Fed fears increase in runaway algos

Risk controls to prevent erroneous trades are not being properly implemented, with runaway algos more common than first thought, according to new research from the US Federal Reserve.

Risk controls to prevent erroneous trades are not being properly implemented, with runaway algos more common than first thought, according to new research from the US Federal Reserve.

The study by the Chicago Fed complied interviews from a range of market participants and found that two of the four clearing brokers, two-thirds of prop trading firms and every exchange questioned had experienced errant algorithms. One exchange added it could detect runaway algos which had a significant price impact, but it was unable to determine if excessive positions had been built up over time.

The prevalence of rogue algorithms was thought to be the result of improper installation or oversight of pre-trade risk management controls that may arise from competition among brokers to offer the quickest market access, as well as delays in receiving post-trade data that prevent real-time monitoring of client flow.

“Most of the trading firms interviewed that build their own trading systems apply fewer pre-trade checks to some strategies,” suggested Carol Clark, senior policy specialist and the Chicago Fed and author of ‘How to keep markets safe in the era of high-speed trading’. “Trading firms explained that they do this in order to reduce latency.”

The ownership structure of exchanges was also called into question, given many latency-sensitive trading firms have stakes in trading venues that could be used to influence the structure of exchange systems.

The study also showed there are times when no single entity has a view of a trading participant's overall exposure to the market. This is an issue under consideration by some risk management technology providers, with Nasdaq OMX-owned FTEN recently installing a solution at Bank of America Merrill Lynch which gives the bank visibility of all its European client’s exposures across markets.

To prevent future losses related to high-speed trading, the paper suggests the imposition of limits on the number of orders that can be sent to a trading venue in a period of time, a kill-switch that can halt trading at one or more levels, and profit and loss limits that restrict losses.

«