Greek electricity retailers face stronger incentives to forecast customer consumption accurately from November 1, as demand prediction becomes a more significant commercial asset. The Greek energy regulator RAAEY confirmed that revised balancing-market provisions on significant, systematic demand deviations will take effect at the start of November, after transmission operator IPTO confirmed technical readiness. The provisions address persistent differences between the electricity demand market participants declare and the volumes their customers ultimately consume.
Rules target persistent forecasting errors
The rules are not intended to penalise every forecasting error: electricity demand cannot be predicted perfectly. Weather changes, industrial production varies and individual customers continually alter their behaviour. The regulatory focus is on significant and systematic deviations that show a portfolio is repeatedly being scheduled inaccurately.
Persistent errors can impose costs on the wider system. When a supplier consistently declares less demand than customers consume, the transmission operator must source additional electricity in the balancing market; when demand is repeatedly overestimated, generation may need to be reduced elsewhere. Either direction can increase balancing requirements, and from November poor forecasting will carry a more explicit financial consequence.
The change affects the economics of electricity retail, where suppliers have traditionally competed on prices, discounts, contract terms and customer acquisition. Forecasting quality now becomes another source of margin: two suppliers buying electricity at similar wholesale prices can have different profitability if one repeatedly incurs greater imbalance costs. The supplier with better forecasts has a structural advantage.
Data and software shape demand forecasts
The rules create a commercial opportunity for forecasting-as-a-service. Specialist software providers can combine historical meter data with weather forecasts, industrial schedules, calendar effects, customer characteristics and real-time consumption information to predict portfolio demand. Machine-learning systems can continually retrain their models as customer behaviour changes.
The technology is particularly relevant to suppliers whose portfolios are changing quickly. A retailer adding thousands of customers cannot assume they will behave like its existing customer base, and a residential portfolio dominated by electric heating differs from one dominated by conventional heating. Hotels respond strongly to tourism patterns, industrial consumers follow production schedules, and EV charging adds another increasingly variable load.
Forecasting is therefore becoming customer-specific, with smart-meter data able to materially improve accuracy. Traditional monthly meter readings offer little detail on when electricity was consumed, whereas interval meters show consumption by individual hours or quarter-hours. Suppliers can identify customers that create predictable peaks and those with volatile loads, using that information not only for billing but also to model their balancing exposure in greater detail.
The same data can support aggregation and demand response. If a supplier knows that a customer regularly uses 5 MW during a particular production period, it can forecast that demand more accurately; if some of the load is flexible, an aggregator may also be able to change consumption. Forecasting and flexibility thus begin to converge, allowing an advanced portfolio manager to correct an imbalance before delivery, rather than merely predict it.
Forecasts can prompt trading and flexibility actions
If a supplier expects demand to exceed its nominated position, it could buy electricity intraday or ask participating flexible customers to reduce consumption. If demand is expected to be lower, controllable loads such as EV charging or thermal systems could potentially be increased. The electricity portfolio then becomes actively managed, creating opportunities for companies that provide automated intraday trading and optimisation.
A modern retail platform could continuously forecast demand, compare it with existing market positions and execute corrective trades as delivery approaches. Human traders would increasingly supervise automated systems instead of manually calculating every adjustment.
In Greece, the shift coincides with a more complex retail market. Customers are increasingly exposed to fixed, variable and dynamic products, while distributed solar changes net consumption and smart meters generate more detailed data. Electric vehicles and behind-the-meter equipment add new load patterns, and data centres and other large consumers are introducing further concentrations of demand.
The result is a system that is harder to forecast using conventional historical averages. The cost of an inaccurate forecast also becomes more important as balancing markets grow more sophisticated.
Forecasting performance may affect suppliers and customers
The change could influence electricity-supply valuations, with customer count alone potentially becoming a less informative measure for investors. Two suppliers with 500,000 customers can have very different risk profiles depending on customer type, data quality, forecasting technology and balancing performance. A high-quality consumption dataset, like a proven forecasting model, could become an intangible asset.
The shift also has implications for industrial customers. Suppliers may increasingly reward companies that provide accurate operating schedules or grant access to real-time consumption data, since large customers with unpredictable demand create balancing costs for retailers. Customers with stable or transparent consumption profiles may become more commercially attractive, and contracts could eventually reflect that difference.
Energy-intensive customers may be offered better terms if they provide production forecasts, allow automated access to meter data or participate in mechanisms that help suppliers manage deviations. The boundary between electricity supply and energy management would narrow further.
Independent aggregators could also benefit. If Greece increases the financial consequences of systematic forecasting errors, suppliers have a stronger incentive to obtain flexible resources that can correct their portfolios. Aggregators controlling refrigeration, HVAC, EV charging, industrial processes or distributed generation can sell that capability, making demand forecasting a potential route into a flexibility market: first predict the deviation, then correct it.
The change will not eliminate imbalance costs, nor should it: the electricity system relies on balancing markets because forecasts will always contain errors. The aim is to discourage persistent behaviour that imposes avoidable costs on the system. This technical regulatory change also makes forecasting monetisable, as software that reduces systematic portfolio errors can directly protect supplier margins.
Smart-meter data becomes more valuable, automated trading more attractive and flexible customers gain another route to market. The Greek retailer’s competitive position may increasingly depend not only on buying electricity cheaply, but on predicting customer consumption, correcting forecasts before delivery and minimising the cost of every remaining error.