The pros and cons of algorithmic trading
Prepared by Tom Middleton of Citigroup
STRUCTURED ALGORITHMS
This class of algorithms is designed to trade as close as possible to a common trading benchmark.
VWAP (volume-weighted average price) This is probably the best-known and most commonly used trading benchmark. Over a given trading interval, the VWAP is the ratio of the total notional traded and the volume. VWAP algorithms use volume profiles to minimize variance, and limit order strategies to optimize performance.
TWAP (time-weighted average price) This is similar to VWAP, except that the volume profile is constant. For instance, while a VWAP algorithm might trade 1,500 shares in the next five-minute ‘bucket’, and 2,500 in the following five minutes, TWAP would trade two lots of 2,000 shares. TWAP tends to be used over short time intervals.
Implementation shortfall This comes in many different flavours. The aim is for the algorithm to manage the trade-off between impact (the cost of trading too fast), and opportunity (the cost of trading too slowly). Many strategies use an empirically derived impact cost model to estimate the optimal way in which to slice up the order.