Remember me

Register  |   Lost password?


Sign up here to let us know if you are interested in joining us for our Introduction to QuantLib Course later in the year.

 

MoneyScience Blog Header 2015 2


Andrew Lo - Computation and the Evolution of Financial Technology

Wed, 06 Jul 2011 06:45:00 GMT

 

 

Even the most sophisticated technology and computational methods may not be enough to prevent the kind of disastrous errors that recently played out across the global economy, suggest these two speakers. Human temperament and judgment will remain critical factors in financial markets.

When there is an economic incentive, says Andrew Lo, “technology has always been put into practice quickly.” Soon after the invention of the telegraph, Wall Street was draped in wires. Lo offers a thumbnail of significant milestones in computer and financial technologies, including macroeconomic modeling, linear programming, portfolio optimization, and the controversial securitization and CDO models implicated in the 2007-08 economic debacle.

Many of these marriages of high tech and finance proved successful and beneficial to millions: stock option valuation formulas enabled the launch of new exchanges, and the evolution of computing has made electronic trading possible and cheap for multitudes. But there have also been “unintended consequences,” says Lo, a “theme that runs across all technology, not just finance.” He shows an index of U.S. residential housing prices from 1890 to 2006, which shows a major uptick at the end – graphic illustration of last decade’s “extraordinary run up of housing prices.” This inflation and the subsequent crash Lo traces to “the confluence of human behavior interacting with new financial technology, allowing us to pump tremendous amounts of money into residential real estate.”

Because of the complexity and size of financial networks and interactions, markets are increasingly interconnected, so breakdowns in one sector inevitably impact others. Lo points to the “quant meltdown” of March 6, 2010, where in 13 minutes, a $35 billion company shrank to $1 million. This kind of flash crash, triggered by human error coupled with all-electronic trading systems, “will happen more frequently, not less,” says Lo. “We have the ability to wipe out a large amount of life savings at the click of a mouse. We have to do better. Technology must account for the frailty of human behavior.”

Despite advances in technology, says John Thain, it is still the case that “garbage in results in garbage out.” Back in 1979, when Thain started at Goldman Sachs, he analyzed financial transactions on large sheets of green paper, and did calculations by hand. A single merger transaction with 12 different structures and 12 different prices required 144 sheets of paper, and mistakes meant literally cutting and pasting. Excel, PowerPoint, and powerful computers have made the work of financial analysts more productive, says Thain, but none of the new technology “will tell you if a transaction is good for shareholders. We still have to make sure the numbers are right.”

When a trade takes less than a millisecond to execute, human errors become rampant. (Thain notes that traders want their servers next to official exchanges to “reduce the speed of light delays caused by the distance an electronic order has to travel.”) Sometimes input filters catch the extra zero or an improbably repetitive trade, but “people are creative in their mistakes,” says Thain. While the New York Stock Exchange features human specialists to catch these, there is a lack of human intervention in the many purely electronic exchanges. He rues the overly complex mortgage tranching schemes that led to “ninja loans” and a hyper leveraged real estate market and disaster for financial institutions worldwide. Thain concludes, “It’s not good to buy things you don’t understand,” and trades with no realistic inputs. The greatest advances in methodology and technology cannot substitute for human judgment.

Via: MITWorld

, , , , , , , , , , , , , , , , , , , , ,