“Kill switches”: Diverting Focus From Automation’s Broader Challenges
Context is of primary importance when taking stock of the proposed initiatives addressed at the recent industry meeting in Washington, D.C. during which the Securities and Exchange Commission (SEC) and the exchanges addressed the state of market infrastructure following NASDAQ’s August 22 three-hour trading halt. (If you missed the reason behind the failure, it’s since been confirmed that the securities information processor (SIP) suffered a technical failure.)
During the dialogue, SEC officials asked U.S. stock and option exchanges to create “kill switches” that would enable them to shut down trading when technological failures occur. And the debate went from there.
From my perspective as a former trader, I think the term ‘kill switch’ has the ability to oversimplify and divert focus from the broader challenges facing the exchanges. However, the comments that circulated in regards to exchanges having ‘homework to do’ are spot on in terms of the work required to assess the situation we are now facing. There is a meaningful body of work required to identify risks and issues, largely coupled with the fact that capital markets have gone through an unprecedented automation evolution in a relatively short time frame.
Breaking down the challenges into functional groups makes sense, especially with respect to the ‘kill switch’ concept. Executing brokers perform a role in the market that is, by nature, different from execution venues. We should ensure we separate the challenges. An exchange not functioning correctly is a risk to the entire marketplace if – as the trading industry is now – everything is interconnected. On the other hand, a broker not functioning correctly is merely a risk to itself and to market confidence, but is not a bottleneck (as other brokers can step in to execute).
The functional groups within this issue are:
- Surveillance – identifying there is an issue. (With infrastructure failures such as NASDAQ this is self-evident very quickly, though with issues related to automated trading, such as Knight Capital, it may be harder to spot immediately).
- Kill Switches for brokers. (To prevent erroneous trading)
- Fail-Over switches for exchanges.
- Disaster recovery procedures to reduce operation downtime.
Exchanges require the ability to transfer the responsibility of price matching, should there be an infrastructure issue that prevents the matching of buyers and sellers on their own platform. As such, I’d rephrase the exchange requirement from a ‘kill switch’ to a ‘fail-over switch’. Also by saying ‘switch’ that should include both technology and procedures. Mistakes happen; if identified in time, the ability to fail-over to a functioning alternative is essential. It should take minutes not hours. It is here that procedures are essential, as is the construct and regular practicing of disaster recovery solutions.
In an institution that is creating and executing orders such as a broker, it is essential to employ appropriate surveillance technology, and then, if a notable change in behavior/an error occurs, be able to shut down trading quickly with a ‘kill switch’. Noting that shutting down trading is not as easy as implied, it requires sequential system interaction to ensure risk is managed appropriately, and, as such, the investment required in both time and technology is meaningful.
In summary, there is a common theme here that isn’t ‘new’ news. And that is the fact that our world has an automation-driven complexity that is not going to change — and in fact, you can count on it becoming even more complex. What has to change is the investment in surveillance and risk management that will ensure processes and technology keeps pace with these changes.
With recent events like the NASDAQ failure, we are seeing warning signals for what may be much larger issues down the road if we don’t make the necessary investments and preparations now.