Björn Carlson

Björn Carlsson took his Ph.D. at Uppsala university in the area of constraint programming. He is now a Vice President of software architecture at NASDAQ OMX Group.

Next generation trading system

The next generation of trading systems are facing significant challenges, given the current quest for the lowest possible latency, and the aggressive growth of cores on a chip. When pairing these strong trends in the industry with the rise of complex instruments and trading strategies, you get an imposing set of conflicting requirements.

On the one hand, the simpler the system is the better, as this opens up opportunities to dramatically simplify the communication between the trading client and the trading engine, i.e. latency is lowered. On the other hand, the more versatile the system is the better, as this allows the exchange to offer more services and products without an explosion of surrounding systems and auxiliary interfaces. A state-of-the-art trading system should therefore efficiently scale with the hardware developments, while preserving rich functionality across a uniform set of protocols. The latter is also important to the overall fairness of the exchange as most members are presented with the same set of entry points into the marketplace.

Specifically, the following items are critical.

• scalability of the system. This includes scaling up, down and out, to make sure the system fully utilizes the cores and IO capacity of the network.
• versatility of the system. This includes supporting both simple (central limit order book of stocks) and complex (fixed-income, derivatives, commodities) trading models.
• cost and reliability. This includes architecting for simplicity and robustness, while adhering to the above principles, and remaining vendor-neutral.