ITRS Insights

Track order status and turnaround times, per client exchange or instrument, helping you observe transaction performance across a number of trading systems


It is becoming increasingly important to track IT activity and performance in real-time to support business functions. This can often involve the need to be able to understand and cross analyse semi-structured data (e.g. logs and alerts), structured data (such as system metrics and application metrics) and time series data (e.g. latency of trade execution throughout the day) in real-time. Previously, this was not possible; now it is.


The Application Support team in an Investment Bank were under pressure from the business to provide more valuable insights to support their business activities, particularly in helping them to maintain and improve relationships with key clients and to ensure that the business can safely trade in an increasingly real-time and connected world. With millions of trades flowing through a complex trading estate and into various venues for execution, there were many potential points of failure or delay, not only across the many different systems but also across the many different stages (and hops) of the trade lifecycle.
The business wanted to not only track the latency of its trading activity but to understand the data produced as quickly as possible. This was so they could ensure that latency was “normal” and that there were no weak points in their trading estate which could result in losses, both financially and/or attrition of end clients. The business recognised that any insights were incredibly valuable in feeding their strategies and operations, especially if they were real-time insights as this could give them a competitive edge or prevent the problems caused by poorly performing IT systems.
Previously, in order to track trades as they progressed through different applications (or hops) across the trade lifecycle, the Application Support team would need to log into several different systems and read time stamps to calculate the current latency. With rapidly changing trading volumes, it was difficult or impossible for a person to see poor performance as it was happening. And poor performance is a relative term. Is the current performance poor relative to a normal working day, or for the current volumes of trades. An absolute measure of latency can’t determine this.
Finally, if latency did increase outside of "norms", the Application Support team wanted help in identifying the likely cause. With so many variables to consider and previously, without help from tools, the likely cause was often hard to identify and rectify. ITRS Insights provides assistance with ‘causal analysis’ – using a combination of probabilistic causal algorithms and time window correlations to help identify factors which are likely to be contributing to the unusual behaviour.

Introducing ITRS Insights

ITRS Insights performs an impressive range of analytics across both real-time and historical data at a scale required by today’s big data world. This makes it suitable for running analytics on your business activities, your underlying technology or a combination of both. In this scenario, ITRS Insights not only tracks the trades throughout each hop, it also runs real-time analysis to calculate average latency and show it’s variation against the “norm” for this time of day or month.

Powerful queries: perform sophisticated temporal or text based queries in real-time.

Pattern discovery and anomaly detection;

  • Anomaly detection: advanced algorithms enable you to interpret periodic patterns in your data, so you can identify anomalies and future trends, and automate processes to protect against the occurrence of issues. Insights ability to analyse and store multiple streams of data from different sources means more information is readily available to query and compare, helping you find your answers in real-time. 
  • Machine learning: the machine learning algorithm tracks patterns by clustering similar patterns together and learning what a “normal” pattern/trading day looks like. By segmenting the patterns against time, for example by trading days, you can use the patterns to predict anomalies. As more and more data is fed into ITRS Insights over time, it increases its ability to recognise patterns seen before and to pre-empt scenarios or flag issues before they occur, increasing your confidence in foreseeing and protecting against potential issues.

Collaborative sharing of information: your queries, research and investigations are pulled together into one or a series of notebooks that can be easily shared with different audiences and stakeholders.

Analyse real-time vs historical data: now you can achieve a complete view of your data’s behaviour using one application. For instance, by comparing historical trades to current messages you can track trade flows that go through multiple hops, in different formats, from one place.

Secure and scalable;

  • Immutable: your data streamed into Insights is stored in a read-only format.
  • Clustered architecture: if a node goes down, the cluster can continue without issue, ensuring constant availability of both data access and analysis.
  • Protection of confidential data: “tenancies” enable users to access the data they need, whilst segregating commercially sensitive data.  ITRS Insights also allows a user to create a stored searching algorithm which can be used by other users without revealing the details of the search criteria, e.g. if the searching of log files for trade IDs were confidential, then the calculated latency can be exposed without being able to see the trades that calculated it.
  • Simple integration: ITRS Insights’ RESTful interface allows you to easily deploy and seamlessly integrate with a wide variety of data collection or visualisation tools, or your own proprietary systems. It also integrates fully with ITRS Geneos to allow you to capture all the data that is flowing through the Geneos estate.


ITRS Insights tracks the flow of trades throughout the estate and highlights any issues such as missing trades or unexpected latency. Using the text search capabilities, you can not only see the journey that trades are taking, you can also quickly calculate the average latency at a given point in time. This can be used to highlight any delays or potential issues. By running the real-time latency values into the time-series repository, a pattern is calculated for a similar working day (non farm payroll day, ECB announcement day, etc) and shows any variations from the current latency against the “norm” for this time of the day or month. Or by correlating the latency versus the trade volumes, show if latency is unusual for the trade volumes that the estate is processing – in real-time.  Finally, the Bayesian causal network analysis combined with time window based correlation helps the support person to identify the factors which may be contributing to the observed poor performance. 

Insights Product Tour

Take a look inside ITRS Insights

Request a call from one of our team