The Enduring Auction Capability (EAC) has been used for procuring response services for delivery since Friday November 3rd 2023. Now that the first set of results is available, we can see if any bidding patterns have emerged in these early stages.
The three main changes to the bidding process were:
- Co-optimization: Participants are able to submit multiple offers across the different response services in each EFA block.
- Splitting: Units can deliver multiple frequency response services simultaneously.
- Negative pricing: Sell orders can be priced below £0/MW/hour, which essentially means paying to deliver a service.
Participants are using the various new functions in the EAC
So far, 44% of units in each EFA block are submitting multiple baskets containing orders for different services that would clear through co-optimization. So far, slightly over 20% of units have submitted orders that would allow for splitting.
14% of baskets so far have included bids for high services at a negative price. Dynamic Regulation has seen the biggest impact from this, with the clearing price in the high service turning negative in 37 of the first 42 EFA blocks since the EAC launched.
As a result of the introduction of co-optimization, each service is now seeing more bid volume than ever before.
In the first seven days of the EAC, the average volume bid into Dynamic Moderation was 450 MW on average. This is almost three times what it was in the week before the launch of the EAC. In Dynamic Regulation, it was almost twice as much at 570 MW.
Bidding strategies in the EAC have varied significantly
The EAC has introduced some new potential complexity to bidding for frequency response due to the range of options when bidding. One of the simplest measures of complexity can be seen in the number of baskets submitted in each auction and the number of orders within each basket.
Orders are each individual item in a basket. For example, a parent order for Dynamic Containment High and a child order for Dynamic Containment Low within the same basket are two different orders.
High variation in basket quantities
There is a lot of variety in the number of baskets being submitted by each asset. Many are averaging six baskets per auction, which is essentially one basket per EFA block. However, multiple assets submit 24 baskets per auction, four baskets per EFA block. This brings them close to the maximum of 25 baskets per auction.
Participants will submit multiple baskets in each EFA block to co-optimize their bids. Participants can bid across several services.
On the other hand, participants will generally submit fewer baskets if they are looking to just participate in one service or if they have volume contracted in other markets during certain EFA blocks.
Low variation in order quantities
There is less variety in the number of orders being submitted within each basket. Most units have been submitting seven or fewer orders in each basket. Conrad Energy is one exception, submitting 13 orders per basket on average for its Winchester asset. Conrad Energy has been making use of splitting by submitting multiple child and substitutable child orders across multiple services - in a single basket.
In addition to splitting, the benefit of submitting additional child and substitutable child orders within a basket is that it allows participants to bid in different quantities for any services at different prices.
Complexity in the EAC
At a high level, optimizers submitting many baskets with many orders use a more complex bidding strategy. Assets submitting fewer baskets containing fewer orders are submitting simpler bids.
By this measure, Tesla run the most complex bidding strategy by submitting many baskets with many orders across their assets.
At the other end of the scale, operators such as SMS and Flexitricity are submitting fewer baskets with fewer orders across their assets.
But has a more complex bidding strategy actually resulted in increased revenues?
Complexity does not necessarily mean higher revenues
One measure of the complexity of bidding strategies in the market is the number of baskets submitted within each auction. While some of the best-performing systems are Tesla-operated, submitting 24 baskets a day, these are also all 2-hour systems. Meanwhile, some of the best-performing systems elsewhere are those submitting the minimum standard number of baskets per auction - six.
By contrast, the number of orders submitted per basket does show a stronger relationship to revenues (although there are exceptions). Many of the baskets for the 2-hour systems operated by Tesla contain orders for just one service (in both directions), in other words, either Dynamic Containment, Dynamic Moderation, or Dynamic Regulation. However, they have a high order count as these baskets also use child orders heavily. Each basket has several child orders for the service in both the low direction and high direction.
2-hour systems use a higher number of orders per basket than shorter-duration systems. They are mostly submitting multiple child orders for Dynamic Regulation High and Low. Shorter-duration systems mostly taking part in Dynamic Containment or Dynamic Moderation are using fewer orders.
However, those using multiple child orders have been able to earn higher revenues on average than those using a small number of orders.
A Participant submitting baskets that contain multiple child and substitutable child orders will do so in order to bid in additional quantities at different prices. This allows the algorithm to select as much volume that falls under the best price.
So far, this indicates that increased complexity in bidding strategies can result in higher revenues, but mostly by making use of the increased number of child orders that are now available rather than increasing the number of baskets. Meanwhile, there are clear exceptions to this.
One factor at play, alongside the new bidding functionality, is pricing strategy. So, how is this affecting revenues?
Acceptance rates in the EAC are driven by price
The proportion of volume accepted provides an insight into the pricing strategies being used. Those with the highest acceptance rates (even those at 100%) focus on a price-taking strategy, submitting bids at very low prices. These same assets are also those with the simplest bidding strategies regarding the number of orders submitted for each auction.
Meanwhile, assets that submit more complex bids are seeing lower acceptance rates. This is a result of more complex pricing structures (for example, pricing volume at increasing amounts through child orders), and also the option of mutually exclusive options for ESO to choose from through co-optimization.
Despite this variation in the market, there is no clear relationship between acceptance rate and revenues. Some systems with simpler strategies and high acceptance rates are performing well, but some of the systems with the most complex strategies and lower acceptance rates are too.
Overall, there is currently a lot more affecting revenues than pure bidding strategy. Optimization in other markets is also playing a key factor.
The future of the EAC
This is just the beginning of the EAC, and new or clearer patterns may develop over time as optimizers identify the best strategies. However, the huge range of options means that it may take a while to get there, if at all. One clear sign, however, is that revenues have fallen since the launch of the EAC regardless of strategy. This is a result of lower available prices in general, which the EAC was designed for.