banner icon

Client case study: Reducing equity trading losses in complex environments

Achieving the level of transparency that inspires confidence

Keen to ensure the reliable low latency execution of trades across a highly complex equities trading environment, our client wanted to swiftly identify latency sources and when an issue emerged, pinpoint its location facilitating the problem’s quick resolution.

Our client was operating a highly complex equities trading environment, offering Direct Market Access (DMA) services using a Smart Order Router (SOR), algorithmic trading engine and an internal crossing system as part of their execution processes.

Trading on behalf of their customers, ensuring reliable low latency execution proved critical to their customer’s satisfaction and ultimate retention.

The Challenge: Latency detection across the complete equity execution process from order origination to fill

Our client wanted to be able to effectively identify any sources of latency within their infrastructure adversely impacting trading success and when an issue emerged, be able to quickly ascertain its location and diagnose the root cause.

To achieve this, our client required the ability to trace the path taken by loss making trades, so they could measure the latency experienced at the different stages through infrastructure monitoring.  This was complicated by the fact that their customer orders entering the environment could be executed through various means including:

Exchange Orders

To avoid market exposure, it was common practice to divide larger orders into a series of smaller orders for execution, often distributed across different markets.  These orders would be managed by the SOR.

In these situations our client wanted to understand the timing metrics between the customer order arriving at the SOR and the system then emitting the first exchange order, so they could measure responsiveness.  In addition to this, our client also requested that the timings for all other exchange orders executed as part of the original client order were measured, so the timings between the placement of the first and final order could be assessed.

Internal Crossing

Prior to sending orders to the exchange, the SOR would check if the order could be internally crossed using the client’s own book.  If this was the case, the order would be executed at this point, if not it would proceed as an exchange order.

Our client wanted to understand if unnecessary latency was being introduced through this process.

Algorithmic Orders

The decision to trade certain orders was done through our client’s algorithmic trading engine.  On generating an order, it would then be sent to the SOR for execution.

Our client also wanted to understand the latencies introduced by this stage of the process.

These combined processes created a situation where single clients orders could generate multiple algorithmic orders, and each algorithmic order could in turn generate multiple exchange orders, as demonstrated by the following diagram:


Keen to track the relationships between all of these orders, gain timings at each stage and track the execution and acknowledgement reports through the entire process, our client could then reveal latency hotspots within their equities trading environment.


The Approach: Performance monitoring at multiple points throughout the entire order execution process


The monitoring points Velocimetrics was instrumented in.

To effectively monitor this highly complex environment, Velocimetrics chose to capture data at multiple points throughout the entire process, as demonstrated by the diagram to the right.

Internal protocol decoding to monitor middleware messaging traffic

Our client had employed the services of a broker-less message bus system in an effort to minimise messaging latency.  Velocimetrics was able to write a protocol decoder so the conversations taking place on this bus between the client’s algorithmic trading engine, its SOR, crossing engine and exchange connectivity infrastructure, could be monitored at the network level.

Business flow analysis to identify parent and child trading relationships

Once the data had been captured, it could then be re-inflated into the business objects it represented such as an order and all required information instantly sent to the Velocimetrics in-memory cache for analysis.

Through a process of correlation, related information was then linked back together to reconstruct a seamless end-to-end flow for each business object.  Association techniques were then applied to identify the relationships between different flows, such as parent and child orders, and timings for each individual hop recorded.


The Results: Complete trade performance analysis enabling latency hotspots to be rapidly revealed 

Our client was able to search for any given order and understand its complete path, regardless of whether or not a common trade ID was retained as the original parent order was divided into multiple algorithmic and exchange orders.

By measuring both hop-by-hop and end-to-end latency across these paths, our client was able to gain all of their required performance metrics, use this information to identify latency hotspots and drill down to detect the issues’ exact location, facilitating its speedy resolution.  This approach achieved the client’s ultimate goal of minimising the impact of latency on trade execution speeds.

Interested in seeing what Velocimetrics' technology could offer for your firm?

Schedule a demo and see how uncompromised visibility accompanied by performance-improving analytics could transform your business insight

Find out More

Client summary:

  • Firm type: Investment bank
  • Asset class: Equities
  • Location implemented: London


  • Identify latency sources across complex order execution processes
  • Effective trade performance analysis


  • Multiple monitoring points
  • Complete trade reconstruction for parent and child orders


  • The ability to trace the path taken by any order regardless of whether a common trade ID survives throughout the flow
  • Effective detection of latency hotspots
  • Reduction in trading losses