CameronFIX Survey Captures Latest Industry Views on FIX
By Annie Walsh
CameronTec conducted a worldwide CameronFIX user survey last month to capture latest industry opinion concerning the adoption and broader usage of the FIX Protocol; movement in FIX volumes since the financial crisis hit; and the greatest challenges currently facing FIX, its users and FPL itself. Annie Walsh, Chief Marketing Officer for CameronTec sums up the findings while drawing on industry participants for further comment.
Respondents to the survey represent both buy- and sell-sides, across large to smaller firms, as well as exchanges and Independent Software Vendors (ISV’s). In terms of geographic location, respondents cover every major financial precinct.
For many firms where technology is a key market differentiator on their trading desk, the need to maintain a competitive system with all the requisite functionality has importantly included FIX.
Over 75% of FIX users reported an increase in FIX volumes; with FIX usage considerably greater since the financial crisis hit in 2008. FIX investment has remained stable with 60% of firms reporting no downsizing to their FIX team over the same period.
While overall FIX usage is up worldwide, many firms did however reign in investment on new FIX infrastructure, translating to strong stability for existing FIX frameworks. Both the versions of FIX used (72% are still using FIX 4.2) and the age of their infrastructure (four years on average) indicates that firms preferred to keep with existing frameworks over anything new during the crisis, but they are now shifting gears as they look forward.
“Performance is always on the radar but in harder times project prioritisation is more a factor for long-established firms,” notes Max Colas, Chief Product Officer of CameronTec. “At the same time, the emergence of High-Frequency trading has driven stronger demands on FIX engine performance pushing the vendors to continuously deliver.”
Adds Colas: “Long-established firms have looked at recent innovations and performance upgrades in FIX engines with mixed appetites, but survey responses indicate the need for upgrades is brewing with many looking to the future and making plans with partners that will still be there tomorrow.”
There was an overwhelming consensus among respondents for the need to explore ways to reduce latency within FIX, although this enthusiasm was also met with some reservations that such efforts could, in reality materialise with a tangible outcome.
“The major challenge for big financial houses, brokers and trading firms is to achieve ultra-low-latency per inbound and outbound FIX messages in direct market access flow,” says Erika Bajer-Jurkovic at Deutsche Bank. “It is very important to accurately measure and present these latencies in the real time and flag delays. An easy to follow, unambiguous latency-measurement standard on the top of the FIX Protocol could introduce and shape new metrics with a business to be considered, especially in the area of algorithmic trading.”
Commenting on buy-side firms driving FIX latency requirements, Jet Tek’s CEO Greg Orsini says, “While there has long been pressure from buy-side firms to continually shrink market data and transaction latency, there has been no standard way to measure and communicate total latency as well as all of the constituent point-to-point latencies. Recently, the FIX Interparty Latency Working Group has been tasked with defining FIX extensions to formalise how these discrete latencies can be presented. This will provide sell-side firms the opportunity to distinguish themselves from their peers by providing new information, which their customer can use to investigate network elements, make real-time routing decisions and eliminate slow liquidity providers.”
Vendor latency also continues to be a highly scrutinised. Despite some high frequency traders capable of using FIX today to trade in around 250 microseconds, in reality only a small number of FIX engine providers are currently producing engines anywhere near capable of consistently beating or surpassing 30 microseconds for message processing to deliver throughput of around 50,000 messages per second.
For many FIX users, another important issue is the growing gap between more recent FPL initiatives compared with the boiled down reality in the trenches. Industry-wide stickiness of older FIX versions (80% of FIX traffic still uses FIX versions 4.0, 4.1 and 4.2) continues to impinge on what would otherwise be a natural upgrade cycle.
John Heaney, Client Connectivity Manager for electronic market making firm Winterflood Securities points out: “We have a large number of connections trading multiple asset classes either on our internal liquidity or via our DMA platform; and within these connections a mixture of FIX versions 4.0 and 4.2 are being utilised. Winterflood, however, is yet to see later versions of the Protocol adopted for pure order routing among clients. While we currently implement FAST FIX to achieve low latent market data feeds and new FIX 5.0 trading connections to exchanges, we are not being asked by clients to implement FIX 5.0 connections.” Stickiness of older FIX versions also resonates with Deutsche Bank. According to Erika Bajer-Jurkovic: “In the real world there is a gap between versions being developed and versions being maintained with a majority of clients using FIX 4.2, and FIX 4.1 and FIX 4.0 still being present.”
“In general,” Bajer-Jurkovic continues:
“Unwillingness to upgrade existing FIX versions is widely present as there is a perception that it would introduce instability into trading patterns due to possible different interpretations of FIX fields and domino effects on the surrounding systems.”
Changes are afoot however, with respondents reporting current adoption of newer versions of FIX such as 4.4 at 18% and to a lesser degree 5.0 at 3%, but with demonstrable intent to upgrade: 63% of our surveyed clients say they will have us upgrade their FIX engine this year and of these 31% will upgrade in the next 6 months. Reasons put forward for upgrades are principally two-fold; firstly the promise of higher performance (60%) and secondly access to latest functionality in the engine (45%).
Of FPL’s initiatives during the past several years, a large majority of users find FASTSM, SCP, FIXT and FIX 5.0 less relevant. FASTSM however appears to counter this trend with a more enthusiastic approach to adoption of the FIX Protocol’s algorithmic trading definition language evident.
“While historically we found FIX easy to implement and that it works fantastically well as a method of sending and receiving order flow, posting IOI’s/trade ads and allocation messages, we are now looking into FIXatdl as a way of allowing clients to electronically instruct us exactly how they want their orders executed,” says John Heaney. “Whether this advantage will introduce new complexity, we are yet to find out.”
Erika Bajer-Jurkovic adds:
“Algorithmic trading is another fast growing area and while FIXatdl is still in its infancy there will be more and more in-house developed specifications which will be difficult to migrate later on.”
CameronTec’s CTO Steve Christinson highlights the varying costs associated with FIX operations: “For some firms obtaining and maintaining productive FIX connectivity can be a relatively inexpensive process. At the opposite end of the spectrum, creating a large-scale, continuously-operating FIX network serving tens of thousands of users can be a costly proposition.”
John Greenan at BNP Paribas Investment Partners says: “It’s important to ensure that the increase in complexity around FIX does not lead to increased costs for the end user in terms of support, implementation and management. Historically the simple focus of FIX to allow buy-sides and sell-sides to transmit orders and executions meant that FIX was ‘easy’. The expansion into a whole new range of business areas and workflows has meant an explosion of complexity. We, the FIX community have to find ways to keep it manageable even if we can no longer keep it simple.”
“FIX Protocol’s development roadmap is weighed down under pressure to maintain backward compatibility while addressing future feature evolution demands,” says CameronTec’s Christinson, “And the widespread adoption of FIX and its usage across an ever expanding range of market situations, will only amplify this dilemma further.”
“Many firms are finding that neither the build nor buy approach works, at an acceptable price point, and in isolation. Increasingly it is about finding FIX solutions and components which can be used standalone and mix and matched with existing infrastructure without comprising quality, performance and levels of service from the vendor.”
“Given the richness and complexity of the specification itself, defining FIX Rules of Engagement can be difficult and often counterparty specific. Complexity can arise around operations management; for deployment, configuration and monitoring of the FIX installation. In addition, integration with non- FIX systems plus measuring, achieving and maintaining best of breed performance e.g. when compared with native Exchange APIs, can present problems.”
“Up until very recently there was clearly a lack of affordable automated regression testing tools and suites available, although this is now changing with a number of entry level tools recently made available from vendors such as Jet Tek and Lasalletech, as well as the open source community, in particular the VersaFix initiative.”
In a true measure of the dedication endemic to those within the FIX community, CameronFIX survey respondents whole-heartedly endorsed FIX as future-orientated; with most believing it will still prevail in five years time. Presumably, this also means they expect the current drawbacks of FIX such as latency and over-customisation to be resolved along the way, which leaves as many challenges to address for the community.
What happens at the frontier of FIX with regard to communicating with other protocols remains unclear, and uncertainty persists over the broad course of convergence. A speculative view would say that users do not necessarily see it as a challenge, more so an issue of relevance.
The survey was carried out early May 2010 with a global sample of CameronFIX users representing 200,000,000 FIX messages per day and more than 4,000 FIX sessions.