely, or even mostly, the case. It seems possible now that many of the intra-day price moves we see are not caused by human investor sentiment at all, but rather are a reflection of human-programmed but otherwise fully automated algorithms engaging in a complex dance across stocks, bonds, currencies and commodities.
I liken it to a computer conducting an orchestra versus a human conducting an orchestra. The computer can be programmed to actuate mechanical arms and hands that mirror the movements a human composer would make, and it’s even possible that a computer might be programmed to “listen” to the music and alter its composing style based on the audio feedback from the sound of the instruments. But the computer isn’t really feeling anything. And if it goes wiggy, things can go downhill fast.
It’s really the same any time you introduce computer technology into any sequence of events. Take the electronic engine control module on my motorcycle versus the carburetor on my wife’s bike. Most days I like the ECM. It gathers data from an oxygen sensor and automatically adjusts the air-fuel ratio to provide the optimal mixture for the given inputs of the moment based on preprogrammed criteria. In theory, if I tinker with the air flow in any wayâ€”either intake or outflowâ€”the computer will adjust the fuel delivery based on the new information from the O2 sensor. On my wife’s bike, if I change the air flow, I have to take out my tools and tinker with the carburetor by hand to adjust the fuel delivery.
On the down side, though, if any part of the electronic engine control system fails, I’m dead in the water. A computer glitch can derail my ride. My wife’s carbureted bike has no computer that can fail.
This is my analogy-laden take, anyway. It’s also worth noting, as Journal story commenter “Adam Hendricks” did, that there’s not a lot of variety with computers. Many of these algorithmic systems aren’t all that dissimilar from one another, and many have similar goals. So when they all start moving fast, they’re also moving together. That’s bad from a correlation standpoint. Here’s Hendricks’ take: “Everyone is running virtually the same software, and the software did what it was supposed to do: follow trends, run predictive model algorithms, and send high speed buy and sell orders based on the model results on market behavior that is presumed to be caused by human behavior patterns. Such software was originally designed for small boutique investment groups to glean profits by mathematically modeling the behavior of the market as a whole. Such software doesn’t work well at all when every major trading house is using it. The model driven high speed trades feed off one another. The market is not being driving by human trading behavior anymore. It is being driving by similar software reacting in high speed to feedback loops on each other. That’s why it’s been such a roller coaster ride and will continue to be so.”
Hard for me to improve on that, but you all are welcome to do so in your own comments here.