Remember me

Register  |   Lost password?










mikeohara's Blog

In brief, colocation beats the speed of light

April 9, 2015 by mikeohara   Comments (0)

Sea-based XbandA recent commentary article Physics in finance: Trading at the speed of light (Mark Buchanan, Nature, 11 February 2015) made a seemingly logical statement: In order to take advantage of price differences between two distant financial exchanges, it is best to station a trading strategy computer at the midpoint between the two exchanges, for example, on a ship in the middle of an ocean. The geodesic midpoint is where the two price levels can be compared at the earliest possible time. Therefore -- the enticing logic goes -- this is also the ideal point where the arbitrage decision can be made -- buy at a low price on one side and sell at a higher price on the other side.

read more...

, , , , , , ,

How to evolve a risk-based FPGA strategy into a value add strategy

April 8, 2015 by mikeohara   Comments (0)

Written by Dr Marcus Perrett, Director of Technology and Development at Fixnetix. 

The use of FPGAs in finance was originally driven by regulation. Brokers were encouraged (or mandated by regulation) to have controls in place to monitor and, if required, cancel or stop clients trading. As clients had traditionally enjoyed direct access to the market via a broker, a system that was positioned between a trading system and the exchange needed to be a fast as possible to reduce the impact to a client’s trading strategy. Hence, FPGA technology was employed to great effect in a number of configurations; some banks built their own systems and several vendors built their own. Fixnetix has enjoyed success using that model in Canada, where regulation has driven the need for a low latency risk solution for latency sensitive clients trade flows via a broker.
 
However, regulation alone has not proven to be the only driving factor for the use of FPGA in the finance area. For example, in Japan, the native exchange (TSE) utilises a complex and verbose protocol called arrowhead. In addition, the rules around erroneous trades and protocol discipline are onerous and for potential new entrants into that market it can be a daunting task to overcome.
 
Several brokers have solved this issue by offering FIX to arrowhead translation services to allow easy and ubiquitous trading, the downside being that such systems are very slow due to the amount of processing required to perform the translation into arrowhead and even more so in generating FIX compliant messages in the return path; this requires Persistence (storage of incoming FIX tags for later use) and Enrichment (calculation of values not available in arrowhead message such as Average Price).
 
What did Fixnetix do?
 
Fixnetix took on the challenge of replicating such functionality in an FPGA and furthermore, incorporating a single-to-many architecture that allowed a single FIX connection to trade at a high rate with the translated arrowhead orders spread over multiple virtual servers. This functionality was combined with APAC specific risk checks that are far more onerous than the SEC's 15c3-5 or IIROC & CSA's NI 23-103, requiring checks such as a live per-stock position and multi-tick and lot size checks. The resulting system performed in 2.5uS what it took a software system 1-2mS to achieve and is the only example of such capabilities encapsulated within an FPGA, with the risk abilities being almost secondary to the translation abilities that are realised.
 
FPGAs - a value add service?
 
An implementation such as that described above highlights that, rather than FPGAs being viewed as fast gatekeepers, they can offer value add services that make a material difference to the fill rates client achieve. Fixnetix has been able to leverage its unique architecture (where the client and the venue are independent at a network level) to enable such value add functions and has in development several others.
 
One example is offering clients who are trading on FIX markets the ability to send smaller binary protocol messages to the FPGA, which translates them to FIX extremely quickly; the benefit being that some clients can generate binary orders much faster than an equivalent Fix message which, when viewed holistically, provides real end-to-end latency improvements.
A second example would be some intelligent order routing and generation capabilities. Here the FPGA could bundle similar orders together or split them up to achieve best execution, route them to an internal (to the broker) dark pool or stock holding, or translate order types as required. Such capabilities are only possible via the EMS-on-FPGA approach, as they require not just application message inspection for risk, but parsing, processing and synthetic message generation.
 
A future with FPGAs
 
It can be seen that the journey of using FPGAs for financial applications is still in its infancy and as use cases extend many more applications beyond those detailed here will appear. Fixnetix envisage that within 3 years, the application of FPGAs as risk filters will be both saturated and commoditised and only those able to add more value into the trade flows will benefit. It is here where the EMS-on-FPGA approach positions Fixnetix at the forefront of this new charge into value creation with its iX-eCute product range, backed up by its pedigree and current deployment footprint across the globe.
 

, , , , , , , , , , , , , , , , , , , , ,