The official explanation of the Basic Rules for Electricity Market Operations emphasises common foundational institutions and an operating framework for the market.[1] The explanation of renewable electricity pricing reform further describes market pricing and associated settlement arrangements.[3] Companies therefore need to understand production capability, contract structure and cash flow together. Actual participation still depends on the relevant market and applicable rules; a national policy direction is not an operating manual for an individual trading product.

ZHICE sees the trading team's role as more than anticipating prices. It must make executable commercial choices with incomplete information, limited time and physical constraints. Generators need to understand available output, maintenance and fuel constraints; retailers and energy users need to understand customer demand, load characteristics and contractual obligations. The risks differ by business type, but the common problem is getting timely and complete business facts into the decision.

When forecasts remain inside technical systems, production plans in dispatch conversations, contracts in documents and trade results in post-transaction summaries, the organisation struggles to explain profit and variance. The objective should shift from adding a trading platform to establishing one shared operating record. That record identifies the effective version, applicable period, owner and known limitations, enabling departments to discuss the same exposure within the same time boundary.

Identify four capabilities by decision responsibility: commercial management, independent risk control, operations and settlement, and production coordination. This does not require every company to create four separate departments. Smaller teams may combine activities, but critical checking and approval should not all rest with one person. The proposal owner explains assumptions and boundaries; the risk role independently checks constraints; the authorised decision-maker accepts residual risk; and settlement verifies results against the final rules and records.

Production coordination should not be a notification sent after trade execution. Changes in maintenance, equipment status, fuel or demand affect the ability to perform obligations. Specify which changes must reach trading, when notification is required and how receipt is confirmed, with a conservative response when confirmation cannot be obtained in time. Trading must likewise communicate contractual commitments to production and business management so internal plans are not changed without regard to external obligations.

Role procedures should explicitly address exceptions: unavailable systems, absent key personnel, approaching market deadlines, large forecast deviations and temporary rule changes. Naming deputies, pause conditions and escalation routes in advance is more useful than asking everyone to 'coordinate better' after an incident. Exception approvals need a duration and scope; a one-off emergency authorisation must not permanently expand a role's powers. Complete records and a review are still required after normal operation resumes.

Group operations also need to define the relationship between legal-entity responsibility and centralised management. Central teams can share forecasting, analysis and operating support, but contracts, authority and settlement obligations of different entities must remain distinct. Establish which exposures can be observed in aggregate, which require separate management and whether internal coordination creates additional obligations. Preserving entity, regional and product boundaries prevents an apparently balanced group dashboard from concealing an individual entity's inability to perform.

Daily and periodic work can follow the same loop: gather business constraints, produce explainable forecasts, construct alternatives, obtain authority and execute, then reconcile delivery and settlement. A forecast does not issue a trading instruction; it is one decision input. Comparing alternatives also requires existing contracts, available flexibility, business objectives and risk limits, including a response if the forecast proves wrong. This prevents a model output from being mistaken for a completed business judgement.

Freeze the information available for each formal decision, including forecast time, data scope, production constraints, market information and applicable rules. Freezing does not prevent updates; it makes later adjustments traceable. When new information arrives, record which assumption changed, why the plan required adjustment and how that change was approved. Otherwise, review judges an earlier decision using later knowledge, offering little help for model improvement or fair assessment of roles.

The operating cadence should match the actual deadlines of the market concerned rather than copy another region's practice. Work backwards from the deadline to data confirmation, proposal submission, risk checks and authorisation, allowing appropriate buffers. Handovers should include unresolved exceptions and next actions, not just results. Across shifts, regions or legal entities, identify who has final interpretive authority so one transaction does not acquire two effective versions in different teams.

Forecasts should communicate their conditions of use and principal uncertainties, not only a number. Business staff need to know whether inputs are complete, whether structural changes have occurred and which periods need additional human judgement. Forecasts materially beyond the range of historical use should trigger further review. Record which assumptions support decisions under different scenarios, then test them against outcomes. This connects model improvement to commercial choices more effectively than one aggregate retrospective error measure.

Risk policy works only when it translates into concrete action. Separate limits into business boundaries, outstanding exposure, operating behaviour and exception triggers, defining the object, period and calculation for each. Contract volumes, forecast generation or demand and adjustable capacity need consistent time periods before uncovered exposure can be assessed. Risks differ across entities and products. This report proposes no universal numeric limits; companies should approve them according to their own capacity to bear risk.

Limit breaches should not rely on retrospective email alerts alone. Systems and processes should distinguish a warning, a prohibition and a permitted exception. A warning calls for explanation or increased monitoring; a prohibition stops the relevant action; an exception requires an authorised person to approve the reason, duration and remedy. For exposure already created and not immediately reversible, specify a feasible reduction path and track its completion. Acknowledgement of a risk is not closure.

Authority design should also address incentive conflicts. If traders choose assumptions, set risk thresholds and explain settlement variances themselves, independent checking is absent. Changes to critical parameters, permissions and manual forecast overrides need searchable records. Management can sample actual transaction chains to test whether decision evidence is recoverable, contemporaneous authority was respected and exceptions were genuinely resolved, rather than checking only whether policy documents exist.

The official explanation of the Basic Rules for Electricity Market Information Disclosure emphasises truthfulness, accuracy, completeness, timeliness and usability, while distinguishing disclosure scopes.[2] Internal data governance should connect with these requirements. Information used externally needs a source and review owner; information used internally needs a defined permission scope. Visibility to one role does not make the data available for every business purpose, model-training activity or external communication.

Create a common business dictionary for forecasts, contracts, metering, market notices and settlement, paying particular attention to time zones, trading intervals, units, positive and negative directions and version updates. Apparent model weaknesses often arise from misaligned time slices, late data or mixed revision versions. The data platform should retain original records and correction relationships so any business result can identify the version used and when that version became available.

Rules also require version control. Map rule provisions to business processes, parameters and reports, identifying effective dates, applicable entities and the owner of interpretation. After a new rule is issued, assess its scope of impact, change and validate parameters, train relevant staff and confirm the switchover. For disputed provisions, retain the verification process and interim treatment. An unconfirmed oral interpretation should not become an embedded system rule affecting trading and settlement indefinitely.

Separate expected settlement, actual settlement and cash receipt. The business team's rule-based estimate, the market operator's settlement, accounting receivables and payables, and cash flows may relate to different processing dates. Identify differences arising from data revisions, metering, rule interpretation, contract terms or timing. Assign who reconciles each item, who raises a dispute and when the matter should be closed.

Operating attribution should avoid treating one total return as a measure of trading ability. Price effects may be external; volume differences may arise from production or customer demand; execution differences may reflect operating actions; forecast errors may come from data or models. Establish a repeatable comparison baseline and the calculation order for different factors to avoid counting the same contribution twice. Effects that cannot yet be separated should remain explicit unexplained items, rather than disappear into a residual measure.

Performance assessment needs both process and outcomes. An authorised, well-evidenced decision can have an adverse result, while an unauthorised action may make an accidental profit. Reward sustainable judgement and cooperation rather than short-term outcomes that weaken risk discipline. Include recurring errors, exception-closure quality, forecasting improvements and cross-role responsiveness, calibrating standards through specific cases so review becomes a tool for improving operations.

Dispute handling also needs a complete chain. Retain original evidence, calculations, communications and applicable deadlines for metering or settlement discrepancies, with a named role responsible for raising the issue. Internal confirmation does not establish external agreement. Reports should distinguish pending verification, submitted, confirmed and resolved states. Finance and the business should use data at the same status, reducing misunderstandings between performance review and cash planning and drawing attention to long-standing open items.

Stress scenarios should combine changes that the organisation finds difficult, rather than merely raise or lower a price curve. Examples include a critical equipment fault near a trading deadline, missing forecast inputs alongside changed customer demand, or a settlement-definition change during a system upgrade. These are hypothetical exercises, not descriptions of actual events. The test is whether information reaches the right role, authority is obtained in time and the company can make an executable decision within its boundaries.

For each scenario, identify triggering facts, affected responsibilities, required data, the first action and recovery conditions. An exercise should reveal and correct procedural gaps, not expect staff to fill them through personal experience. In particular, test whether pausing trading or reducing exposure creates new performance obligations, whether backup systems preserve consistent versions and whether independent checking survives the absence of key people. The endpoint is a return to controlled operations, not the production of an incident report.

Examine external dependencies as well: data services, trading tools, forecast suppliers and system support. Agree in advance what substitute inputs are available during an interruption, whether data can be exported, whether historical records are complete and whether supplier parameter changes require revalidation. Exit and recovery capabilities may be invisible on a normal trading day, but determine whether the organisation can keep working during exceptions. Include them in procurement and operational acceptance.

Select a clearly bounded business and complete one cycle from forecasting through settlement. Map actual work, recording handovers, information delays and repeated entry, then establish common definitions and an authority register. Retain necessary manual checks while testing whether responsibilities are clear and results explainable. Real problems from this pilot provide a stronger basis for development priorities than designing a large platform for every market and product at the outset.

The next stage translates stable processes into system capabilities: version retention, constraint checks, exception approval, settlement reconciliation and attribution. Before expanding to a new region or trading product, assess rule differences and organisational readiness, then decide which components are reusable. A system feature does not establish the company's ability to manage its risks. Business authority, staff readiness and contingency exercises should all be prerequisites for expansion.

Management should ultimately accept three capabilities rather than a number of reports: can roles explain the same transaction consistently, can exceptions receive timely authorised decisions, and can outcomes produce explicit improvement tasks? ZHICE recommends retaining only review actions with an owner and completion conditions, and checking their effect in the next cycle. Once this mechanism operates consistently, power trading becomes a sustainable organisational capability rather than expertise held by a few individuals.

Learning after expansion should retain failures. Periodically select a few representative decisions, reconstruct the information and alternatives available, and discuss which judgements are reusable and which are context-specific. Avoid treating correlation as causation or widening risk-taking after one favourable outcome. Adjust strategy or authority only after accumulating enough evidence under different conditions. Matters still in dispute should remain under observation, with the next facts needed for judgement explicitly stated.