High-Frequency Trading Changes the Game
Risk doesn't obey stop signs.
May 6, 2010
High-frequency trading accounts for 70% of the volume on U.S. markets, and an increasing share on Canadian markets. A report from the Financial Markets Group at the Federal Reserve Bank of Chicago suggests that that increase the potential impact of error.
“Certainly, trading losses due to errors are routine in an open outcry trading environment, and numerous trading errors in electronic, screen-based trading environments have resulted in individual losses amounting to hundreds of millions of dollars. Although algorithmic trading errors have occurred, we likely have not yet seen the full breadth, magnitude, and speed with which they can be generated,” writes Carol Clark.
She also provides a helpful overview of a fast-evolving sector of the market.
“A small group of high-frequency algorithmic trading firms have invested heavily in technology to leverage the nexus of high-speed communications, mathematical advances, trading, and high-speed computing. By doing so, they are able to complete trades at lightning speeds. High-frequency algorithmic trading strategies rely on computerized quantitative models that identify which type of financial instruments to buy or sell (e.g., stocks, options, or futures), as well as the quantity, price, timing, and location of the trades. These so-called black boxes are capable of reading market data, transmitting thousands of order messages per second to an exchange, cancelling and replacing orders based on changing market conditions, and capturing price discrepancies with little or no human intervention.”
While there may be little human intervention, there are human-designed risk controls for members of an exchange’s clearinghouse to mitigate losses. These include margin requirements but also limits on the size and price of the order.
However, these controls get in the way of speedy trade execution. As a result, some non-members of a clearinghouse are getting direct access to an exchange’s trade matching engine, either sponsored by a clearing member with some pre-trade risk controls, or “naked” – with none at all.
As in any innovation, there is good news and bad news. “There is evidence that high-frequency algorithmic trading also has some positive benefits for investors by narrowing spreads—the difference between the price at which a buyer is willing to purchase a financial instrument and the price at which a seller is willing to sell it—and by increasing liquidity at each decimal point” Clark notes. “However, a major issue for regulators and policymakers is the extent to which high-frequency trading, unfiltered sponsored access, and co-location amplify risks, including systemic risk, by increasing the speed at which trading errors or fraudulent trades can occur.”
And, there is anecdotal evidence that “accidents happen.” Clark reports one such incident: “For example, in 2003 a U.S. trading firm became insolvent in 16 seconds when an employee who had no involvement with algorithms switched one on. It took the company 47 minutes to realize it had gone bust and to call its clearing bank, which was unaware of the situation.”
She doesn’t arrive at any firm conclusions. The target, perhaps, is moving too quickly. But the risk management systems of the broker-dealers have to be as fast as the technology they provide to high-frequency traders, she suggests: “Therefore, of paramount importance is the speed at which clearing members receive post-trade information from the clearinghouse and incorporate this information into their risk-management systems so that erroneous trades can be detected and stopped.”
Otherwise, clearing members may only be 47 minutes away from bankruptcy.