Opportunities for arbitrage were analysed from custom built market data sources, trading on price discrepancies in different markets. This analysis then fed into a trading model that would participate in short trading windows, made available by inefficiencies in global markets. The trading model used a number of different factors to make fast trading decisions, and constantly monitored varying latency and jitter on different execution venues for optimum operation.
Trade loss rates of ~2% where achieved via highly informative price signals, that were quickly subsequently embedded in wider market prices. A number of proprietary techniques for optimising communications and data processing ensured excellent fill rates. Trade obfuscation was employed to minimise information leakage, and preventing predatory counter-measures.
Completely automated, the trading system coped with unexpected market movements - and failures to ensure fail-safe operation. The trading system included a custom-built dynamic UI to allow manual tuning of the algorithm in real-time.
When: 2008-2012
Model: Custom built
Client: Confidential
Technologies: C++, C#, F#, WPF, SQL
