I, Robot: Who will most benefit from the growing automation of energy trading?
Concerns about the impact of automated trading on futures exchanges stem from the fear that computers will run out of control, a concern reinforced by events such as the 'flash crash' of May 6, 2010.
However, the rules that make up automated trading algorithms are no more than the mathematical formulation of human perceptions about markets and price behavior. The most likely beneficiary in the long term are exchanges.
On March 3, Chief Executive of the CME Group Craig Donohue revealed that about 45% of all futures volume on the New York Mercantile Exchange, which deals primarily in energy commodities, was made up of proprietary electronic trading and that a further few percent comprised high frequency algorithmic trading.
Speaking at the Reuters Future Face of Finance Summit, he said that computer-driven trades in cash equity markets were often in excess of 70%.
Article continues below...
Request a free trial of: Energy Economist
Platts premier analytical newsletter, Energy Economist, combines incisive judgments and detailed data sets that deliver ideas you can profit by. This unique monthly publication presents succinct features that emphasize the cross-sectoral and long-term implications of current events and also includes regular updates from key energy policy centers around the globe.
The high level of electronic trading should be no surprise, but 'proprietary electronic trading' has been taken to mean automated trading, which represents different things to different people.
The move from open outcry to electronic trading platforms is automated trading, as is computerized order-matching internally by individual companies. These are both progressive and benign improvements that increase the efficient workings of markets.
However, in the context of Donohue's comments, 'proprietary electronic trading' appears to refer to trades generated by computers rather than humans, or at least that is how the comments were interpreted.
That automated trading has taken off in energy markets should also be no surprise, it is well entrenched in forex trading and equities.
There are a plethora of automated trading platforms available to day traders on the internet. Energy is in fact the latecomer.
The advantage is clear: computer power allows the rapid identification and implementation of multiple buy and sell possibilities.
If, in any market, a trader develops a successful strategy, one that might return 5% on turnover, for example, then the ability to enhance the number of trades executed means that turnover grows and so too do total returns.
The point-and-click human trader is physically limited in a way that a computer simply isn't.
Automated trading breeds
Automated trading comes in a number of different forms distinguished by the relative importance of the human and machine elements.
At its simplest, a trader uses a computer to identify and execute trades, but each trade is approved by the trader before execution.
A second form is arbitrage identification -- for example if bids and offers for a troika of currencies momentarily move out of line with each other creating a no-lose trade, the computer executes.
The computer can be left running 24 hours a day with minimal risk. If the conditions of the automated trade are not met, then the trades aren't made.
Third, a computer can be used by a human to execute a buy or sell strategy. If the trader wants to execute a large trade, he or she can instruct the computer to do so via lots of automated small trades to disguise from the market the size of the total order.
Fourth, there is what might be called 'true automated trading' in which the trader sets a more complex set of rules based on which the computer will execute buy and sell orders automatically.
This may be an algorithm based on relationships between different contracts. Or it may be simpler: a computer can be programmed to buy crude futures whenever the price rises above the 30-day average. Or to buy and sell when it identifies a certain movement in prices, which the trader believes indicates additional price movements to be likely.
Determining rules assumes that price behavior is to a degree predictable, i.e. that past observations about price behavior will be replicated in the future.
This might involve momentum trading on a single commodity or be based on more complex relationships between different commodities.
If these relationships change, as they almost certainly will -- remember these are economic relationships not Newtonian ones -- or prove erroneous, then the strategy will become redundant or start to make losses.
These should be contained by stop rules, allowing the trader to put their head back under the mathematical bonnet and revise the assumptions lying behind their commands.
The use of the phrase 'algorithm' adds to the mystification of such activities. Most of us understand calculus for a brief period between the ages of 15-17, but algorithms are a step beyond -- terminology that most people will not even try to understand.
Whoever uses one must be very clever indeed, and if they are using 'complex algorithms' they must have attained heights of intelligence to which mere mortals can only aspire, and more importantly cannot ask questions about.
In fact, an algorithm is just a set of rules that can be complex or otherwise. It is these rules that are the trading strategy.
Next story: Are the concerns over automatic trading overstated?