PhaseCapital, a Boston-based “quant” investment adviser, announced their selection of StreamBase as their CEP platform. There has been a lot of interest in this announcement, for many reasons (e.g., here's the front-page coverage in Inside Market Data (warning: IMD is password protected)). I was involved in some of the interviews with Eric Pritchett, CEO of PhaseCapital, and there were 5 especially interesting things that struck me:
1. Increased use of CEP as infrastructure. PhaseCapital uses StreamBase at the heart of their high-frequency, low latency trading infrastructure. It's a better way to build event driven systems, as Eric was quoted in the IMD interview: "StreamBase technology allows PhaseCapital to abstract the rest of its trading architecture from the feed handlers themselves, enabling the firm to make changes to its data without having to re-engineer other parts of the system." In the past year, we've seen an increase in firms deploying CEP as event-driven application infrastructure, rather than an embedded component of, for example, and algo trading application.
2. Expansion of CEP use cases. PhaseCapital's application of CEP is very broad - it's at the heart of their portfolio management system and risk engine, for U.S equities, futures, and equity options. They use it for market data management of multiple exchanges to their trading system, and to order state handling and smart order routing. OK, I'll quote myself on this one: "CEP isn't just for algorithmic trading anymore." It's becoming a key part of enterprise infrastructure.
3. A comprehensive customer-driven CEP product evaluation. PhaseCapital performed a comprehensive evaluation - they installed and tested "all the major CEP products," and ran real use cases against the products. They looked at language, performance, usability, tools, and scalability of the CEP architectures. Eric summed it up nicely here: "Our
team consists of scientists and professional engineers; we are techies
at heart, so we appreciate the value of a well-designed system that
supports our development and deployment philosophy."
4. The continued rise of the need to automate. In the article The Five New Technology Priorities, I wrote about the first priority: automation. In one interview last week, Dr. Geoff Goodell, who ran the evaluation, described how critical CEP is to their company: "We're a small team of 15 at PhaseCapital - this CEP-based system IS our trading operation - everything is automated." I think the need to automate systems is at the heart of StreamBase's recent growth - firms need to do more with less.
5. StreamBase: the new market leader? I've spent almost a year watching CEP evaluations play out from StreamBase's side and I'm starting to observe patterns of who's winning business and why. In just the last six months, there's been a shift - publically, StreamBase has scored wins with BNY ConvergEx, RBC Capital Markets, BlueCrest, and now PhaseCapital. That's more customer wins than all the other CEP vendors combined in the same period. What's going on? It seems to me that the field is separating into "first generation" and "second generation" CEP. For example, most of Apama's attributes are first generation (single threaded server architecture, unmaintainable code generation for their CEP language, non-SQL oriented language, lack of debugging tools, and an adapter architecture that adds unacceptable latency due to its heritage of having CORBA embedded). Need proof? As Dr. Goodell described, Apama didn't even make the first cut in their evaluation. On the other hand, Aleri was the other final CEP product in PhaseCapital's evaluation, partly because they fared well relative to Apama because of its multi-threaded server architecture.
The real test of a market and products aren't what the vendors say about their product; it's what their customers say about their product, and it's exciting to see the CEP landscape and buzz about new applications, new industry trends, and new product category leaders as the market continues to grow.