Death race: High-Freqency Trading is speeding out of control

Last Updated May 10, 2010 5:00 PM EDT

If the recent bank industry bailout highlights the dangers of "too big to fail" financial firms, concerns over high-frequency trading, or HFT, reveal another truth -- speed kills.

The perils are related. Large financial institutions have spent the last three decades seeking to grow ever larger. And when they collapsed, they left a commensurately large crater in the global economy.

Trading firms, including big banks, are now locked in a similar battle to grow ever faster. Both trends have their roots in eras of intense deregulation. In a hyper-connected age, both also illustrate how markets can unravel at warp speed, whether for subprime loans or for equities. But if the dangers around huge financial firms are, by now, glaringly evident, the risks of HFT are only now coming fully into view.

Here's what we're beginning to glimpse: Just as our biggest financial institutions have grown too big to control, the pace of the markets has grown too quick to manage, let alone regulate.

To figure out how we got here, let's take a brief historical detour. In 1998, the SEC passed the so-called Regulation Alternative Trading Systems rule, expanding it seven years later with the "Regulation National Market System." Together, the rules were aimed at making markets more efficient and competitive. And it worked. The changes caused a proliferation in electronic trading networks, cutting trading costs by upwards of 500 percent.

But it also gave birth to new, and worryingly opaque, markets, such as "dark pools," where participants can trade anonymously. In a related development, narrowing margins set off a "technological arms race" among providers of high-frequency, or algorithmic, trading services. Today the fight is over how to shave milliseconds off trades.

The standard defense for HFT is that it makes markets more liquid, cheaper and swifter. Although research is limited, there's some evidence that this is correct. Academics at the University of California, Berkeley, Columbia University, and the Netherlands' Vrije Universiteit Amsterdam have found that algorithmic trading boosts liquidity for heavily traded stocks, reduces the "spread" between buy and sell orders, and increases price transparency.

Just not always. During last week's meltdown, some electronic trading exchanges continued to trade, others slowed and a few seized up altogether, wreaking havoc with prices. Trades balled up in some markets and surged in others, as some algos programmed to sniff out opportunities during such spasms went on a buying spree.

HFTs "don't supply liquidity when they need to. They provide liquidity when they want to," said Joe Saluzzi, co-head of equity trading at brokerage firm Themis Trading.

In other words, algorithmic traders tend to add liquidity and tighten spreads when the market is in equilibrium. When all hell breaks loose, as it did on May 6, they may withdraw liquidity at will. That can raise prices, stir volatility and sow panic.

This should ring a bell. When loans started to curl at the edges in 2007, the credit markets froze solid because no one believed in them anymore. It was a useful reminder that liquidity isn't an inherent feature of markets, but rather an article of faith. Once that wavers, in short, God is dead.

Meanwhile, let me make an obvious, but still noteworthy, point regarding the algos behind HFT: They're written by people, not machines. And in this case, the programs are developed by the very traders that have a financial interest in using them (Yes, I'm ringing another bell.)

You'll recall that during the housing bust, lenders and credit rating agencies concocted all sorts of fancy formulas to justify issuing loans, derivatives and other financial goodies that later proved totally unjustified. The math is only as good as the assumptions behind them.

Similarly, HFT algos can reflect economic reality, but they can also create it.

Also note that our researchers up above concluded that HFT raises liquidity, but chiefly for large-cap companies. That's because such traders focus on widely held, liquid shares, as do institutional investors. The effect on smaller players is less certain.

Perhaps the biggest argument against HFT is that it favors speculators over longer term investors. The gamblers, of course, play an important role in keeping the markets in balance. But not if the shorts, blasting out thousands of trades per second, dominate trading while moving at breakneck speeds in the name of market "efficiency."

Indeed, there are algos out there that will scan and interpret economic news before you, me or anyone else saddled with gray matter has a chance to read it, let alone decode what any of it means. That tilted playing field presents all sorts of thorny problems, like whether retail investors should even be in the game.

All of this said, we should recognize that the HFT genie is never going back in the bottle. The question is how we control it, recognizing that speed, like size, has a cost. Writes Larry Tabb, head of market advisory firm Tabb Research, in a recent client note on last week's crash (registration required):

Maybe we don't want to have the most efficient markets in the world. Maybe we don't want penny and sub-penny markets where liquidity is increasingly fractured.... And maybe we don't want the fastest market in the west.

Image from Flickr

  • Alain Sherter On Twitter»

    Alain Sherter covers business and economic affairs for