Search

The beauty of co-location

7 May 2014

By

At FIA EPTA, we think objections to co-location most often stem from not understanding what it is and how it benefits the market, or from the mistaken belief that co-location is some kind of ‘special club’ that only certain people get invited to join. 

Co-location is the practice whereby exchange members place the computers running their trading algorithms in the same data center as an exchange’s matching engine. While renting rack space at a co-location facility is a comparatively recent phenomenon, trying to get as close to the center of price discovery as possible (i.e. being co-located with it) is an age-old ambition. The difference now? – provided exchanges and MTF’s offer this service to all market participants on equal terms, co-location of servers makes markets fairer than they used to be.  

Let’s review history for a minute: it wasn’t all that long ago that trading took place exclusively in open outcry markets on physical exchange trading floors (or “pits”). This was just another way to get close to the center of price discovery, where the pits were the equivalent of today’s data centers, and the floor traders, in their color-coded jackets, were yesteryear’s trading servers. To be allowed to place a trader on the floor, you had to be a member of a special club by virtue of having bought a “seat” on the exchange (historically each member sat in an assigned seat to participate in buying and selling stocks). However, not everyone could buy a seat: the club was limited in membership, no matter how much money you were willing to pay to join. As a result, only a select few had the privilege of owning a seat on the exchange.

Remember how it used to work? On the floor, you may have been standing on the other side of the room from the broker with whom you wanted to trade. Trying to get your order placed was a matter of physical brawn: you had to shout louder than everyone else and literally elbow your way through to reach them. In regional markets, sometimes even that wasn’t enough. French floor brokers, for example, had a reputation for trading only with their friends; if you were a non-French trader on the floor, even if you offered a better price, you’d never get to deal.  The old concept of “best execution” was thoroughly flawed, and customers suffered as a result. Thanks to the development of automated trading technology, and the practice of server co-location, direct access to the exchange is no longer restricted to this special club. Anyone can pay for direct access to the exchange if they wish.

Other aspects of trading also suffered under the floor model. Price discovery was delayed or outright wrong: under a fast market circumstance, prices quoted were stale and fully out of sync with what was happening on the floor. Market abuse could thrive and was much harder to detect: the social dynamic of the floor lent itself to unfair practices, such as specialists who changed their prices when they saw on what side the broker wanted to deal, as buyer or seller. Trade confirmations carried enormous uncertainty: a mistrade would only come to light once the clearer failed to find the opposite trade in the next clearing slot. This could take an hour or more, leaving trades outstanding and both traders and customers at tremendous risk for long periods of time. Again, automated trading has significantly improved pricing efficiency, and the practice of co-location is a significant part of this. The speed at which market participants receive data directly reflects on their ability to create pricing efficiencies between markets. It doesn’t make economic sense for all trading/investing models to pay for this access, but that doesn’t prevent them from enjoying the benefits of efficient price discovery created by those market participants who do use the technology.

Even before pits and floors, location played a critical role in trading. In the origins of the stock markets, the so-called 'flagman' was an important communicator. The flagman was the person who occupied one of the many hills in New Jersey between the New York Stock Exchange and the Philadelphia Stock Exchange. His function was to signal orders and other information to the next flagman, and the next, until they finally accessed the exchange on the far other end. Records show it took about 19 minutes for a signal to make it through the line; at least 38 minutes to receive a confirmation of a transaction or partial fill.

In today’s world, it’s inconceivable for a trader to lose sight of an order for so long. Investors and professional traders get nervous if they don’t receive an immediate message back from the exchange acknowledging the order or confirming an execution. The techniques used by  the flagmen are now automated at the co-location site (through switches and GUIs and FPGAs). 

To come back to the present, the idea of co-location as we now know it was introduced to Europe by Chi-X in 2007. Chi-X was the first successful European multilateral trading facility (“MTF”) (a concept that was made possible by the competition provisions under the original Markets in Financial Instruments Directive (“MiFID”)). In so many ways, Chi-X changed the way things were done in the market; and made modern markets better than ever before. First, Chi-X invited all clients to place their computers at the location where the platform matched trades. It threw open access to its market data to all parties active on the platform. Finally, it broadcast its data instantaneously to all vendors. These changes meant the crowd space in the virtual “pit” was now unlimited; everyone could participate equally, and everyone could see data at the same time. 

Members struck their own deals with the most commercial data vendors and telecoms providers to lease lines into the co-location center from their offices around the world. There was a cost attached, yes. For an investor trading his personal portfolio twice a week, it would be considered expensive, but the prices were reasonable to professional brokers, banks and traders. The system had become a level playing field at last.

Co-location is not something that every trader needs.  If you're a retail investor with long-term investment goals, you probably don't need to invest in a spot at a co-lo facility. The important point for good modern market structure is that every market participant should be able to choose whether or not to make this investment: if this principle is upheld by all then co-location is one of the most transparent and fair aspects of modern market structure.

There is one aspect of co-location that could be improved, however.  After Chi-X introduced co-location in Europe, other exchanges opened co-location facilities as well.  In some cases these looked the same as Chi-X, but were significantly less transparent. Members were not invited to locate at the premise of the exchange's matching engine, but rather could take a space close to the address at which the exchange operated. This ushered in a real estate rat race, where firms fought to get as close to the co-lo facility as possible. In addition, some exchanges offered firms no choice of telecoms provider, so an easy price hike was created where firms had to pay five-fold the street price of data lines to co-locate with them. 

So yes, co-location does make markets fairer, but we can and should continue to seek improvements.  Access to co-location facilities should be regulated, transparent and provided fairly to all. 

The views expressed in this blog post are the personal opinions of the author and do not necessarily reflect the official policies or positions of the FIA European Principal Traders Association or the Futures Industry Association.

  • EPTA
  • Audio
  • Blog