01 May 2008 Comments

 

Algo trading in energy markets

James Davies, Trayport

Currently the penetration into energy markets by hedge funds and algorithmic trading desks is far below that of other asset classes. Why is this?

Algorithmic trading has been with us for many years. The evolution from options pricing models that produced pricing sheets for pit trading through to the complex event processing systems from Apama, Streambase, Rulecore and others, has happened remarkably quickly. We have seen the introduction of commercial off-the-shelf spreading tools through application programming interfaces for vendor systems, exchanges and multilateral trading venues. In all marketplaces, algorithmic trading is highly significant, and in many it dominates volume. It appears that the more established a market and the more open it is to allowing access to any group of traders, the more likely algorithmic trading will play a significant role. This makes the fact that this is not the case in European over-the-counter energy markets a little odd.

Currently, when compared to other established marketplaces, algorithmic trading is underdeveloped in percentage volume terms in European OTC energy markets. At first glance this is somewhat surprising, as certainly a number of the markets are deep and liquid enough to be target markets for algorithmic trading. In particular, German, UK and French power markets, along with Dutch and UK gas, should be of interest to a would-be energy algorithmic trading group.

Many of the reasons for the lack of algorithmic trading in these asset classes are largely historical. Starting with the collapse of Enron in 2001, many of the smaller players, who traded through Enron for hedging and speculation, decided to exit the market. Post-Enron, the market was reliant upon utilities companies and the major investment banks.

Since Enron, it has been harder for these groups to return to the energy markets. There are dual barriers to effective algorithmic trading in energy, these being bilateral credit agreements and having the basic building blocks in place to begin algorithmic trading. The major issue with credit agreements is that there is no prime brokerage function taking place in the market. Therefore, every trading company needs an individual bilateral credit agreement with each potential counterparty. In general, for the market that exists at the moment, the bilateral credit risk is considered low as most of the counterparties are fairly sizable. However, a number of large utilities and banks would be suitably wary about signing too many credit agreements with lots of small players.

This will change at the end of 2010 when the Markets in Financial Instruments Directive capital requirements will come into effect across these groups in Europe. This means a number of players currently involved in the market will be effectively barred. Some hedge funds are already examining the possibility of attaining prime brokerage relationships for energy trading. For the prime brokers there will be a need to scale to make it a profitable business. It is likely this will then make it possible for many smaller players to enter the market in the near future.

The basic component issue is a little more abstract. In general, the proliferation of algorithmic trading has been achieved by creating basic building blocks, then applying these componentised rules and methods to each new asset class. This has been getting easier, with customisation of existing building blocks happening quicker, and the reuse of existing code increasing. However, this may not be as true for some of the energy markets. In these markets, it is common for limit orders to be marked “all-or-nothing”, presenting an added degree of complexity to some basic operations necessary to perform algorithmic trading. For many, this complexity is likely to be an attraction, as it will drive away some of the less capable and give a potentially longer life to algorithms.

As the above barriers come down, the involvement of many more participants will change the dynamic of the marketplace and bring about a large increase in the number of algorithmic trading participants and trade volumes. The willingness of current participants to innovate and experiment in order to take technologies and staff from other instruments has seen this market grow quickly in 2007. Given the continued uncertainty in credit markets and turmoil in equities, funds are looking even harder than normal for other sources of return. This combined with the continued deregulation of the energy markets in Western Europe and the increasing influence of carbon quotas means it must be expected that these markets will continue to build in size and sophistication.

James Davies, head of trader systems, Trayport