Auto Trading in energy markets – more than just a set of algorithms
Auto-trading and algo-trading are the prevailing trends on the trading floors of public utilities and energy traders. Automated trading systems are known from the financial sector, where they have established mainly in high-frequency trading. Those methods and systems have now arrived in the energy industry and are spreading wherever near to real-time trading, liquidity and acceptable transaction costs meet – as in continuous intraday trading at EPEX SPOT.
No fear, no greed
The increasing market share of trading algos is easy to understand. Auto-traders can react faster to new information, they can efficiently manage many small positions and be a huge help, especially with time-critical strategies. Auto-traders know neither fear nor greed. They precisely execute any strategy, always respecting the limits. If this is considered a quality, they are superior to human decision makers.
In an ever more competitive environment energy traders may use algorithms to reduce costs of trading while at the same time improving their trading results – just to remain one step ahead of the competition. No matter if that’s just a few pips per MW. Public utilities and smaller traders of renewables who do not operate a 24/7 trading floor yet are currently evaluating carefully whether the investment for a 24/7 desk is necessary, or whether an intelligent machine could handle the unpopular night and weekend shifts. Another motivation – especially for public utilities searching for new business models – is the idea of offering market access and trading services to third parties.
There’s no crystal ball needed to realise that auto-trading will spread in the energy world. One might even ask why the transfer of established methods is not progressing faster. Is it the industry itself, which is considered traditional and risk-averse?
Though energy markets are converging with the financial models – fundamental differences remain. Ultimately, they depend on the physics of generation, storage and distribution. Resulting tasks, such as high-frequency position management for complex portfolios and the correct nomination of all trades to the transmission system operators, are all part of this business.
As a result, auto-trading in energy industry means more than just executing a couple of algorithms.
In intraday, traders must be able to rely on positions being shown in real time, precisely and correctly. Schedule management and nomination require great care and attention. Communication with many different players (stock exchanges, transmission system operators, customers, forecast or information providers) is essential. The auto-trading algorithm is only a part of a much bigger picture: the creation of an automated trading platform which integrates a range of internal and external systems, data and information sources. A glance at the image clearly spells out this challenge.
Concept of a trading platform
The key element of such a trading platform is a central position management that calculates the current position cyclically and whenever there is new information. It’s obvious that a wide range of data sources for forecasts, optimisation results or incoming customer orders need to be reliably connected – with secure communication protocols and powerful, monitored interfaces.
A reliable interface to the stock exchange is also essential – here, however, the EPEX SPOT itself provides quality assurance by subjecting communication via the M7 interface to certification. But also every deal at the stock exchange must be captured and transmitted to the position management, which recalculates the resulting position immediately. In this way, position management acts as the central hub where all flows of information are managed.
In the background, portfolio and book structures ensure that information stays transparent: For every deal it remains clear to which portfolio it belongs and which customer orders or deals have initiated it. This information is used to calculate the reimbursement of revenues to customer orders or, as applicable, to internal units (e.g. generation, sales).
Nomination secures trading results
Once intraday trading covers more than one control area, nomination becomes a critical success factor. Even the best deal is vain if it is not properly nominated to the TSOs.
For a short-term trading platform, this results in a further process chain that requires reliable automation. In the presented vision of a short term trading platform deals are transferred from central position management, validated, consolidated to schedules and dispatched. The replies from the transmission system operators are received and displayed. Performance and speed are equally important here, as nomination must keep up with the frequencies of trading.
Also required are automatic compensation rules for deals across control areas. Bidding, management and nomination of capacity rights for cross-border deals can be another important part of the story. No wonder, that trading and nomination are thus increasingly being seen as an entity. High-performing, reliable and automated nomination is the natural counterpart to the fast reacting trading algorithms.
But let’s have a closer look what we mean by auto- or algo-trading.
Auto-trading and algo-trading
Automated trading is currently going through two stages of evolution. At the first, we see auto-traders integrated in trading solutions as “black boxes”. At best, they can be configured to cover a wide range of strategies. Robust, proven and rapidly available, they represent a good way to get into automated intraday trading. The only drawback of auto-traders: they don’t offer a specific competitive advantage, as other market participants are also using the same or very similar mechanics.
A genuine competitive advantage can only be achieved with an algo-trader. This involves a “white box” – something like a scripting tool the trader can use to set up and execute complex, custom strategies. Those can be combined and continuously developed. Trading knowledge and algorithms remain the exclusive property of the company.
Before a self-developed algo-trader is released on the market, it’s advisable to carefully run backtests – using a suitable simulation environment and prevent any undesirable effects of the algo-trader. Additional benefit: market knowledge will grow and result in better understanding of situations in which a specific algorithm works well and others where another set up may be more appropriate.
Algo-traders offer maximum flexibility. They also require specific knowledge and resources for development and simulation. What does that mean for the traders? Most likely they will spend less time on deal execution or nomination. Instead more time will be available for analysis and strategy development. Instead of clicking deals, future traders will orchestrate the development and deployment of auto- and algo-traders.
Compared to the financial sector, where we are already familiar with the “race for algorithmic supremacy”, there are still low-hanging fruits in energy trading. It may represent already a considerable improvement for a small renewables trader when new forecasts are automatically and reliably closed at the stock exchange. A rather small set of algorithms may be sufficient for this case. Not enough to exceed the vwap (volume-weighted average price) but good enough to approach it. At least it’s better than paying balancing energy.
Next-generation trading platforms
Both auto-trading and algo-trading will rapidly grow in the energy sector. Automated trading in the energy industry, however, requires more than a set of algorithms. It needs trading platforms to integrate complex IT infrastructures and diverse communication processes and then monitor and keeping them running around the clock. These platforms can also be a basis for the upcoming, fully digitised business models in energy trading.