Skip to main content

Curve Building Methods

OpenDataDSL supports a range of curve-building methods that can be applied individually or combined to produce a complete forward curve. Each method is suited to different market structures and data availability scenarios.

Overview​

MethodBest For
BootstrappingStripping continuous curves from liquid instruments
InterpolationFilling gaps between known price points
ExtrapolationExtending a curve beyond available market data
BlendingCombining multiple source curves into one
Shape / SeasonalApplying a seasonal profile to a base curve
SpreadDeriving a curve from the difference between two others

Bootstrapping​

Bootstrapping constructs a daily (or hourly) forward curve from a set of traded instruments — such as monthly, quarterly, or seasonal contracts — by decomposing them into their constituent periods.

note

Bootstrapping is the most common method for building power and gas forward curves from exchange-traded products.

Example: A Q3 gas contract covering July, August, and September is bootstrapped into three individual monthly prices, ensuring the average of the months matches the quoted quarterly price.

Interpolation​

Interpolation fills in missing tenors between known price points. OpenDataDSL supports several interpolation algorithms:

  • Linear — straight-line interpolation between two points
  • Cubic spline — smooth curve through multiple points
  • Flat forward — holds the previous known value until the next point
// Example: linear interpolation between monthly prices
interpolation = "linear"
tip

Cubic spline interpolation produces smoother curves and is preferred when the curve will be used for mark-to-market or risk calculations.

Extrapolation​

Extrapolation extends the curve beyond the last available market quote. Common approaches include:

  • Flat — holds the last known price flat
  • Seasonal repeat — repeats the seasonal pattern from a prior year
  • Long-run marginal cost (LRMC) — reverts to a fundamental anchor price

Blending​

Blending combines two or more source curves into a single output, typically using a priority or weighted approach:

  • Priority blending — uses Source A where available, falls back to Source B
  • Weighted average — applies configurable weights to each source
note

Blending is commonly used to combine exchange settlement prices in the near curve with broker or OTC quotes for longer tenors.

Shape / Seasonal Adjustment​

A shape curve applies a multiplicative or additive seasonal profile to a base forward curve. This is useful when:

  • A base curve is only available at monthly granularity
  • You need a daily or hourly shape that reflects seasonal demand patterns

Spread Curves​

A spread curve is derived by computing the difference (or ratio) between two existing curves:

// Example: spark spread = power price - (gas price * heat rate)
sparkSpread = POWER_CURVE - (GAS_CURVE * heatRate)

Common spread curves in energy markets include:

  • Spark spread — gas-fired power generation margin
  • Dark spread — coal-fired power generation margin
  • Crack spread — refinery margin between crude and products

Combining Methods​

Smart Curves can chain multiple methods together. A typical power curve build might:

  1. Bootstrap exchange quarterly contracts into months
  2. Blend with broker quotes for the far curve
  3. Interpolate to fill any remaining gaps
  4. Apply a shape to convert from monthly to daily resolution

Next Steps​