Routing is important for markets that have a range of execution venues.
When choosing where to send orders, many different factor should be analyzed.
Routing mechanisms constantly make these decisions, based on the available data, to try to ensure orders are sent to the optimal destination.
As we saw for liquidity based trading algorithms, a common way of tackling the difficulty of trading in fragmented markets is to use liquidity aggregation. A virtual order book can be created by collecting data from all the possible execution venues.
Having decided this when we need to convert our requirements into actual orders. This transformation may be performed by the actual trading algorithm, alternatively, it can be delegated to a routing based execution tactic.
The virtual order book has been decomposed back into its constituents.
These are ordered by price, probability of execution and size. The hidden orders are only estimates, and so are placed after visible orders. Each entry is also flagged with the venue it originated from, whether is from the primary exchange an alternative trading system ( ATS) or an electronic crossing network (ECN).
When trading pairs of assets, the trading algorithm focuses on the trigger, whether the ratio or spread is favorable. Ensuring that each of the assets are successfully bought or sold can be delegated to an execution tactic.
Since execution cannot be guaranteed, legging it is an important consideration. If a reasonable amount of legging is permitted then each order can almost be worked separately. The execution tactic can choose to use standard mechanisms, much like any other static, price or liquidiy-based approach. Only when the legging reaches its limit will it need to try to intervene.
The relative liquidity of each asset is also important , if one asset is significantly less liquid than the other then this becomes a bottleneck. In fact, the orders for the least liquid asset tend to drive the whole process.