Managing Risk in High-Frequency Trading
In a post-Flash Crash world, high-frequency trading has a new place in the spotlight. With the proliferation of automated trading systems, order execution can happen faster and in far greater frequency than ever before. There is great opportunity here but, at the same time, the risk of quickly wiping out positions and accounts is significantly increased.
Faulty algorithms, defective implementation, and unexpected market conditions can all cause automated trading system malfunctions. Since software defects are inevitable and unexpected market conditions should be expected, protective risk controls are essential.
We frequently see risk management systems relying on post-trade checks where clearing firms vet for algorithm and black box problems after the trades come through. While this model is functional and often cost-effective, it has an obvious weakness. The big issue, as the name implies, is that post-trade mechanisms function post-event. Relying solely on post-trade risk can mean taking serious losses before reacting to any problems. This danger is only amplified in the case of HFT where systems may be executing hundreds of trades a second. By the time the problem is identified, an individual trader may be wiped out, brokerages may be jeopardized, and, in worst-case scenarios like 2010's sudden nose-dive, entire markets can be affected. This can hardly be called risk management if it occurs only after damage is done. While necessary, post-trade risk management alone does not sufficiently protect markets and market participants in a high-speed, high-volume environment.
On the other hand, deploying a diverse set of pre-trade controls can overcome the issues described above, creating broad protection against risk before a dangerous situation develops. With multiple, simple pre-trade controls in place, traders guard against risk but also mitigate latency impacts, a critical factor in an ultra-competitive industry. Simple and effective tools include basic checks on order and position size, as well as order rate flow control. Effective pre-trade risk management may also include time-based limits such as daily traded lot maximums or ceilings on total contract value traded.
With these safeguards, traders gain risk protection tailored to individual tolerance and capacity. These controls also place minimal burden on processing time and trade execution speed. Pre-trade risk management brings additional value to trade system development by preventing errors from becoming problematic. As a result, traders gain opportunity to examine and debug algorithms and create better trading tools in the process.
In a basic scenario where one system trades one market, it is easy to see the value of pre-trade protection. Things get more complicated when liquidity-seeking traders deploy automated strategies across several markets around the world. In dynamic conditions, incorrectly implemented risk controls can burden a system to the point of speed disadvantage. I've argued that speed is a commodity, so when everyone is fast, how do you control risk but still capitalize on the fleeting opportunities for profitable trade that form the basis of high-frequency strategies? One way to tackle this issue is to set pre-trade risk management as the gatekeeper of the last leg to the exchange. Rather than returning risk-related feedback to a central location, adjusting, and resending messages, risk management should occur before an order is sent to the exchange.
In the trading community, there is a certain degree of reluctance to embrace risk management due to perceptions of increased latency with added cost and complexity in system development and implementation. This reluctance is short sighted as, when done right, necessary speed can be maintained and when it comes down to it, complexity is a requisite element of aggressive and successful trading. Traders might also regard risk management implementation costs as providing both insurance against potentially catastrophic future losses and a feedback system pinpointing key software weaknesses.
Naked market access puts all players at risk, so general acceptance and effective use of risk management is critical. Smart risk management strategy in HFT should incorporate both post-trade and pre-trade controls. With this multi-level protection, situations where trading systems lead to destruction visible beyond individual traders can be avoided. This should lead to increased participation and greater liquidity in the markets.
Yuriy Shterk is vice president of product development for CQG, the order execution, charting, and analytics provider.