Spark Spreads and Capacity Revenue
How merchant generators earn from spark spreads (hub energy minus gas cost) and PJM capacity revenue, using planning marks of $50/MWh and $250/MW-day.
Educational analysis for professional use. This guide is not investment advice or a recommendation to buy or sell any security, and it is not personalised.
Merchant Revenue Splits Into Energy and Capacity
Contracted yieldcos price off PPAs. Merchant generators price off two wholesale legs: energy (spark spread) and capacity (availability payments). Vistra’s majority-merchant fleet and NRG’s integrated Texas retail model both depend on how those legs clear each period, unlike a contracted yieldco with a fixed PPA price.
The contracted premium guide maps the multiple gap. This page covers the merchant revenue build.
Spark Spread: Energy Minus Gas Cost
The spark spread answers: after buying gas, does the plant earn more selling power into the hub?
Spark spread ≈ wholesale energy ($/MWh) − Henry Hub gas ($/MMBtu) × heat rate (MMBtu/MWh)
Henry Hub planning is $3.00/MMBtu (2025 actual ~$3.52). PJM wholesale energy planning is $50/MWh. PJM 2025 realised load-weighted average LMP was $50.73/MWh.
Worked examples in this guide use demonstration prices; check current market prices when modelling.
Heat rate is plant-specific and not filed uniformly across the comp set. The concept matters more than a single benchmark number: when gas rises at fixed output, the spread narrows; when hub energy rises at fixed gas, it widens. Merchant EBITDA moves with that margin. Contracted PPAs insulate the yieldco side of the sector from the same swing.
Capacity Revenue: $/MW-Day Times Cleared MW
Capacity auctions pay plants to be available, independent of dispatch. The revenue build:
Annual capacity revenue = cleared MW × $/MW-day × 365
| Mark | Value |
|---|---|
| PJM planning capacity (BRA) | $250/MW-day |
Planning uses $250/MW-day. Cap-bound auction years can clear well above that norm; those clears are not a planning base case.
Worked Example: 1 GW PJM CCGT at Planning Marks
Using planning inputs on 1,000 MW cleared capacity with 58% combined-cycle gas turbine (CCGT) capacity factor (planning norm):
| Revenue leg | Calculation | Annual $M |
|---|---|---|
| Capacity | 1,000 MW × $250/MW-day × 365 ÷ 1,000,000 | ~$91.3M |
| Energy (MWh) | 1,000 × 0.58 × 8,760 | 5,080,800 MWh |
| Energy revenue | 5,080,800 × $50/MWh ÷ 1,000,000 | ~$254.0M |
| Total (energy + capacity) | ~$345.3M |
Capacity is ~26% of this simplified revenue stack at planning marks. Energy dominates because 58% capacity factor on 1,000 MW produces ~5.1 TWh at $50/MWh.
Cap-bound auction years can inflate capacity revenue materially versus the $250/MW-day planning norm on the same cleared megawatts. Models should not anchor on auction clears; use planning capacity for through-cycle work.
How the Comp Set Maps to Merchant Revenue
| Company | Merchant exposure proxy | Fleet context |
|---|---|---|
| VST | Majority merchant without LT PPAs; forward ~half 2026E EBITDA from Retail + LT PPAs | 43.6 GW; gas 62%, nuclear 15%, coal 20% |
| NRG | Integrated retail + merchant; no filed PPA% | 12.3 GW pre-LS Power; coal 52%, gas 43% |
Vistra’s 6,448 MW nuclear includes ~3.8 GW under long-term PPAs (AWS, Meta). That slice behaves like contracted baseload, not spark spread. NRG’s retail margin dampens wholesale volatility but is not a PPA. Both add contracted-like cash flow that a pure spot-merchant plant does not have.
The illustrative merchant worked example (5.0 GW, 70% merchant energy, PJM $50/MWh plus $250/MW-day capacity, $1.2B Adjusted EBITDA, 6.5× EV/EBITDA anchor) packages these legs into a single valuation template.
Capacity Factor Links the Two Legs
Energy revenue scales with output; capacity revenue scales with cleared MW regardless of dispatch. A CCGT at 58% capacity factor (planning norm) generates far more MWh than a solar site at 24% on the same nameplate. The capacity factor guide covers the LTA math. For merchant screening, CCGT norms feed the energy leg; cleared MW feeds the capacity leg.
What Matters Most
Merchant generators live on spark spreads and capacity auctions. Planning marks: PJM energy $50/MWh, capacity $250/MW-day, Henry Hub $3.00/MMBtu. Build energy revenue from MW × capacity factor × 8,760 × $/MWh; add capacity from MW × $/MW-day × 365. Contracted yieldcos skip this stack entirely, which is the main reason they trade +3-5 turns higher on EV/EBITDA.
Renewables & IPPs Sector Primer
A multi-period DCF splits contracted PPA years from the merchant tail, pricing each off realised power price and term, then checks the value against spark spread and CAFD.
The Excel model is the primer's two archetype builds live across 10 sheets: a multi-period DCF with a contracted-or-merchant toggle, step-down discount rates for the PPA term and the merchant tail, a spark-spread build for the merchant case and a CAFD-and-payout bridge for the yieldco. Change the realised PPA price, the contracted discount rate or the merchant spread and the value per share moves; the premium and payout sheets update alongside it. It also splits the contracted premium into the part the pipeline explains and the part contract quality explains, splits the PV between contracted years and the merchant tail, and reads EV per kW on both archetypes.
Frequently Asked Questions
- What is a spark spread in power markets?
- The spark spread is the margin between wholesale electricity revenue and the gas cost of generating that power. Conceptually: hub energy price ($/MWh) minus Henry Hub gas price ($/MMBtu) times the plant heat rate (MMBtu/MWh). A positive spread means the CCGT earns more selling power than it spends on gas. Merchant generators like Vistra and NRG see EBITDA move with this margin; contracted yieldcos do not.
- What is capacity revenue for a power plant?
- Capacity revenue is the auction clearing price ($/MW-day) times cleared megawatts times days in the period. PJM planning capacity is $250/MW-day. Cap-bound auction years can clear above that planning norm; models should not anchor on those clears. Capacity pays for availability; energy pays for dispatch.
- What wholesale energy prices should models use?
- Our planning marks: PJM $50/MWh, ERCOT $35/MWh, CAISO SP15 $33/MWh, MISO Indiana Hub $42/MWh. PJM 2025 realised load-weighted average LMP was $50.73/MWh. Henry Hub gas planning is $3.00/MMBtu from our energy planning marks (2025 actual ~$3.52). Run merchant examples on planning values, not live hub prints.
- Why do merchant generators trade below contracted yieldcos?
- Merchant EBITDA re-rates with spark spreads and capacity auction outcomes. When hub energy falls or gas rises, the spread compresses. Contracted PPAs cap that downside. The EV/EBITDA gap is ~11-15× for contracted yieldcos versus ~6-9× for merchant-heavy gas (our screening convention, Jun 2026). Retail integration and nuclear or coal baseload can lift a merchant book above the pure-gas anchor even without a filed PPA.