Peering into the Black Box

Was Lehman Murdered by a Thousand Trades?

Faulted by its detractors as a nefarious presence and underestimated by its defenders as benign, high-frequency trading is truly the darkest part of Wall Street's black box, a trading movement that is so new and so dominant that it has triggered both of the market's extreme passions -- fear and greed -- at the same time.

This controversial mechanized speed trading is responsible for about half the volume in the stock market and probably an equal share of the happy hour conversations within a few blocks of the New York Stock Exchange. Are computers better than humans? Can they be trusted? Are they taking advantage of the traditional investor? How can we invest in them?

And, with the calendar nearing the one-year anniversary of the collapse of Lehman Brothers, a question lingers: Did the black boxes bring about its fall? To many, it's still unclear if Lehman's mortal wounds were self-inflicted or if its death was hastened by a thousand trades.

Enter Rishi Narang, founder and head trader of Telesis Capital LLC, a fund company that evaluates and invests in quantitative trading strategies, including high-frequency trading. Mr. Narang, angry at what he believed was the media's misunderstanding of the quantitative world, decided in the fall of 2007 to write a book explaining the practice. In doing so, he tried to make one point clear: computers don't manipulate markets, people do.

To be clear: Mr. Narang is of the opinion that as a whole, quantitative trading and its sub genre, high-frequency trading, are good for the markets. Such trading strategies create liquidity in the markets and -- in the case of high-frequency guys, he argues, can stabilize volatility by providing the other side of the trade. They buy when there is a preponderance of sellers. They sell when buyers are dominant.

Mr. Narang, whose book Inside the Black Box: The Simple Truth About Quantitative Trading expands on the point, says high-frequency traders took the wrong side of trades as bank stocks were tumbling around the time Lehman Brothers was failing. That's the opposite of what many believed was quantitative profiteering on Lehman and other big banks a year ago.

"Quants were hurt in the run-ups and run-downs," Mr. Narang said. "They were very hurt because of historical fundamentals. A lot of these guys were going long banks as they were collapsing."

A "Human Decision"
By the time Lehman's future was in a crisis, many quantitative traders essentially pulled the plug on their models and the computers that ran them, he said. Mr. Narang took his portfolio to cash in the days before Treasury Secretary Henry Paulson and Lehman management met to discuss a bailout that didn't happen.

"What happens is you're trading off of market information that you can actually utilize and quantify," Mr. Narang said. "These bailouts were not quantifiable pieces of information. There was nothing predictable about them.

"Most quants reduced risk and that was a manual, human decision and a very wise one."

Mr. Narang concedes that in the bigger space of hedge funds, there could have been rumors floated to support short positions. He recalled an email message he received in March 2008 about bets against Bear Stearns Cos. shortly before the investment bank's orchestrated rescue. He said such trades were so prescient they almost had to be the result of insider information or some kind of manipulation.

"It clearly wasn't a quant," Mr. Narang said.

The point is the big trading decisions in all sorts of investing are made by humans, not computers. Suggesting high-frequency trading and its potential dangers are a problem caused by machines is to make a fundamental mistake. It's not the machines behind rapid-fire trading that threaten to make the markets an uneven playing field, it's the people.

The problem, of course, is that people have been given new, trading tools of mass destruction. In the right hands, a high-frequency or passive quantitative program can be left running with minimal human interference. There's balance and liquidity in the markets, and the traders make a tidy profit.

"Horrendous Things"
Put those powerful tools in the hands of the less scrupulous, however, and suddenly the rules have changed. Someone is not only playing the game but controlling the roll of the dice. In August, a federal prosecutor asked a judge to put Sergey Aleynikov in jail because the computer code he allegedly stole from Goldman Sachs Group Inc. could be used to "unfairly manipulate" stock prices.

If someone could do it with a few megabytes of code, what's to stop traders at a big brokerage, or a big hedge fund or anyone with real resources from juicing the trading desk even if for a few trades? Stock manipulation is a crime as old as the markets, but have the markets ever seen such a powerful force as a computer that can crunch numbers -- including incoming customer orders -- in milliseconds?

Regulators, including the Securities and Exchange Commission, admit they have scant resources to police all those trades happening faster than the speed of light. So-called flash trades, a kind of high frequency trading that some have likened to front-running, may be banned. But day-to-day workings of the high-frequency market go largely unwatched.

Mr. Narang believes that almost all high-frequency trading is either neutral or beneficial to the markets. He even thinks the market would be healthy if as much as 95% of all trading became automated. But he can't guarantee that someone will misuse the technology. Lehman may have been spared the black box a year ago, but what about the small investor who's beaten to the punch on a stock by a trader who has programmed a microchip to seek and destroy?

"If you leave people to their own devices they often do horrendous things," Mr. Narang said.

That's why looking closely at the black boxes will only get us so far. The real answers lie with the hands holding them.

Peering into the Black Box -
Wonderful article. And, I myself think there needs to be some limit on technology. For instance, it should be flat out illegal to create a human-like AI, which would then cause a technological singularity (human level AI learns how to program other human-level AIs, which then essentially supersede human beings and take away any reason for human beings to exist...though frankly, I don't think this will happen any time soon).

That said, technology has never been evil. In fact, I'd argue that technology has for the most part done far more good than harm. After all, we take for granted technological breakthroughs such as indoor plumbing, electricity, flight, etc...

At some point in history, these things were unheard of.

Even the atomic bomb has also resulted in the nuclear reactor (although perhaps it would have been better if we never discovered the mushroom cloud).