Bringing Predictive Intelligence into Trading Operations11.13.2014
By Theo Hildyard , Software AG
We hear increasingly these days about how the focus of electronic trading is shifting from speed to intelligence. In fact, my previous blog post was about how it is essential for firms to be not only fast but also smart.
But within the context of trading operations, what exactly does “smart” mean and how can it be achieved?
To my mind, intelligence in this context is essentially about the combination of three factors:
- assimilating many (and fast moving) data points
- predicting likely scenarios based upon that data – the basis for decision making
- taking the best course of action based upon those predictions, often fully automated
This is where the real advantages of Complex Event Processing (CEP) and streaming analytics platforms come to the fore, in that they enable all three of those factors to be combined.
By aggregating, correlating and analysing historical and streaming event-based data from multiple sources in real-time, normal and abnormal patterns of data or associated behaviour can be identified and detected. But more importantly, predictions can be made about when something abnormal is likely to happen. This in turn enables decisions to be made ahead of time so that appropriate action can be taken.
This is powerful stuff and has a wide range of uses in financial markets. So it is perhaps not surprising that firms are increasingly implementing these platforms to bring predictive intelligence into multiple points along the trading logistics chain.
One example is in Market Surveillance and Monitoring, where market participants can go beyond one size fits all real-time rules and instead create a very granular view of what is normal for that client in that asset class at that time of day. Armed with this “intelligence”, streaming analytics can not only detect deviations for normal but pre-empt them, giving risk and compliance teams a vital window in which to act. And of course because the data required to deliver this “intelligence” is big and the environment that consumes it is low latency, CEP plus some cumbersome database is not enough. In-memory data management is needed to deliver the application performance at scale and low latency messaging is need to glue it all together.
Another example is where a broker uses this technology to adopt a more customer-centric approach to pre-trade risk by providing pre-emptive updates and alerts to customers on positions, use of limits etc., throughout the trading day. This contrasts with the typical image of pre-trade risk – a system that is driven by enforcing hard limits in order to protect the broker.
In conclusion, early adopters of streaming analytics platforms are already seeing significant benefits from using this technology. The kind of predictive intelligence they offer, particularly when combined with other technologies such as low-latency messaging and in-memory data management, is enabling firms to access a mine of hitherto untapped opportunities.