Traditionally acquiring data from exchanges has taken one of two routes – connecting to the exchanges directly or obtaining the data from a third party, such as Reuters. Many institutions have deployed both.

In recent years the technologies used to acquire the data has focused on the speed of delivery that is, acquiring the data as quickly as possible. This is especially important for algorithmic trading. In turn, this has put the emphasis on institutions connecting to the exchanges directly. However, more recently, the third parties are offering solutions that effectively connect to the exchanges directly.

This raises the question of which is the best solution.

AA INFO has worked with banks, suppliers and exchanges to develop the best solutions. This continues to included writing state of the art systems, developing infrastructure renew understanding the total costs of ownership and the trading market.

Many factors influence the speed of data through systems. This includes message size, message volumes, message rates, protocols and normalisation. The hardware, networks, software engineering and the relationship between them need to be understood. The business driver, including total costs of ownership and understanding the whole picture, is paramount.

A recent case study

A bank reengineers its direct exchange connectivity obtaining nanosecond latency through feed handlers. Further investment is made in DMA and RDMA techniques to increase speed by 30%. They now deliver data to the trading application quicker then anyone else. The trading application is moved to the new environment and becomes a bottleneck. Trading continues to be profitable. The application is fully reengineered to take advantage of the new environment at significant cost. On trading in dark liquidity pools, the new application is proven to be far ahead of the market. All the rights boxes have been ticked. Trading loses profitability. Analysis discovers that the application is so far ahead that it is trading against itself.