The Case For Test Symbology
With Michael Rude, COO REDI Global Technologies, Gary Stone, Chief Strategy Officer, Bloomberg Tradebook, and Eugene Finkelman, Executive Director, Electronic Client Solutions, J.P. Morgan
Michael: Regulators are increasingly focused on electronic trading systems, and are enforcing rules and regulations to govern the processes, procedures and controls to promote a more stable trading environment. Market participants are generally responding with a more rigorous software development lifecycle and with improved documentation and testing practices. The net effect of these changes could be a more controlled, compliant and stable marketplace.
That said, there is opportunity for improvement around go-live testing. Market participants who are releasing system changes to production are required to perform a pilot test to ensure the system is performing as expected. However, most exchanges do not allow ‘test’ orders in production, or any order that does not have an economic purpose. Absent the availability of test symbols in production, market participants are limited in their ability to validate market readiness before the market opens.
It’s not entirely clear why the exchanges resist the adoption of test symbols in production. In some cases, it’s because the addition of test symbols requires technology changes for the exchange. But it seems incongruous to have such extensive rules and regulations governing the development of electronic trading systems, but such limited tools to evidence the efficacy of our systems in a live environment.
Gary: I cannot understand why regulators are silent on this. There are regulations that we cannot test with live orders. Yet we are forced to because we have no other method of testing connectivity. Fragmented markets are also a real concern. Regulators need to standardise the test tickers across markets in a single asset class for Smart Order Routers to understand that they have connected and are sending appropriate orders IN PRODUCTION. Just because beta works, it is an assumption that it will work in production.
Michael: Whether it’s from an execution standpoint or from a clearing standpoint, it is universally viewed as a systemic risk if we can’t test systems across asset classes. There are certain test symbols that are available for use on NYSE for example, but such use is not permitted globally across asset classes.
While we are happy to acknowledge that it makes perfectly good sense to test your systems, at the moment, before we start rolling out into production we’re not allowed to do that last leg; we’re not allowed to certify in a live environment. Even the brokers aren’t allowed to send in “test orders”; they have to be real orders with real economic purposes behind them.
Eugene: Whether testing a change or validating production status, market participants need a safe way for interacting with production environment.
As part of the FPL Risk Mitigation Symbology Working Group, we are in discussions with one of the exchanges right now trying to connect two and two together: we want to test and a dedicated symbol to test will help in that effort. If we do testing, should it be a dedicated symbol or other random symbols? Not everyone sees the need to use test symbology just yet.
How will these work?
Eugene: The working group has put out requirements around the creation and expected behavior of these symbols.
At a high level, the key is for these instruments to mimic the exact behavior of any other instruments on that market, except these would be zero-funded no-risk securities.
They should be listed, have standard identifiers (Rics, ISINs, etc..) and should support standard order types in those markets.
They should be available for trading regular market hours in addition to pre and post market sessions where participants can route orders on these and receive standard electronic acknowledgements and executions.
Eugene: This would help exchanges to make their environments safer, raise overall market confidence and establish consistent protocols for everyone to follow.
Achieving all this will provide an edge for early adapters as it would be an innovative and effective tool a liquidity venue can offer to clients in helping them address their risk concerns, which may in turn lead to increased flow and participation in such venues.
Michael: It’s possible that exchanges could have an economic incentive- for example, the Hong Kong Stock Exchange charges for the use of its test environment.
So the top level aim of the entire project would be that live testing symbols be used so that you don’t have to test orders live; which increases the stability of the orders that are going in, builds confidence in your own systems and also builds confidence in the market itself?
Eugene: The whole project is designed to provide the industry with a safe means of validating production environment, by establishing test symbology for all asset classes and recommended best practices around use of such symbology. This helps to establish a consistent and reliable business process thus reducing the risk behind production changes and increasing confidence for market participants.
Gary: It could be argued that Knight would not have happened if they could have linked to a test ticker and ran before using real tickers.
Why hasn’t this being done already?
Eugene: Participants need time, resources and money to make this happen. This could mean establishing processes to make sure that they have everything available to support this. For example, there’s not going to be natural liquidity out there so you need to be able to provide liquidity for these test symbols. It starts out with the exchange or whatever liquidity venue first. If at least we have one, two, three venues going then we can use this in turn to start having buy-sides, sell-sides or OMSs supporting this.
One of the points to be discussed is whether the exchange could set a threshold on their side: if there are concerns that someone could flood the system with a big volume of test orders they can set up thresholds to only take X number of tests. That should be part of the best practices around this, for the exchange to think through.
The best practices document is being scrutinised by one of the big exchanges right now. Best practices include recommendations for all market participants to use this test symbology for verification of all production changes whether implemented by buy-side, sell-side, OMS or the exchanges themselves. Verification of connectivity and systems are part of this. This could be used by firms in the morning as a start of day check to make sure that they are connected properly and that everything is working as expected. Then as a tool for post outage verification, where if an issue is encountered intra-day firms can send test orders to confirm everything is backup prior to resuming live trading.
If you create a test symbol that you can use to place an order, get an execution, and complete the allocation and clearing process – you can then truly test out every piece of the infrastructure within the lifecycle of the trade to ensure that everything is working as designed.
Eugene: The aim of the working group is to have test symbology across all asset classes; Starting with Phase1, and the current focus, to at least have these available for electronic order routing and execution on the major markets. With the ultimate goal of the whole project to be able to do the full end to end lifecycle, including allocations, clearing, et cetera.