The Rise and Rise of Electronic Trading

An Electronic Communication Network (ECN) is a computer system that facilitates trading of financial products outside of stock exchanges. The primary products that are traded on ECNs are stocks and currencies.

ECNs came into existence in 1998 in the US when the SEC authorized their creation. There is a good entry in Wikipedia on Electronic Communication Networks which goes into this in more detail. What is interesting is that the introduction of ECNs and other classes of Alternative Trading System (ATS) has accelerated the adoption of electronic trading by the traditional exchanges. This has resulted in significant M&A activity with ECNs such as Archipelago and Instinet being acquired by the exchanges themselves.

Some analysts have gone as far as to say that at this point ECNs and exchanges are virtually indistinguishable. A good example is Celent.

The relentless rise in program trading (driven by the Hedge Fund industry in particular) has further increased the demand for automated electronic trading.

The Need to Virtualize Electronic Trading

Against this backdrop, creating an energy efficient computing infrastructure leveraging virtualization, provisioning and orchestration to virtualize electronic trading and deliver a consistently low latency high throughput non-stop quality of service i.e. achieve a specific set of performance service level agreements (SLAs) has become paramount, particularly as the many of the exchanges are seeking to reinvent themselves as service providers hosting third party regional exchanges as well as their own core exchange services. In Europe a good example is Atos Euronext Market Solutions.

The Challenge

However to create this kind of infrastructure and virtualize an electronic market requires an understanding of the way in which electronic trading works in practice. In any given market, each instrument (stock, option at a particular strike price, etc) traded electronically has its own order book. Each order book is maintained by a book manager which is responsible for implementing whatever trading rules are in place (inverting, crossing, matching, sweeping, etc) as well as the top of book (best bid/offer) and depth of book (active bids and offers in price/time priority). Furthermore orders or adjustments to orders sent by any trader (human or program) must always be handled in the order in which they are submitted by the trader and must never be lost or duplicated. This is called preserving partial ordering and is a fundamental requirement.

Therefore the moment you start to think about dynamically reallocating resources and migrating order books from one resource to another you have to ensure that you can guarantee partial ordering i.e. the partial ordering SLA cannot be violated under any circumstances.

This is a special case of the general problem of dynamically distributing a mediated service that can be logically partitioned into well-defined segments that applies to not just electronic trading but also to betting, online gaming and so on.

How to meet this challenge while at the same time taking full advantage of the benefits that virtualization and virtualization management brings to the table is something I'll be covering in more depth in my next posting.