NEMSEC common mistakes
NEMSEC · Understand the traps and how to avoid them.
Treating locational marginal price (LMP) as a single system-wide energy price and ignoring that it decomposes into energy, congestion, and loss components, then misattributing a price spike to generation scarcity when it is actually binding transmission congestion.
Candidates learn the merit-order concept first and mentally collapse LMP to a marginal cost curve. The three-component decomposition and the role of shift factors in congestion pricing are abstract, so under time pressure they default to the simpler mental model.
What to do instead
Work through the LMP decomposition formula explicitly for at least ten worked examples, including cases where the system is unconstrained (congestion and loss components zero) and where one line binds. For every price-spike scenario, write out which component moved and name the binding constraint before concluding anything about scarcity.
Confusing real-time and day-ahead settlement by applying day-ahead prices to actual delivered volumes, or forgetting that the real-time deviation is settled at real-time prices while the day-ahead position settles at day-ahead prices.
The two-settlement system is conceptually simple but the arithmetic is easy to invert, especially when a question describes a generator that was scheduled in day-ahead but deviated. Candidates conflate 'scheduled' with 'delivered' and lose track of which volume gets which price.
What to do instead
Build a settlement worksheet for every practice scenario: columns for day-ahead schedule, day-ahead price, real-time deviation, real-time price, and total revenue. Compute the total as day-ahead volume times day-ahead price plus deviation times real-time price. Practise until the worksheet is automatic before attempting narrative questions.
Answering market-power questions by identifying high market share alone, without applying the deliverable-capacity or pivotal-supplier test that actually establishes the ability to raise prices profitably.
Candidates import a generic antitrust reflex (share equals power) that does not fit electricity markets, where transmission constraints and inelastic demand make a small supplier pivotal. They also skip the profitability step (the residual demand curve must be steep enough).
What to do instead
For any market-power question, run a fixed checklist: define the relevant geographic market using transmission limits, compute residual demand faced by the supplier, test whether the supplier is pivotal in any hour, and then test whether withholding raises profit. State each step explicitly in your reasoning.
Classifying every unusual trading pattern as manipulation, rather than distinguishing legitimate economic withholding, normal portfolio optimisation, and bona fide arbitrage from conduct that meets the legal elements of manipulation (intent plus artificial price effect).
Surveillance scenarios are written to look suspicious, and candidates pattern-match on 'unusual' rather than on the legal test. They also underestimate how often rational bidding produces outcomes that look anomalous.
What to do instead
Learn the statutory elements of manipulation and apply a three-part filter to every scenario: was there an artificial price or misleading signal, was there intent or reckless disregard, and would a competitive benchmark explain the conduct? Write the benchmark explanation first, then test whether the facts defeat it.
Treating value-at-risk as a complete risk measure and ignoring tail risk, liquidity risk, and the mismatch between VaR horizon and the actual time needed to unwind a physical or illiquid position.
VaR is the most memorised metric, so candidates reach for it as a universal answer. The assumptions (normal returns, liquid markets, static portfolio) are rarely stated in exam stems, and candidates do not question them.
What to do instead
For every risk question, state the VaR assumptions explicitly and then name at least one limitation relevant to the scenario (fat tails, illiquidity, correlation breakdown in stress). Pair VaR with expected shortfall and a stress-test result in any recommendation.
Assuming that compliance with a market rule is satisfied by filing the required report, without checking the substantive accuracy, timeliness, and completeness obligations that attach to the filing.
Candidates treat compliance as a documentation exercise. Real enforcement actions frequently turn on late, incomplete, or misleading submissions rather than on the underlying conduct, and exam stems exploit this gap.
What to do instead
For each regulatory obligation, memorise the four dimensions separately: who must file, what must be filed, by when, and with what accuracy standard. In scenarios, audit the filing against all four and flag any dimension that fails, even if the report was submitted.
Applying a generic cybersecurity control framework to energy-market systems without mapping to the operational-technology and market-participant interfaces (bid submission channels, settlement data feeds, ICCP links) that carry the actual risk.
Candidates study IT security concepts and assume they transfer unchanged. Energy markets blend IT and OT with strict availability requirements, and the exam tests whether you can localise controls to the market data path.
What to do instead
Draw the data path from participant bidding system through the market operator to settlement, and label each interface with its threat model and required control (authentication, integrity check, segregation of duties, logging). Answer security questions by naming the specific interface and control.
Confusing the roles of the market operator, the regulator, and the surveillance unit, and attributing enforcement powers to the wrong body.
Governance structures differ across jurisdictions, and candidates generalise from one model. Exam questions often describe a referral or investigation and ask which entity has authority at each stage.
What to do instead
Build a one-page table of the bodies in scope with columns for rule-making, monitoring, investigation, and sanctioning authority. For each practice question, identify the stage of the process before naming the body.
Scheduling a unit or portfolio without accounting for ramp limits, minimum up and down times, and start-up costs, then reporting an infeasible or uneconomic schedule as optimal.
Candidates optimise on energy price alone because the arithmetic is tractable. Physical constraints are listed in the stem but not integrated into the calculation under time pressure.
What to do instead
Before computing any schedule, list every constraint mentioned in the stem and check the proposed schedule against each one in sequence: ramp, minimum run times, start-up cost versus margin. Reject any schedule that violates a constraint before comparing economics.
Ignoring the treatment of losses and the reference-bus assumption when reasoning about congestion and price separation, so that a line loading is misjudged or a price difference is attributed to the wrong cause.
DC power flow with a slack bus is a modelling abstraction that candidates learn procedurally but rarely interrogate. Losses and the slack bus are usually invisible in simple examples, so their effect surfaces only in harder questions.
What to do instead
Practise small three-bus examples by hand, varying the slack bus and adding loss factors, and observe how flows and prices change. When a question involves price separation, state whether congestion, losses, or both explain it before selecting an answer.