Monday, January 3, 2011

The Need for Speed: Brave New World of Trading

Graham Bowley wrote a chilling article for the Sunday New York Times Business section which should give pause to anyone who still believes that stock exchange trading is accomplished by flesh-and-blood traders shouting buy and sell orders in the pits. Uh uh. The new world of trading stocks, bonds, commodities, and derivatives is done by high-frequency algorithmic trading programs talking to each other in huge barns in New Jersey.

Trades don’t take place anymore in the blink of an eye. They take place in a particle accelerator:

Nasdaq, for example, says its time for an average order “round trip” is 98 microseconds—a mind-numbing speed equal to 98 millionths of a second.


So what’s wrong with machines untouched by human hands matching buy and sell orders? Isn’t such innovation faster, cheaper and more democratic? Now anyone can play with the click of a mouse or the tap on an app.

Nothing is wrong if you’re one of the high frequency traders (HFT) that these super-fast electronic exchanges cater to.

Despite its claims of leveling the playing field, this brave new world is funded by a concentrated block of financial institutions. Bowley discusses Direct Edge, an electronic trading platform that is backed by Goldman Sachs, Knight Capital, Citadel Securities and the International Securities Exchange. Its largest shareholder is JPMorgan.

The HFT programs aren’t making qualitative, slooooow decisions based on concerns such as management competency or the future of the new products pipeline. These machines make money zipping in and out of trades taking advantage of very slight discrepancies in prices across all the exchanges. You don’t make a lot from each trade, but ½ cent multiplied by millions in nanoseconds adds up:

By using such techniques, traders make only the tiniest fraction of a cent on each trade. But multipled many times a second over an entire day, those fractions add up to real money. According to Kevin McParland of the TABB Group, high-frequency traders now account for 56% of total stock market trading.


HFT gives almost total advantage to the entity with the fastest machines, which means the entity that can afford to invest in them. Ordinary investors need not apply. Except they are, because they own the assets being traded.

What can go wrong with machines running things without human intervention? Back on May 6, 2010, the Dow Jones Industrial Average plunged 700 points in minutes before prices recovered. This was dubbed the “flash crash.” I remember this well, especially because Accenture, a large global consulting firm, was trading at about $41 a share at 2:40pm. Within seconds, its price was driven down by the “flash crash” to one penny per share.

The Securities and Exchange Commision (SEC) and the Commodity Futures Trading Commission found that the drop was caused by one sale, dropping a $4.1 billion block of E-Mini Standard & Poor’s 500 futures contract into the marketplace. When the sell order came in, it seeped through all the exchanges and “buy” orders plugged into the computers shut down. All the algorithms stopped buying at once, and the market went into free fall. In the past “human” traders might have stuck around to intervene to prevent such distortions simply because they could discern them, wait patiently, scrutinize asymmetrical discrepancies and adjust for them.

The May 6 “flash crash” could be the first of many to come. The system certainly can be gamed:

One debate has focused on whether some traders are firing off fake orders thousands of times a second to slow down exchanges and mislead others. Michael Durbin, who helped build high-frequency trading systems for companies like Citadel and is the author of the book, “All About High-Frequency Trading,” says that most of the industry is legitimate and benefits investors. But, he says, the rules need to be strengthened to curb some disturbing practices.

“Markets are there for capital formation and long-term investment, not for gaming,” he says.


HFT is so fast the supercomputers use a trading technique known as flash orders. This allows them to see the buy/sell orders of other traders before they move onto the wider marketplace where everyone supposedly has access to the prices.

You can’t put the genie back into the bottle. But we can (hopefully) learn from the past.

No comments: