Where imbalance comes from
A plant sells next-day output in the day-ahead market (GÖP) and adjusts it in the intraday market (GİP) close to delivery. When actual output differs from what was sold, the difference is imbalance. For wind and solar it is unavoidable: even the best forecast model cannot pin down exactly when a cloud bank passes or when the wind turns.
Imbalance creates cost in two ways. The first is the energy imbalance charge: surplus energy settles below the day-ahead price (PTF) and missing energy settles above it. The second is KÜPST, a charge for deviating beyond tolerance from the final generation schedule submitted to TEİAŞ.
Under-delivery → MAX(V, PTF, SMF) × (1 + kᵤ)
Over-delivery → MIN(PTF, SMF) × (1 − lᵤ)
In practice this means a plant faces expensive imbalance when it deviates in the same direction as the system and cheap imbalance when it deviates against it. When the system is long, SMF falls below PTF and a plant that over-produces effectively sells its surplus at SMF. When the system is short, SMF rises above PTF and a plant that under-produces covers its shortfall at SMF.
The problem for renewables is that their deviations usually point the same way as the system. On a sunnier-than-forecast noon, most solar plants in a region over-produce at once and push the system long. In 2025 the annual average SMF (about 2,526 TL/MWh) was below the average PTF (2,620 TL/MWh); the system was long for most of the year.
Since 1 January 2026 this asymmetry is also built into the coefficients. k and l, previously 3% in every hour, are now set per settlement period according to the system’s direction, and a party deviating with the system faces twice the coefficient. A floor price (V) was also introduced for negative imbalance in very low-price hours: under-delivery settles at no less than 150 TL/MWh even when PTF and SMF are below it.
Coefficients are set by EPDK Board decision and can be changed with advance notice. Calculations should use the current decision.
The change doubles the cost of deviating with the system and ties forecast quality directly to revenue. The value of portfolio netting rises by the same measure: when plants deviating with the system are offset by plants deviating against it, the more expensive coefficient never applies.
The portfolio effect: deviations cancel out
The core idea of aggregation is simple. Imbalance is calculated at portfolio level, not plant level. One plant’s surplus covers another’s shortfall, and only the portfolio’s net deviation goes to settlement.
Take two plants in the same hour. A solar plant sold 8 MWh but produced 6 MWh. A wind farm sold 5 MWh but produced 7 MWh. On their own they create two separate imbalances. In the same portfolio the net deviation is zero.
Assumptions: PTF 2,500 TL/MWh; SMF 2,000 TL/MWh when the system is long and 3,200 TL/MWh when short; 2026 coefficients (6% with the system, 3% against). Cost is the difference against settling the same energy at PTF. KÜPST not included.
One point stands out: in each scenario most of the cost falls on the plant deviating with the system. In a real portfolio deviations do not cancel this neatly. But as the portfolio grows and diversifies, the net deviation stays well below the sum of individual deviations.
What drives netting: correlation
The size of the portfolio effect depends less on the number of plants than on how similar their forecast errors are. Correlation measures this. Plants whose errors are independent net strongly. Plants under the same weather system err in the same direction.
The chart below shows the net imbalance of a portfolio of equal-sized plants as a share of the sum of individual imbalances. It uses a standard portfolio formula that assumes plants have similar error distributions.
Two results stand out. First, with correlation near zero a 20-plant portfolio removes more than three quarters of imbalance. Second, as correlation rises the benefit of adding plants falls quickly. At a correlation of 0.6 the remaining share never drops below 77%, however large the portfolio.
Ten solar plants in one region net less imbalance than five different assets in different regions.
A good portfolio is therefore built on diversity rather than size. Geographic spread, technology mix (solar, wind, hydro), consumption sites and storage are the main ways to lower correlation. Solar and wind forecast errors are weakly related in most hours, and consumption deviations are largely independent of generation deviations.
Balance responsible groups versus aggregation
Portfolio-level imbalance netting is not new in Türkiye. Balance responsible groups (DSG) have played a similar role for years. The Aggregation Regulation that took effect on 1 January 2025, followed by amendments to the Balancing and Settlement Regulation, drew a clear line between the two.
This limit is where the difference shows most. In a DSG, a plant’s full imbalance is netted within the group only if its imbalance ratio stays below a threshold, which fell from 10% to 5% on 1 January 2026. For renewables, hours above 5% are not rare. An aggregator portfolio has no such limit.
The aggregator’s responsibility widens accordingly. It cannot trade wholesale beyond balancing its portfolio, but it posts collateral and carries imbalance for every plant in it. Under amendments made at the end of 2025, a participant that moves all its licensed plants into an aggregator portfolio leaves its DSG, and if it is the group’s responsible party the group is dissolved. The industry also expects that the DSG structure may be phased out entirely.
How portfolio management works in practice
Netting happens automatically at settlement, but most of the portfolio’s value comes from management before delivery. The aggregator decides on the portfolio’s total position, not plant by plant.
The intraday benefit is often overlooked. For a standalone 5 MW plant, closing a deviation of a few hundred kilowatt hours in intraday may not be worth it given transaction costs and order book depth. At portfolio level those deviations add up or cancel out, and the remaining net position can be closed in one trade of a liquid size.
Storage in the portfolio is the last line of defence for what remains after netting. A battery can cover the portfolio’s remaining shortfall or surplus in real time and cut imbalance further. Because that capacity cannot be used for arbitrage or ancillary services in the same hour, the more valuable use has to be worked out hour by hour.
Who owns the netting benefit?
This is the question investors considering an aggregator ask least, and it matters most. Imbalance cost falls at portfolio level, but how are the savings shared between plants?
The method is a matter of negotiation and the most important clause in the contract. Under a fixed fee the risk moves to the aggregator, but its price is built into the fee. Under sharing models the plant’s own forecast quality feeds directly into revenue. For unlicensed and small plants, given the cost of building their own forecasting, a fixed fee is often a sensible starting point.
Questions to ask an aggregator
The investor’s view
Imbalance cost is often a bigger line than expected in the income statement of a merchant renewable project. Feasibility studies usually assume a fixed percentage; in operation it changes every year with forecast quality, system direction and coefficient rules.
Aggregation does not remove this cost, but it lowers both its level and its volatility. For financing the second matters as much as the first: a predictable imbalance cost fits far more easily into a lender’s cash flow model.
At Frekans we build our aggregation portfolio for diversity of technology and region, and manage forecasting and bidding at portfolio level. We share the allocation of the netting benefit and plant-level imbalance data openly with our investors.
