Part 1 of 4
Complex Event Processing is, always has been, and will continue to emerge as much more than just a platform for algorithmic trading. Yet, "Algo Trading" is hot and CEP is definitely a new technology that can be leveraged for strategic advantage in support of algo trading.
While you may need a PhD in Mathematics to develop the hot new strategy, it certainly doesn’t take one to see the synergies between algo trading and CEP. Increasingly pronounced is the uptick in buzz around the Complex Event Processing market and its ability to service the emerging algo trading demands. While a lot of the buzz may be around algo trading, people are quickly realizing the power of applying CEP technology across a wide range of problems – anywhere there is a need to analyze fast moving data in real-time, with minimal latency.
At Aleri, we’re certainly seeing the interest in CEP for algo, but that only represents a distinct minority among the places this technology is being applied.
Before I get into a discussion of the applications being built with this technology, however, it’s worth a short diversion to look at the evolution of the technology to get it where it is today - the fact is the CEP market has evolved quite a bit in a short time.
There are a number of newer players, some with broad solutions, many more with niche offerings around everything from tick-capture to order book management to compliance. And while all of the providers of general purpose CEP platforms claim to be able to address a broad set of use cases, this is where it can become difficult to distinguish which products are suitable for a particular application.
The first implementations of CEP tended to treat event data as transient, having no lasting value, and focused primarily on what I’ll call “situation detection”. The presumption was that the utility value of the data diminished immediately after reading the event information, like a stock price.
Most of the algorithms were built to effectively do complex filtering, looking for a situation that warrants a response. As you might expect, this was well suited for algo trading applications. What is wasn't well suited for, however, was many of the wider range of event processing applications that firms were contemplating, such as managing order books, performing market depth analysis amid fragmented pools of liquidity, determining smart order routing, doing a real-time P&L mark-to-market, much less any risk aggregation.
The CEP market today has evolved in line with the market demands. Our discussion here is too limited to probe in depth the various paths in which this technology developed, but it is clear that the trend is towards:
* the ability to manage stateful event data in a streaming environment
* capabilities for more advanced processing logic extending beyond rules to include both relational operations and procedural logic
* better capabilities in passing data into historical repositories
* enterprise-class attributes including scalability, high availability, security, data recovery and manageability
[Next week, in part 2, Don will explore some specific applications of CEP, beyond algo trading]