A custom analytic module can be added to the engine using the Plugin Framework. This framework allows custom functionality to receive the same performance benefits as the Velocity functions by executing in the same process space as the data store. APIs are available for C, C++, C#/.NET, and Java.

A module can take advantage of the following functionality within the Analytics Engine:

• Receive data from the Event Processing module
• Manipulate this data and insert it into the generic data store using a Velocity function
• Query additional data using a Velocity function
• Publish data to clients using the EventStream interface
• Respond to queries from clients using the query interface

Modules can use the Query Interface and EventStream Interface in combination to communicate with external applications or to support an external interface.

The query interface allows data to be interrogated; a result set is generated and returned to the external application. This interface parses queries and invokes the Velocity functions which operate in the data store to return the appropriate data to the client. Velocity supports the Tcl scripting language.

The EventStream Publisher is used to generate event triggers, complex analytics, or other real-time notifications in an integrated module. EventStream also interfaces to all of the major off-the-shelf publisher applications. The amount of new code to be written to enable event publishing is minimal as the Velocity platform comprises many reusable modules that can be plugged together using an XML configuration file. This enables the user to focus on the business logic and not the underlying plumbing.

EventStream provides the ability to enrich incoming data using algorithms deployed directly in the data path within the engine. In order to act instantaneously on trading opportunities, the EventStream code is executed prior to data storage to ensure low latency. This is especially critical when running hundreds of simultaneous analytics. For example, EventStream is capable of publishing the Volume Weighted Average Price (VWAP) of each component of the Standard & Poor’s 500 to multiple trading applications subscribing to the information, all in a matter of milliseconds.

Because algorithmic trading analyzes past patterns to identify new opportunities, EventStream was architected to provide access to incoming ticks, intraday data and years of historical tick data all in one module. Complex alerts can be configured to scan the entire depth of book in memory and identify trading opportunities that other applications don’t have the power to recognize. The EventStream APIs were designed to make publish and subscribe development quick and easy, with built in failover capability.