Study ERP topics in two passes: first as physical market mechanics, then as risk measurements built on those mechanics. Test yourself with scenarios where the physical reality (location, storage, volume flexibility) contradicts the financial abstraction, because that is where hedge design decisions actually live.
Why textbook risk metrics behave differently on energy prices
Energy spot prices are mean-reverting, jump-prone, and sometimes negative, so the distributional assumptions behind standard metrics need explicit checking rather than silent borrowing from equity or rate markets.
Mean reversion reflects a real mechanism: when prices spike, demand responds and supply is dispatched, pulling prices back toward production cost levels. Jumps reflect discrete events—outages, weather fronts, transmission constraints. Negative prices appear when inflexible generation must keep running. Each mechanism is a physical fact that shows up statistically as fat tails, autocorrelation, and bounded-but-violent behavior.
The risk implication is that you should be able to say, for any metric you use, what it assumes and where energy data violate that assumption. A metric estimated on calm periods will describe tail events poorly; a metric ignoring mean reversion will misjudge how long a spike lasts. Trace each mechanism to its statistical signature, then to the measurement choice it forces: heavier tail modeling, horizon-specific volatility, or scenario analysis alongside distribution-based measures.
- Mean reversion → spikes are typically short-lived but frequent; horizon choice changes the answer
- Jumps and outages → tail events are larger and more frequent than a normal distribution implies
- Negative or near-zero prices → volatility cannot be summarized by one number across price levels
- Regional segmentation → correlation between locations can collapse exactly when diversification is needed
Price risk, volume risk, and the spark spread: three concepts that must not blur
Price risk is exposure to the market price per unit; volume risk is exposure to how many units are actually bought or sold; spreads link a fuel input to an energy output through a conversion efficiency.
A heat rate expresses how much fuel energy is needed to produce one unit of electricity; the spark spread is the revenue margin implied by the power price minus the fuel cost at that efficiency. A dark spread is the analogous margin for a coal-fired unit. These are conversion concepts: they matter because a generator's exposure is not to power price alone but to the margin between two prices moving with different drivers.
Volume risk is categorically different from either. A power producer may hold a contract granting a buyer the option to take varying quantities daily—up to some maximum, with notice requirements. Even if the producer locks the power price with a forward, the quantity delivered is uncertain, so the hedge that is correct at full volume over-hedges at low volume and under-hedges at high volume. Distinguish, in every practice problem, which dimension—price, spread, or quantity—the question is actually about before selecting a tool.
Basis risk: why a hedge on the wrong delivery point leaks
Basis risk is the residual exposure created when the hedge instrument settles at a different location, grade, or time than the physical exposure, so the two prices can diverge even when both move.
Energy markets are physically segmented by pipelines, transmission lines, fuel grades, and storage. Two hubs in the same country can correlate at 0.95 in normal conditions and decouple when a line is congested or a unit trips. A forward sold at hub A does not offset exposure priced at hub B; it offsets the broad market move and leaves the differential—the basis—as a live position.
This is why hedge documentation should name the delivery point, index, grade, and settlement period of both the exposure and the instrument, and why a basis swap or a freight/transport cost adjustment is the standard patch when the points differ. In case-style questions, check the settlement terms on every instrument quoted. A mismatch there is not a rounding issue; it is a second, unhedged position hidden inside a supposedly flat book.
Scenario 1: the generator hedge that sat on the wrong hub
A gas-fired producer hedges its expected output with a forward at a liquid benchmark hub instead of its own pricing point, converting a price hedge into an unhedged basis position.
Setup: a plant prices its output at a regional index tied to a constrained zone, while the only liquid forward contract settles at a major benchmark hub. A plausible mistake is to sell forwards at the benchmark hub for the full expected volume and record the book as hedged. The better decision is to recognize the locational mismatch, size the hedge for the correlation actually observed between the two points, and consider a basis instrument or a transport-cost-linked contract for the remainder.
Why it matters: in a congestion event the benchmark price falls while the plant's local index falls further or behaves differently, so the hedge fails precisely when protection matters. Compare the choices in the table. The learning point is procedural: before evaluating any hedge economics, write down the settlement terms of the exposure and the instrument side by side, and treat any difference in location, index, or period as a separate, quantified risk requiring its own treatment.
| Decision element | Plausible choice | Better choice | Why it changes the outcome |
|---|---|---|---|
| Hedge location | Settle at the liquid benchmark hub | Match the plant's pricing index, or layer a basis trade | The differential becomes an explicit, monitored position instead of a hidden one |
| Hedge volume | Full expected output | Scale to observed correlation between the two points | Over-hedging a weakly correlated exposure adds risk in stress |
| Residual treatment | Assume basis is negligible | Document basis as a named exposure with a monitoring trigger | Congestion events are identified before they hit earnings |
Scenario 2: a swing option that made a hedged book risky
A trader hedges the notional of a flexible delivery contract at full volume, but the buyer's optionality over quantities means the effective position changes with exercise decisions.
Setup: a supplier holds a contract letting a utility buy up to a maximum daily quantity within a monthly cap, deciding day by day. The supplier locks the power price with forwards for the maximum volume. A plausible mistake is to treat this as fully hedged because notional matches notional. The better decision is to model the volume range and its correlation with price—typically the buyer exercises most when prices are high—then reduce the forward position accordingly and note the remaining volumetric exposure in the risk report.
Why it matters: when prices spike, the buyer takes maximum volume and the full forward offsets it; when prices fall, the buyer takes little and the supplier is left over-hedged, profiting on paper while the underlying business loses. The forward is the right instrument for the wrong quantity. The general lesson: when an instrument embeds optionality over quantity, the hedge must respond to exercise behavior, and the risk metric you report must describe the range of volumes, not just the maximum.
| Viewpoint | What it assumes | What it misses |
|---|---|---|
| Hedge at maximum notional | The buyer always takes full volume | Low-price periods leave the supplier over-hedged |
| Hedge at expected volume | Volumes average out | High-price periods leave the supplier under-hedged |
| Volume-aware position | Exercise correlates with price | Requires monitoring quantity behavior, not just price marks |
A daily mark-to-market journal with a self-check rubric
Build one small, ongoing exercise: track a fuel price and a power price for two weeks, compute daily returns and one spread, and write a one-line risk comment each day.
Exercise: pick any publicly quoted natural gas price and power price series. Each study day, record both prices, compute the daily percentage changes, and compute a spark spread assuming a stated heat rate of 7.5 MMBtu/MWh—so at $3.00/MMBtu fuel and $40/MWh power, the spread is 40 − (7.5 × 3.00) = $17.50/MWh. Write one sentence: what moved, and would your hedge have helped or hurt? This takes minutes but forces the physical-to-financial translation daily.
Rubric: score yourself 0–2 on four checks—a correct spread computation with units stated; an observation of whether the two prices moved together or diverged; a comment connecting the move to a mechanism (weather, outage, congestion) rather than just direction; and a hedge statement naming location, index, and period. A useful learning milestone is consistently scoring 6 or more out of 8 after two weeks, with every hedge statement passing the settlement-terms test.
An adaptable preparation sequence from market mechanics to decisions
Sequence study in four layers: physical market structure, pricing and derivative mechanics, risk measurement, then integrated case decisions—so each later topic rests on concepts already built.
Layer 1: map the physical system—generation types, fuel supply chains, transmission and pipeline constraints, storage—and name where each creates price behavior. Layer 2: derivatives on commodities—forwards, futures basis and convergence, options, and how conversion spreads (spark, dark, crush-style logic) frame exposures. Layer 3: risk measurement—volatility behavior on energy data, sensitivity measures, stress scenarios, and their documented assumptions. Layer 4: cases—hedge design, documentation, and professional standards applied to the first three layers.
Allocate review time so each layer feeds the next: after Layer 2, re-test every Layer 1 mechanism by asking what instrument would trade on it; after Layer 3, re-derive which assumptions break. Readiness checks before final review: you can compute and interpret a spark spread with units correct; you can state the settlement terms that matter on any quoted hedge; you can explain why a metric's assumptions fail on energy data; and you can write a two-paragraph case decision naming price, volume, and basis exposures separately. Treat these as learning milestones, not score predictions.
- Layer 1 → physical structure: fuels, grid, transport, storage; each tied to a price behavior
- Layer 2 → instruments: forwards, futures, options, conversion spreads
- Layer 3 → measurement: volatility and tail behavior, sensitivities, stress scenarios, assumptions
- Layer 4 → decisions: hedge design, documentation, ethics, integrated case reasoning
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.