Many institutional investors (including fund managers) utilize some form of automated trading. High-frequency trading (“HFT”) is a distinct subset of automated trading that makes its money from rapid entry and exit in positions throughout a trading day and can be based on a number of different strategies. Positions are generally held for fractions of a second (faster than an eyeblink in some cases).
Normally under the radar, HFT took center stage in the investing world in 2010 with two events: on February 3, HFT firm Infinium Capital Management’s algorithm malfunctioned, entering 6,767 orders to buy light sweet crude oil futures on the NYMEX, enough of which were filled to send the market soaring. More well-known is the May 6, “flash crash,” which saw US equities plummet, helped along, regulators said, by HFTs. However, of the two, the February 3 incident more clearly highlights some of the compliance issues relating to HFT.
A NYMEX business conduct panel investigated the February 3 incident and found a number of risk management failures, ultimately issuing a $350,000 fine to Infinium. Interestingly, the panel determined that Infinium breached its own protocols in developing the algorithm that went rogue on the NYMEX:
• Infinium’s normal testing process takes approximately 6-8 weeks; whereas this algorithm was finished the day before it went on the market and was only tested for a couple of hours;
• Features designed to automatically shut down the algorithm failed, attributed to errors in the code; and
• An employee used a colleague’s trading ID to place positions that would offset the firm’s undesirable exposures.
While the flash crash and the Infinium incidents were unusual in their scope (and resulted in a fine in the latter case), smaller versions of these kinds of issues take place every day.
While there are benefits to the marketplace associated with HFT (such as greater liquidity and lower transaction costs), firms utilizing it should build a robust compliance program around these activities, considering the following:
• Enhanced risk controls because of competitive time pressure to execute trades without the more extensive safety checks normally used in slower trades;
• Establish and document the process for developing, testing and using algorithms. Out of control or “rogue” algorithms are fairly common; though immediate causes may vary, a root issue is the care that is taken (or lack of it) in building the algorithms in the first place, or managing them once in use. Examples include:
o Algorithm goes out of control and submits unexpected orders;
o Trader sets parameters that cause an algorithm to trade too aggressively;
o Algorithms used simultaneously (e.g., to outbid one another) get into a negative feedback loop.
• Write “kill switches” into algorithms so trading can be stopped at certain preset levels;
• Impose limits such as:
o The number of orders that can be sent to an exchange within a set period of time;
o Intraday position limits that set the maximum position a firm can take during one day;
o Profit-and-loss limits that restrict the dollar value that can be lost.
• Security risks such as terrorist activity, hackers, disgruntled employees, or others obtaining and interfering with algorithms and/or related systems. Firewalls and physical barriers should be used and other measures taken to limit both internal and external access. Security measures should be tested periodically to ensure that they have not been tampered with, or accessed inappropriately;
• Determine whether the firm’s HFT activities could be considered front running, and whether this activity violates applicable regulations or the firm’s own Code of Ethics.
The SEC responded to the May 6 flash crash and it is likely that the regulatory landscape will continue to evolve. A round up of regulatory activity is listed below:
• In place:
o The SEC introduced “circuit-breakers” for individual stocks that stop trading across all markets. While these can be modified to permit continued trading, this is only within set parameters;
o The SEC established uniform policies for canceling trades struck at clearly irrational prices;
o The SEC eliminated “stub quotes,” which allowed market-makers to buy good stocks for a penny if there are no other bids;
o The London Stock Exchange abolished liquidity rebates.
o Differences in data conventions among the dozens of markets may have exacerbated the flash crash;
o Banning flash orders;
o Require issuers to notify exchanges when they expect material information will be revealed during trading hours so that the exchanges can halt trading before the news arrives;
o Regulators/governments should release major information only when the markets are closed or at pre-announced times;
o The technology required to run an HFT platform is very expensive. As trading speeds increase, HFTs are investing more money into faster technologies. A concern here is that ultimately, only a few HFTs who can afford the ongoing expenditures will remain – undoing the benefits that HFT can have for the marketplace. It has been suggested that this “arms race” can be halted by requiring all exchanges to delay the processing of every order instruction they receive by a few milliseconds.