China's Spot Market Rules Let Solar and Storage Bid as One Entity

China's National Development and Reform Commission (NDRC) released draft Basic Rules for Electricity Spot Markets on 19 October 2026, and Article 8 does something no previous national rule has done: it lets a wind or solar plant and its co-located battery participate in spot trading as a single joint bidding entity — across bidding, clearing, dispatch, settlement and performance assessment. The rules would replace the 2023 interim framework, take effect on 1 January 2027, and run for five years. Public comment closes 18 November 2026.

Read it as a market-design story and it is a modest liberalisation. Read it as a control-systems story — which is the correct reading for anyone who models these assets — and it is a structural change in what a battery is for. Co-located storage stops being a separate revenue-seeking participant and becomes an instrument of the plant's delivery schedule, with the plant taking price risk on the combined position. That is a different optimisation problem, a different EMS, and a different degradation duty cycle.

What the draft actually says

Per pv magazine's coverage of the draft, Article 8 supports renewable facilities and their co-located energy storage participating jointly in market bidding, clearing, dispatch, settlement and performance assessment. The practical consequence is that the battery can be scheduled against the spot price signal inside one offer rather than as a second counterparty — shifting deliveries toward higher-priced periods where technically and commercially feasible, and soaking up deviations between forecast and actual renewable output.

The draft also broadens the definition of new market participants to include new-type energy storage, virtual power plants, load aggregators, smart microgrids and green electricity direct-connection projects. It sets day-ahead, intraday and real-time trading arrangements, with real-time markets scheduling dispatch over the next 5 to 15 minutes. Locational marginal pricing is named the preferred clearing approach, intended to reflect both supply-demand balance and transmission constraints. Settlement adopts daily clearing with monthly billing, moving toward price-difference settlement in regional and provincial markets. Frequency regulation, reserves and ramping are flagged for gradual joint clearing.

Critically, and as is common with central-government instruments in China, the national draft does not specify the bidding formats, metering arrangements, battery charging restrictions or settlement calculations for combined operations. Those will be set by provincial and regional implementation rules. That is where the returns will be determined — and where two projects in adjacent provinces may face materially different economics for identical hardware.

The reform did not arrive from nowhere. Shandong already operates joint bidding at provincial level, and the NDRC promoted the mechanism through 2025. The 2025 electricity pricing reform had already required renewable generation to enter the market with prices set by transaction. This draft takes a provincial practice national.

Why the settlement clause is the whole story

Most commentary on joint bidding has focused on the upside: the battery can arbitrage on the plant's account, so on-site storage gets a clearer path to commercial value. That is true and it is the headline. But joint bidding cuts both ways, and the second edge is sharper.

If the plant and battery settle jointly, then the plant's forecast error is no longer absorbed silently by the grid. It becomes a deviation the combined entity is financially accountable for. Storage participating in performance assessment means the battery is the plant's first line of defence against imbalance penalties — and the resource that pays for them when it cannot cover the gap.

Model it. Let the plant bid a quantity Q for a given interval, and let its realised generation deviate from schedule by a normally distributed error with standard deviation σ, expressed as a fraction of Q — a 1σ forecast error band of 5%, 10%, 20%. The battery can shift energy across intervals but cannot create it, so under joint settlement it can only fill the shortfall if it has energy and headroom available. In the limit where the battery holds no reserve for the deviation — the pure arbitrage case — the expected share of scheduled energy that is not delivered at the contracted price is the Gaussian shortfall expectation:

E[(x − σ)⁺] / E[x] = [φ(σ) − σ·(1 − Φ(σ))] / φ(0)

with φ and Φ the standard normal density and distribution. At zero error the penalty is zero by construction. At a 5% 1σ band, 6.1% of stored energy goes uncredited. At 10%, 12.0%. At 20% — entirely plausible for a day-ahead wind or PV schedule in a monsoonal or high-cloud regime — 23.1%. That is the number in Figure 1, and it is the reason a joint-bidding EMS cannot treat the battery's energy budget as free arbitrage capacity.

Change one assumption and the result changes: a battery sized with reserve headroom and dispatched by a forecast-aware controller claws back most of that loss. But it does so by holding energy back, which is exactly the behaviour that destroys the naive arbitrage revenue case. Joint bidding therefore converts a duration-sizing question ("how many hours?") into a co-optimisation question ("how many hours should be held in reserve against forecast error, at what price?").

What this does to the EMS

Under separate bidding, the battery's objective is straightforward arbitrage: charge at low prices, discharge at high prices, respect SOC limits, protect SOH. Under joint bidding with a co-located renewable plant, the objective function acquires three terms that did not previously interact.

1. The deviation term

The controller must decide, each interval, how much SOC to hold back as imbalance cover versus how much to commit to the price spread. That is a stochastic problem and it needs a forecast-error distribution, not a point forecast. An EMS built on a single deterministic PV or wind profile — the standard in most Chinese co-located projects today — has no way to price this term and will systematically over-commit energy to arbitrage in high-variance weather.

2. The forecast term

Coordinating generation forecast with the charge/discharge schedule inside one bid means the quality of the renewable forecast directly sets the battery's economics. A 3% better day-ahead PV forecast is worth more to a jointly bidding plant than to a separately bidding one, because it reduces both the deviation exposure and the reserve that must be withheld. Forecast improvement becomes a capex-justifiable line item rather than an operating nicety.

3. The degradation term

This one is under-appreciated. Imbalance-covering duty is not arbitrage duty. It is irregular, partly unpredictable, and frequently shallow — many small cycles in response to deviations rather than one deep daily cycle. Cycle counting by throughput (the standard first-order approach) will understate damage if the battery is doing high-frequency partial cycling, and overstate it if the deviation cover is largely absorbing energy that would have been curtailed anyway. The degradation model has to be evaluated at the actual C-rate and cycles-per-day the joint strategy produces, not at the nameplate arbitrage cycle.

The missing half: provincial rules will decide the returns

The national draft is a framework, not a settlement methodology. The elements it leaves to provincial implementation matter more than the elements it fixes:

The parallel with other markets is instructive. Europe's co-located and hybrid asset rules — and the UK's more recent treatment of hybrid connections — have run into exactly the same ambiguity on metering and on whether a battery may hold a capacity agreement while also arbitraging on the plant's account. The answers differ by jurisdiction and they dominate returns. China's provinces will follow the same pattern, and a developer building a portfolio across three provinces should assume three different settlement methodologies.

What it means for the co-located fleet

China has added storage at a pace that makes it the world's largest market by installed capacity, and a very large share of that capacity is co-located with renewables rather than standalone. For that fleet, Article 8 changes the operating mandate in three specific ways.

First, the battery stops being a merchant. A storage asset whose revenue was arbitrage against provincial spot prices is now, when co-located, an instrument of the plant's delivery performance. Its revenue is the plant's realised price, minus imbalance cost, and its value shows up as a system-level improvement in the plant's capture price rather than as a standalone P&L line. That is a cleaner story for a lender, provided the settlement mechanics are nailed down.

Second, duration gets re-sized. A four-hour battery sized for two daily arbitrage cycles is not necessarily right for a plant with a 15% day-ahead forecast error during the shoulder seasons. The reserve requirement is not a fixed number of hours; it is a function of the forecast-error distribution, the penalty curve, and the correlation between error and price. Higher price during the error — precisely the case when the grid needs the energy — makes the reserve more valuable and raises the optimal duration.

Third, dispatch strategy has to be evaluated as a strategy, not a rule. Rule-based dispatch with fixed SOC thresholds cannot price deviation risk because it has no distribution to reason about. Economic dispatch with a forecast-error term can. The choice between them is now a revenue decision worth modelling explicitly rather than a platform preference.

There is a quiet upside too. Joint bidding gives on-site storage a defensible commercial function beyond arbitrage — one that survives low price spreads, which China has. Provinces with heavily suppressed midday prices and thin evening peaks have made standalone arbitrage marginal; a battery earning its keep as forecast-error cover and imbalance avoidance has a value that does not depend on the spread. Whether that value is enough to cover capex is the open question the provincial rules will answer.

The design question to settle before January

Between now and 1 January 2027, developers with co-located projects have a finite window to re-model their portfolios against a settlement regime that does not yet exist in final form. The work that matters is not forecasting provincial rules — it is building models that can be re-run when they land. Concretely: an EMS that takes a forecast-error distribution as an input rather than a point forecast; a degradation model evaluated at the duty cycle the chosen strategy actually produces; and a financial layer that can be re-solved across netting-period and penalty-curve assumptions.

The projects that win will not be the ones that guessed the provincial rules correctly. They will be the ones that could re-run the optimisation in an afternoon when the rules were published.

For context on how degradation duty cycles translate into long-run value, see how SOC thresholds shape BESS operation and the comparison of rule-based and economic dispatch strategies. Both matter more under joint bidding than they did under separate participation.

Sources