Algorithmic Trading Usage Patterns and Their Costs

Abstract

Using algorithmic trading data across seven strategy types over 2009 and 2010, we examine usage patterns and performance for a sample of buy-side firms served by a multiplicity of brokers. Strategy usage is categorized by demand for liquidity, volatility, and concentration of orders traded. The data suggest employment of dominant strategies for the majority of firms, and shifts in strategy use are marginal across time and market conditions. In terms of performance, dominant strategies constitute a sensible approach at two ends of a spectrum: for easy orders and for situations that are extremely demanding in terms of liquidity and volatility. Performance matters, but does not distinguish individual strategy types in either regime. In all other circumstances, strategy shifts are possible and potentially profitable, given performance differences.