What makes Short-Term European Power Market Surveillance So Difficult?
By CubeLogic’s Shane Henley, Global Head of Monitoring Solutions and
Kristof Salwiczek, Product Manager of CubeWatchTS
December 1, 2022
You hear this question a lot these days. Ask any Compliance manager responsible for European power market surveillance – they will tell you that there are unique features and complexities inherent in markets such as EPEX Spot and Nord Pool which makes the effective surveillance of these markets very different from standard futures markets such as those on EEX and ICE.
In recent years the significant increase in an intermittent renewable generation has driven a major shift toward short-term power trading. More recently, the European energy crisis has intensified this trend in part due to the skyrocketing cost of margining and collateral associated with forward trading.
From a compliance perspective, Regulators and market participants have struggled to effectively monitor these markets for abusive behaviour. While market participants are not mandated under REMIT to have a surveillance capability, many firms are making some attempts to do so with varying success. There are two key drivers behind this trend.
The first relates to the huge proportionate increase in algorithmic trading (“Algo trading”) in recent years in short-term power markets. When performing their market abuse risk assessments, many Compliance managers rightfully conclude that such activity represents a higher market abuse risk despite the absence of an RTS6-like regime that applies to algo trading in financial instruments in the EU. The second relates to the ongoing stresses in European power and gas markets stemming from the Ukrainian conflict which itself was preceded by the French nuclear generation crisis which started in 2021.
This trend is likely to gather pace in 2023 and 2024 as the Agency for the Cooperation of Energy Regulators (ACER) and National Regulatory Authorities (NRAs) respond to the pressure of being seen to be acting as retail and business customers increasingly experience the pain of rising costs.
So what are the things that make the surveillance of these markets so difficult? If like many Compliance officers, you are asked this question by senior management or IT, we hope that the summary below will help to provide you with some answers.
Single Intraday Coupling (SIDC) to Local Activity
The SIDC, formerly known as XBID, creates a single pan-European intraday power market. This market design aims to increase the overall efficiency of intraday trading by increasing liquidity and promoting competition, thereby enabling more efficient use of generation and capacity across Europe.
One of the features of this market is the rolling transition from the SIDC market phase, where orders from across European markets are consolidated into a shared order book (SoB), to the local market phase which typically starts in the last hour before physical delivery.
This transition is extremely important from a transaction surveillance perspective as the market dynamics shift precipitously during this transition, most observable through major changes in liquidity which may promote conditions that enable a trader to intentionally or inadvertently move a price in a particular direction.
Should the product in question, say quarter hour four in a given hour, for example, be considered the same product in both phases or should each phase be isolated and monitored independently of one another, or both? If treated as the same instrument in both phases, how are the technical challenges of linking the SIDC and Local products from a data perspective overcome?
Another consideration is the active and increasing role played by algos and how this might induce your traders to undertake potentially abusive behaviour in the context of the SIDC market design. Traders will acknowledge that algos generally become far more active, and perhaps aggressive, during the shoulder period between phases on both the SIDC and Local market side.
The ability to distinguish algo transactions from manual transactions is important for two reasons. Firstly, to support the ability to calibrate your surveillance tool separately for each type of trading approach. Secondly, you should be able to identify when multiple algos or manual trading strategies are interfering with one another in the same order book.
However, the ability to distinguish manual from algo activity presents several technical challenges. Since ACER does not mandate the reporting of algo IDs under REMIT (as is the requirement for financial transactions under MiFID reporting) most exchanges currently do not accommodate them within their data structures forcing market participants to devise workarounds that often involve the use of unreliable free text fields. Hopefully, over time and with pressure from the market these exchanges will see the value of accommodating this detail within their data structures.
SIDC Order Book Duplication
Another feature of the SIDC market design is that due to the shared order book concept individual orders are mirrored into multiple local order book views. This poses many questions and presents challenges from a transaction surveillance perspective.
How, for example, should the possibility of multiple similar (or identical) order books be handled from a data management and transaction surveillance perspective? Regarding the former, attempting to capture such duplicates/near-duplicates will significantly impact already sizeable data processing and storage costs given that the volume of orders may increase by a factor of ten or more. Does the cost-benefit dynamic justify such an additional cost?
Regarding the transaction surveillance approach, how should the local market view of the order book be monitored in relation to the SIDC/shared order book view? Are two independent but related sets of analysis warranted? If so, for which types of behaviours?
Again, there is no clear market consensus on this topic and limited guidance has been forthcoming from the regulatory authorities who may themselves be tackling the same challenges.
The Ephemeral Nature of Intraday Products
For transaction surveillance solutions to be effective they require transaction data history. This history enables the tool to compare and identify abnormalities across parameters such as trade and order volumes, counts, and prices for most standard surveillance approaches. This is not straightforward when it comes to intraday power where, for example, a quarter-hourly product is assigned a trade ID which is subsequently replaced with a new one for the next equivalent delivery period. Without product history there simply cannot be a transaction surveillance calculation.
At a fundamental level, an effective surveillance solution would be calibratable down to the granular physically delivering product level – but without the “historisation” of this activity, this would simply not be possible as there would be no lookback data available. Transaction surveillance solutions engineered originally for financial markets are often badly exposed when it comes to this aspect of short-term power market data handling.
Intraday Trading and Insider Trading Detection
Insider Trading, as envisaged under REMIT, is a complex topic and a challenging market abuse pattern to perfect from a transaction surveillance perspective. While many firms have implemented some capability to detect suspicious trading activity in the period leading up to and in between the market event and UMM publication, many do not apply this effectively to intraday markets which, in many cases, are the most severe and frequently impacted by such unplanned outage events.
Linked closely to the “historisation” topic outlined above, without the ability to logically classify and group intraday products to build a history, it is virtually impossible to map such products to the intraday delivery periods most relevant to the UMM driving the alert. Without the ability of the transaction surveillance solution to recognise which intraday tradable products are most likely impacted by a market event, the value of the Insider Trading alert as an effective identifier of suspicious behaviour is severely diminished.
SIDC Market Data Quantity and Completeness
Data challenges often arise from constraints in the data made available by the Exchanges themselves. For example, the EPEX Spot API does not make an aggressor flag available in the shared order book/SIDC phase of the market but rather only in the local phase. This presents a challenge for those wishing to perform effective surveillance of the order book – the aggressor status of an order provides important contextual information to the transaction surveillance analyst allowing them to analyse suspicious behaviour more effectively.
In the absence of such data, less-than-perfect workarounds are often necessitated. As the SIDC market infrastructure matures over time, the exchanges may work to make such information available as standard. This would certainly drive a far higher quality of market oversight from the market participants’ perspective.
There are many challenges associated with the effective surveillance of European intraday power markets. Some of these challenges emanate from unique market design features and others from the complexity, completeness, and quality of the available transaction data. It seems clear that most trading organisations and regulatory authorities are still getting to grips with how to effectively monitor these markets.
Market participants that are making little or no attempt at monitoring these markets due to the perceived difficulty and/or complexity of such surveillance are strongly advised to reconsider this pathway. Given the current European energy market crisis, it does not require much imagination to see a significant ramp-up in enforcement activity by ACER and the NRAs in 2023 and beyond. As ACER continues to accumulate and retain data, some may argue that much of what gets market participants into trouble might have already happened.