Financial markets are no stranger to algorithms, having long incorporated them within their daily operations. Now, they are also becoming a common tool on energy and commodity exchanges.

Roman Steden and Hendrik Vollrath set up an automated and systematic trading system at RWE Supply & Trading back in 2015. At the time, they were among the first traders on the EPEX Spot Exchange to buy and sell electricity continuously and around the clock with the help of their own programmes. Roman Steden, team lead of the Automated & Intraday Trading desk, and Hendrik Vollrath, senior algorithmic power trader, talk about the entrepreneurial spirit at RWE, who benefits from algo trading and why it will not be replacing humans any time soon.

In algo trading, computer programmes autonomously trade securities, foreign currencies or goods. RWE Supply & Trading uses algorithms to trade electricity on intraday markets. Does that leave you with anything to do with your day?

Roman Steden: The algorithms make trading easier, more secure and above all much more efficient. This allows us to systematically factor the versatility of our power plants into our hedging and trading strategies. It’s not as if there’s nothing for us to do. Every day, we work on improving the algorithms and adapting them to rapidly changing market conditions. Being a pioneer in intraday and automated power trading, the desk runs a holistic and complex algorithmic framework that continuously trades RWE’s flexible hydro and thermal assets plus batteries, outages, the sharply increasing renewables exposure as well as third-party flow. Saying that, we are responsible for managing a portfolio of over 20 GW installed capacity across various asset classes and play a key role in paving RWE’s energy transition and digitalisation path.

Hendrik Vollrath: Algorithms can process vast amounts of data at a speed that humans can only dream of. We know this from various other sectors, but in trading it is particularly useful because the data sets change in a matter of milliseconds. Even so, people have to keep on top of the software in order to fine-tune the programmes if necessary. We constantly analyse developments in the market and our own trading performance before using this information to come up with new ideas. We are always working together to develop new analysis tools and trading strategies, in order to upgrade the software.

That makes it sound as if the technology is still in its infancy. Isn’t that a cause for concern?

Hendrik: Well, from the word go, automation has actually not only improved our trading performance, it has also made our trading more secure in general. Algo-trading has been commonplace in the major financial trading centres of the world for many years now. And when, in late-2014, we finally got wind from our IT colleagues that this would now also be technically possible for power exchanges, we convinced our executives of just how promising this new technology could be in short-term trading. We have been in the business for ten years now and are undoubtedly one of the most experienced algo-trading companies in the energy sector. However, in today’s age of digitisation, particularly in efficient markets with highly professional market participants, it is all the more important to keep challenging our own approaches.

Roman: It’s probably not by chance that we both worked in risk management prior to joining the team. As a result, we are well-versed in assessing risks and also identifying ways in which to minimise risks with the help of algorithms. At the same time, we always look to identify possible inefficiencies and room for improvement and we use special mechanisms that continuously evaluate the algorithms for this purpose.

Was that the reasoning you used to convince your bosses?

Hendrik: It goes without saying that you have to show your bosses the potential of an innovative development in order to get them onboard. But at RWE (Supply & Trading) we have always fostered a strong entrepreneurial spirit – which extends from the management tier all the way down into the departments.

Roman: One factor was undoubtedly also that we have both been passionate about developing systematic and algorithmic trading strategies for some time now. That’s probably why we came together as a team and were able to tap into our pioneering spirit to bring management round to the idea. The leap of faith taken by our superiors was a real godsend for us. And, to this day, we still have maintained that sense of tenacity to face innovative processes head on, whilst keeping an eye on any potential risks. We are very much aware that not everything always works the first time around. But our environment is flexible enough to achieve solutions step by step.

A godsend for you and for RWE Supply & Trading. Because your business area, Automated & Intraday Trading, has now become well established. What’s next?

Roman: As I mentioned, we are continuously working on actively improving our system. We very much still have the tenacity to break the mould and learn from mistakes. That’s why we don’t simply jump on every high-tech start-up bandwagon that claims it can make a quick buck using new algorithms. We prefer to make informed decisions based on all the lessons we have learned. We are conscientious in our work with the venture capital that we have been entrusted with. We rely on our pragmatic approach, which – together with our IT department and software engineers – has thus far helped us create an ideal foundation, giving us a notable lead in the market. We are looking to maintain this lead and continue our expansion into new markets and products.

Speaking of “making mistakes”: What mistakes do algorithms make?

Hendrik: The algorithms are actually faultless. This is primarily because they have been programmed with security and stability very much in mind. More than anything else, they don’t make the same, manifold mistakes as us humans: from carelessly mixing up numbers to wrong decisions based on complex psychology. It’s simply not possible to avoid these issues when working with real-life traders, no matter their level of expertise. We even go one step further and work with sophisticated security systems that continuously monitor the decisions of the algorithms and subject them to a series of security checks. Only when these checks have been successfully completed are the algorithmic bids transmitted to the electricity exchange’s trading platform.

Let’s assume everything runs smoothly. Commodity trading is not an end in itself. You are paid to connect producers and customers so that energy gets up where it needs to be. To what extent can algorithms help you do an even better job?

Roman: Algorithms can react much faster to new market trends than humans can. This is more important now than ever before, because the balance of supply and demand is constantly shifting due to volatile renewable power sources, such as the wind and the sun. Algo-trading therefore presents a decisive advantage for all those involved, especially when it comes to short-term trading. On the one hand, of course, efficient markets offer good liquidity for trading. On the other hand, the actual optimisation has become much more ambitious than just a few years ago. Managing smaller but also larger volumes as well as being able to correctly quantify the constantly changing positions is what we do best. Our conventional power plants including hydro assets and batteries benefit significantly from this ability, as does the ever-growing market for renewable energy and third-party business.

Hendrik: We’re particularly proud of the innovative services we are able to offer third-party customers thanks to our increased level of automation. Due to the general trend towards algo-trading, the supply of liquidity on short-term markets is now considerable. We have developed a wide range of automated solutions for customers, which help them to access this liquidity easily and efficiently.