Arbitragis Picks Quant House for Arb Algos

March 24th 2008
Vol 23 No 25
DATA CONSUMERS

Arbitragis Picks Quant House for Arb Algos

French proprietary trading firm Arbitragis has begun using low-latency market data from local vendor Quant House to power its algorithmic arbitrage trading strategies, officials say.

The firm employs between five and 10 traders and technical staff in Paris, and trades its own capital using algorithmic strategies, says Arbitragis founder Tuan Nguyen. Because of the nature of its trading operations, the firm sought out a data vendor that could provide the lowest possible latency and help to integrate the data into its proprietary algo trading engines, which are coded by a team of in-house developers.

The firm began using the data earlier this year, following initial discussions that began at the end of 2007. "We found that Quant House had very good technology and they were willing to [customize it to] comply with our systems. In this space, I think customization is really the key," Nguyen says.

He declines to provide any details on the datasets that Arbitragis is sourcing from Quant House, but says the firm was particularly pleased with the flexibility of the vendor's application programming interface (API), which it uses to access the data. "For a start-up like ours, they were willing to do a lot of things that some of the larger vendors would not have done," Nguyen says.

According to Stéphane Leroy, head of global sales and marketing at Quant House, the vendor's market data API uses a publish-subscribe mechanism that allows customers to select individual securities for which they want to receive data updates-in effect creating a customized feed, which enables clients to limit the bandwidth required to connect their trading applications to the vendor's point of presence, resulting in lower connectivity costs.

Allowing clients to choose the securities for which they receive data also helps to control delays in data dissemination that result from spikes in data traffic. "The client is not obliged to receive the entire universe... [so] from a performance perspective, clients just get what they want, which limits congestion," Leroy says.

At present, Arbitragis only uses Quant House's low-latency datafeed, and is not using its trading strategy development toolkit or market data replay service for testing strategies, since the firm has already built its own strategy development framework, along with market data storage and replay facilities to back-test its strategies.

"We need a lot of expertise in terms of computer science to optimize our trading software," to take advantage of Quant House's low-latency data, Nguyen says. In fact, the firm uses such a high level of automation that strategies can even turn themselves off in real-time, without any manual review or intervention, should they stop yielding profits.

Nguyen says that because the firm trades only its own capital, and does not have to answer to clients or shareholders, it has a lot of flexibility in the strategies it trades-including methodologies that might be considered highly unconventional by most investors, but which require fast and sophisticated data. "The models that we use can be derived from seismology, astrophysics or computational biology-anything that somehow connects to our activity in terms of data modelling," he adds.