Case Study – Automated Trading Strategy for a CTA

Background

The CTA was using several disparate systems and technologies to collect market data, develop strategies and trade. As such, there were concerns over the maintainability of the operating environment, the operating costs and the time to market of developing new strategies.

The CTA had several objectives which had to be achieved on an evolutionary basis rather than “big bang”, in order to reduce project risk. The objectives were:

  • Reduce the number of technologies
  • Reduce time-to-market of deploying new trading strategies
  • Reduce latency between signal generation and execution
  • Create and maintain market data archive for strategy back-testing and optimization
  • Reduce total cost of application and infrastructure

 

Implementation

Given the desire for a phased implementation, we agreed and executed the implementation in three stages.

Stage 1: Set-up collection of streaming market data. Recording market data is a pre-requisite for back-testing and calibration. We installed TimeBase and the feed handler for client’s chosen market data provider.

Stage 2: Execute trades from existing signal generation process. Using TimeBase’s custom messaging capability, we developed a QuantOffice execution-only strategy which took as input signals generated from the client’s existing signal generation code. Co-located in Cermak, the execution strategy was run in StrategyTradingServer and trades executed on the CME via ExecutionServer.

Stage 3: Generate signals in Deltix. This final stage involved the client re-writing their signal generation logic as a QuantOffice strategy. By this stage, there was a historical data archive in TimeBase which had been collected since its deployment in Stage 1. As such, the client was able to test the signal generation against his legacy code. Once tested, the QuantOffice strategy comprising both signal and execution logic was deployed in StrategyTradingServer for live trading.

 

Products Deployed

 

The Results

  • Reduced the number of technologies to one
  • Reduced time-to-market of deploying new trading strategies
  • Reduced latency between signal generation and execution (resulting in more efficient execution)
  • Created market data repository for collecting and storing market data for strategy back-testing and optimization
  • Reduced total cost of application and infrastructure