Systematic Trading Strategy
A multi-portfolio, multi-asset trading model automated for end-to-end execution after back-testing and approval.
The strategy and the system
This build covered both a systematic trading strategy and the infrastructure required to automate its complete execution across portfolios and asset classes.
My part
I co-designed and back-tested the strategy using Excel, VBA and C#, then designed and delivered the automation infrastructure.
The interesting bit
I've always enjoyed looking for information edges in data. I worked with a number of very talented people, and we used statistical models to try to find edges we could trade. The part I particularly enjoyed was portfolio construction: turning signals into a portfolio rather than treating each one as an isolated trade.
The hard part
Backtesting a systematic strategy properly takes a huge amount of time. Buying historical data is just the beginning. You have to clean it, get calendars and market holidays right, check the timing of every assumption and keep future information out of earlier points in the test. Otherwise a precise-looking result can give you confidence you haven't earned.
To trust a backtest, you need something close to a laboratory that reflects the real trading environment and minimises bias. Finding an apparent edge is relatively easy. Establishing that it genuinely existed at the time is much harder.
What happened
The strategy operated from April 2011 to May 2012 and returned 10% during significant market uncertainty before being discontinued after a key portfolio manager departed.
Technical notes
- Architecture
- Desktop and command-line processing solution
- Deliverables
- Microsoft Excel workbook · .NET command-line application
- Deployment
- On-premises user desktops
- Technologies
- C# · Microsoft Excel · VBA · ADO.NET · Sophis API · bespoke file-based database
Looking back
We put a lot of effort into the strategy and portfolio. I was disappointed when it stopped after the portfolio manager left, but I respected that it was a business decision by the hedge fund.
I had the chance to help design, build and trade a systematic model with real AUM and meaningful size behind it. I embraced it. It was one of the best coding experiences I've had.