One of the many challenges when operating a successful trading venue is creating a balanced order book that includes activity from across the whole trading frequency spectrum. For example, high frequency market makers' business models depend upon other participants from across this spectrum to hit/lift their quotes.
This diversity means that attracting a suitably broad spread of participants necessitates offering a similarly diverse range of functionality that caters for the specific needs of each category of traders. At present, we see these needs falling into three categories:
• Trading functionality - functional facilities and technology oriented items
• Risk-related functionality and technology - particularly in view of the acceleration of the automated trading process, strong risk controls are essential.
• Market structure - creating a suitable mix of functionality and technology that will encourage a range of types and time frames of trading activity, which will create a balanced order book structure. A critical element in accomplishing this is minimising barriers to entry as well as offering a broad variety of access options.
The complete price picture
The needs of high frequency traders and market makers are typically the opposite of their lower frequency peers. Mainly, high frequency participants just want things to be simple and fast, but one area they are very interested in is market information - particularly the quality and content of order book information.
Understanding the order book structure is a primary interest for high frequency traders. While most trading venues publish aggregated order book data, this provides only a partial picture. One of the most important factors in determining order book structure is not just the total size at each price level, but how it is composed. This is why the updated version of the Eurex market data feed "Enhanced Broadcast Solution" due for release in November will include an indicator showing how many orders each price level consists of.
A trading venue's outright trading speed is obviously important to high frequency traders. However, as round and single trip times have fallen into the very low milliseconds, a different priority has emerged - reducing the standard deviation of a venue's order round trip times. High frequency traders do not just want pure speed but consistent speed, as this makes it easier for their trading strategies to deliver similarly consistent returns.
Providing tight spreads in a fast moving market results in a lot less risk for liquidity providers when the response times of a trading venue are consistently stable, which is why reducing the volatility of trading platform performance is vital. To some extent this is correlated with a trading venue's technological capacity; if a market is running hot, there needs to be sufficient investment in contingency networking and processing capacity to ensure that performance outliers are kept to the minimum.
Eurex Enhanced Transaction Solution Interface - Round Trip Times & Futures Orderbook Transactions
It is also important to note that different high frequency traders have differing priorities in this area. For some, inbound single trip times are all that matters; someone with an aggressive trading style only needs to be fast enough to hit/lift an attractive price before the competition. They will be far less concerned about the return response time from the matching engine. By contrast, a liquidity provider is taking risks all the time as their orders are exposed in the book. Therefore, if they modify an order to a new price they need to know as quickly and consistently as possible whether or not the modification was successful, which means that they are keen to have the lowest possible round trip (as opposed to single trip) times.
A related issue for both types of traders is to understand how they perform in relation to their competitors. Any benchmark data the trading venue can provide, such as anonymous best in class times between an exchange gateway and the client application, adds value. Armed with this type of data, which is provided by Eurex, trading firms can assess their performance relative to their peers and make the most appropriate investment decisions regarding their technology.
Missed, but by how much?
While market makers are obviously concerned with the trading venue's round trip performance, they also need the data that will help them maximise their own performance and its volatility. For example, a market maker might submit an order modification but find that their original price was hit/lifted before their modification reached the matching engine. The key point here is that while the occasional miss by 10 or so microseconds is probably just bad luck, misses of 100 microseconds or more (especially if more than just occasionally) might suggest that the functional trigger for their order modification is being transmitted late. Being able to discriminate between these two scenarios is clearly essential for the optimal tuning of the market maker's internal technology.
Number of Processed Transactions at Eurex Exchange & Response Times
In response to this need, the next release of Eurex's system will include two important additional data items: the trading system will send back an accurate timestamp with each response to an order/quote transaction showing when the order/quote transaction actually hit the matching engine. Furthermore, if an order or quote is hit/lifted, the Eurex market data trade message will contain an additional timestamp showing when the successful aggressing order reached the matching engine. Apart from any immediate application, this additional data can over time be compiled into a valuable database that will allow the market maker to track and respond to trends in their trading technology and algorithms.
The speed with which high frequency trading P&L can swing around makes robust risk controls essential. While it is obviously not a trading venue's responsibility to undertake individual participants' risk management for them, it can at least offer tools to facilitate this and set some sort of boundary conditions. In doing so, it can help mitigate risk for both individual entities and the market as a whole.
One important element in this process is ensuring that participants have access to as much timely and relevant data as possible to assist them in managing their risks. In a high frequency environment, position information alone is insufficient; trading venues and respectively their associated clearing houses, also need to be furnishing firms with their current margin requirements and collateral usage in real time as their positions and market prices change. This not only delivers a more complete risk picture, it also allows firms to better manage their collateral as well as anticipate any additional future margin requirements. A further advantage is that this type of data is entirely non-intrusive to the trading process, so the traditional concern of high frequency firms that risk management delivered by exchanges or clearers causes latency, does not apply.
If inadequately risk managed, high frequency trading can cause disruption to other participants, such as clearing members. For that reason, some trading venues together with their clearing house already provide clearing members with a manual/automated "emergency button" that can be used to stop all trading activity by a client firm whose activity is breaching agreed risk limits.
While effective in its way, this is a relatively blunt instrument. A more progressive (and effective) approach provided by Eurex is to offer a more graduated set of tools:
• A primary trigger that can be set by a trading firm or a clearer that simply sends out alerts to concerned parties if a particular level of margin consumption or risk limit is exceeded. Clearers might use this as an alert to contact the client and warn them of the situation. This would be followed by...
• ...a secondary trigger, which if tripped would enforce a minimum time interval between two arriving orders or modifications for any given product. In addition to handling any problems relating to margin consumption, this type of control is also well-suited for dealing with an automated trading model that is running out of control - or any other sort of technical failure.
• The manual/automated "emergency button" already used.
None of these three risk control levels would have any effect on day to day trading activity and the second and third would only have an impact when the risk situation was already becoming prejudicial.
However, an important additional consideration with this type of control mechanism is the need to keep it simple; for example, by only applying it on a single currency amount per (non-clearing) member. Particularly from a clearer's perspective, trying to apply this on a per trader, per product, per sub-account, per currency basis is impractical, as this needlessly introduces both complexity and the risk of misconfiguration.
In order to achieve the balanced order book that benefits all a trading venue's participants, it is vital to reduce the barriers to traders' market entry. Doing this requires an understanding of what individual categories of traders actually require in order to operate their business. For example, while high frequency traders need the minimum latency and the most comprehensive order book data, others do not. For example, many trading venues have historically focused upon delivering new functional features (such as new order types) to those trading at slower frequencies, but have neglected to reduce the technology barriers confronting these participants.
For example, it is counterproductive to penalise this type of market participant with the costs associated with connecting to and handling a high frequency market data feed. Apart from any subscription costs, handling such a feed necessitates paying for a great deal of bandwidth and storage space. A far more appropriate solution that minimises participation barriers is to offer a netted data feed that transmits only the information the end user actually requires to operate their businesses.
As well as stripping down a feed to the bare essentials required, another valuable step is to minimise the barriers to its implementation by using common standards. FIX is the classic way to accomplish this and also hugely simplifies order management for people already using standard FIX engines and order/portfolio management systems. The traditional objection to this is that FIX is too slow as a protocol for market data transmission, even for lower frequency traders, but the introduction of FIX over FAST now addresses even this hurdle.
All three of the areas outlined above play an important role in stimulating a mix of trading frequencies and styles on an electronic trading venue. However, a critical theme that needs to overlay them all is transparency. Historically, many trading venues have been less than forthright in revealing details of their technological and functional performance. This is no longer appropriate; today, a venue's success increasingly depends upon the quality of its engagement with its customers.
The trading venue that is prepared to share details of its performance with clients - such as the time taken for a message to traverse each hop of its internal architecture - is not compromising itself. Rather, it is effectively collaborating with other interested professionals who can contribute additional intellectual capital for the potential benefit of all market participants.