The marginalizing of the individual investor: the inside story of flash crashes, systemic risk, and the demise of value investing.

• Author: Malmgren, Harald

On May 6, 2010, at about 2:30 p.m., a "flash crash" in the U.S. stock market sent the Dow plummeting by almost 999 points--the deepest intraday plunge ever recorded. From 2:45 p.m. to 3:00 p.m. the Dow recovered by about 600 points. During the flash crash, some individual stocks collapsed to pennies and then rebounded. For the day as a whole, the market closed down about 350 points.

This unprecedented shock galvanized the Securities and Exchange Commission into action to identify the cause and devise mechanisms for preventing recurrence. At first, it was thought that a "fat finger" error might have sent computerized trading into a tailspin. However, it soon became evident that the scale and speed of trading abnormalities could not have been the result of a single misplaced finger. To its chagrin, the SEC found its antiquated computer capabilities could not begin to capture the volume and speed of transactions in today's markets. Moreover, there was no coherent data center which encompassed the disaggregated reporting systems of nine separate U.S. exchanges, the dark pools, and the activities of new ultra-fast, computerized highfrequency trading platforms. Theoretically, it could take months to run the day's records in ultra-slow motion to identify market malfunctions or possible manipulations. The SEC does not have the manpower or technical capability to carry out such a challenging task.

Faced with unprecedented challenges, the SEC resorted to its traditional and only immediate regulatory remedy of setting new circuit breakers for individual stocks. Then it proposed adding a requirement that broker-dealers restrict quotes to within a specified range around prior trades.

Circuit breakers may help reduce the impact of potential breakdowns in specific stocks, but they cannot prevent recurrence of massive disruptions in exchange-traded fund trading. According to Barron's, ETFs were responsible for about 70 percent of all the May 6 flash crash trades that were later cancelled by the exchange. ETFs are a small percentage of securities traded on the stock exchanges, but they were a major factor as liquidity faded away and the ETFs themselves became decoupled from their underlying baskets of stocks, while the prices of some specific stocks swung wildly.

[ILLUSTRATION OMITTED]

Most important, circuit breakers are of no significance if a large-scale implosion of liquidity results from suspension of trading by high-frequency traders, threatening systemic collapse of the entire equity market. What happened that day is that some high-frequency traders did disengage from markets, turning off computers which seemed to their operators to be conveying misinformation.

Recognizing the inadequacy of available information, the SEC has now started developing a new reporting system known as the Consolidated Audit Trail. This system is intended to bring all of the presently disaggregated, fragmentary data into one place to enable "real-time surveillance." Development of the CAT is expected to require at least $4 billion to build, take three years to complete, and require $2 billion per year to operate. Long before CAT is a reality, there are likely to be more crashes. Moreover, there can be doubt whether real-time capacity could ever be achieved given the complexity of high-frequency trading "flash trades" (execute or cancel on receipt) in which a high ratio of bids automatically self-destruct. By the time CAT is finished, technology will have spread globally, posing unforeseeable jurisdictional information disclosure issues as well as new technological challenges.

Ten years ago, it had already been recognized that trading based on computer algorithms probably accounted for about half the total trading volume in the U.S. markets, and dominated the trading practices of major institutional investors. Since then, technology has accelerated, with high-frequency trading platforms ramping up the speed to milliseconds and more recently to microseconds. One of the high-frequency trading CEOs stated that the longest duration of holding a given stock by his firm was eleven seconds, although he considered that to be unusually and unacceptably long.

The combined role of computerized algorithms and high-frequency trading activities varies in intensity during the course of any trading day, but during some time intervals may account for more than 80 percent of total trading volume...