Energy spread options represent the most economically important class of commodity derivatives that most risk systems handle worst. A crack spread call option gives a refinery the right to purchase crude oil and sell refined products at a locked margin. A spark spread call option gives a gas-fired power generator the right to burn gas and sell power at a locked margin. These are real operating options embedded in physical energy infrastructure — and pricing them correctly matters for every refinery and power company in Europe.
What Each Spread Measures
The crack spread measures refinery gross margin. The simplest version — the 3-2-1 crack — prices three barrels of crude against two barrels of gasoline and one barrel of distillate. When this spread is $15/bbl, a refinery with that configuration earns $15 per barrel of crude processed before operating costs. Crack spread options let refineries lock in a minimum margin regardless of where crude and product prices move relative to each other.
The spark spread measures gas-fired power generation margin. The formula accounts for the thermal efficiency of the power plant: spark spread = power price - (gas price × heat rate). For a combined cycle gas turbine with a heat rate of 7.5 MMBtu per MWh, the spark spread = power price ($/MWh) - gas price ($/MMBtu) × 7.5. When the spark spread is positive, it is economically rational to run the plant. When negative, the plant loses money on variable costs alone.
The dark spread measures coal-fired power generation margin using the same logic as spark spread but substituting coal for gas. Increasingly relevant as European coal plants have operational flexibility to switch between running and being mothballed depending on carbon prices and coal-to-gas switching economics.
Why Standard Two-Leg Spread Pricing Is Wrong for Energy
The most common approach to energy spread option pricing treats the spread as a single underlying and applies Black-76 to the spread price. This produces an implied vol for the spread, which can then be used to price other spread options. The problem is that crude, gasoline, distillate, gas, and power each have individual vol surfaces, and their correlations are not constant. The spread vol implied by treating the spread as a single asset will be correct on average but wrong during correlation stress events — which are exactly the events refiners and power generators most need to hedge.
Consider a crack spread call with a three-month tenor purchased in February 2020. Crude oil vol was 25%. Gasoline vol was 28%. Crack spread vol, estimated from the two individual vols and their historical correlation of 0.92, was approximately 10%. In March 2020, the COVID demand shock hit crude and petroleum products simultaneously but at different magnitudes and speeds. Crude fell 60%; gasoline fell 40% initially but then fell 70% as demand collapsed. The crack spread widened dramatically before collapsing. The spread's realized vol during this period was more than 35% — more than three times the pre-crisis estimate from the single-asset model.
A two-factor model that tracks crude and product separately, with time-varying correlation, predicted the crack spread vol rise because it observed the increasing correlation breakdown between crude and product prices during the demand shock. The single-factor spread model was blind to the correlation change.
Gas-Power Correlation and the Spark Spread Regime Problem
European spark spreads have three distinct correlation regimes that have been clearly observable over the past decade. In the first regime — normal conditions — gas and power prices move together because gas is the marginal fuel for power generation across much of the European market. Correlation between UK NBP gas and GB baseload power is typically 0.80–0.90. Spark spread vol is low, around 10–15%, because the two prices move nearly together.
In the second regime — renewable displacement — wind and solar generation push power prices below gas-based generation costs. Gas and power prices decouple as power prices reflect renewable marginal cost rather than gas cost. Correlation falls to 0.30–0.50. Spark spread vol rises to 25–35% because the spread can widen and narrow based on renewable generation levels that are uncorrelated with gas prices.
In the third regime — gas supply stress — gas prices spike while power prices lag because demand-side response and interconnection from neighboring markets partially absorb the gas price move. The 2021-2022 European gas crisis produced exactly this pattern: Dutch TTF gas rose from €25/MWh to €340/MWh, but German and UK baseload power rose much more slowly, meaning spark spreads went deeply negative and power generators were losing money running gas plants even at peak demand.
A spark spread option model that does not explicitly model these three regimes will produce incorrect option prices during regime transitions — which is when spark spread options are most valuable and most important to price accurately. Allasso's spark spread model uses a regime-switching structure calibrated to European gas and power data, producing option prices that embed a realistic probability of regime transitions rather than assuming stable correlations throughout the option's life.
Carbon Emissions and the Dirty Spread
Since the European Union Emissions Trading Scheme (EU ETS) became economically binding in 2018, carbon allowance prices have become a third input to European power generation economics. For coal plants, the variable cost is coal cost plus EU ETS carbon cost (coal emits approximately 0.35 tonnes of CO2 per MWh). For gas plants, the carbon cost is lower (approximately 0.2 tonnes CO2/MWh for efficient combined cycles).
The clean dark spread incorporates carbon: clean dark spread = power price - coal cost × coal heat rate - carbon price × 0.35. The clean spark spread incorporates carbon analogously. As carbon prices have risen from €5/tonne in 2017 to €60-€95/tonne in 2022-2024, the carbon cost term has moved from negligible to dominant in determining plant dispatch economics.
Options on clean spreads must therefore incorporate vol from a third underlying: EU ETS carbon allowances (EUAs). EUA vol has typically been 35–55% annualized, which is higher than either gas vol or coal vol. The three-factor model for clean spark or dark spread options is significantly more complex than the two-factor model, but the carbon contribution to spread vol is large enough that ignoring it misprices options by material amounts — 15–25% of option premium for at-the-money one-year clean spark spread options at current carbon price levels.
Timing Conventions and Delivery Period Matching
Energy spread options require careful attention to timing conventions. A crack spread option must reference crude and product prices at the same delivery date. A spark spread option must reference gas and power at the same delivery hour. Mismatched settlement dates introduce basis that can be $2–$8 per barrel (for crack spreads) or £5–£15 per MWh (for spark spreads) depending on the shape of the forward curves at settlement time.
Allasso enforces delivery period matching for all spread option contracts. When a user enters a crack spread option, the system automatically sets the crude, gasoline, and distillate legs to the same delivery month and verifies that the forward prices used for valuation are aligned. For hourly-granularity spark spread options — which some European gas and power desks use to hedge specific generation hours — the system prices each hour separately using hourly forward price curves rather than applying daily average prices, which would understate vol for peak-hour options.
Structuring Spread Options for Refinery and Power Hedging Programs
Refineries and power generators who use spread options as part of structured hedging programs face a design question that goes beyond pricing: how to combine spread options with outright commodity options to achieve a target margin protection level at minimum premium cost.
The standard approach — buy at-the-money spread options for the target margin level — is typically not cost-efficient. A more efficient structure combines an out-of-the-money spread option (which is cheaper but only activates if margins deteriorate significantly) with a short outright put on crude (which offsets some premium by selling downside protection that the refinery does not need because lower crude prices typically improve refinery margins).
Allasso's structured products module allows users to build multi-leg hedge structures and visualize the total payoff, premium cost, and risk profile across a range of spot and correlation scenarios. The module can optimize a hedge structure for minimum premium given a target protection level, accounting for the correct cross-commodity Greeks throughout. Contact our team in Zurich to see a worked example for your specific commodity exposure profile.