Forward Curves

Overview

Forward electricity prices are essential for MtM valuation, deal pricing, and hedging strategy. Energy Copilot builds forward price curves from ASX Energy Futures daily settlement prices, supplemented with seasonal shape calibration for interpolating between quoted tenors.

ASX Energy Futures

ASX operates electricity futures markets for the NEM’s five regions. The primary instruments are:

Product	Tenor	Frequency	Settlement
Base Load Futures	Quarterly	Q1–Q8 (2 years forward)	Arithmetic average of all 30-min intervals
Peak Period Futures	Quarterly	Q1–Q8	7am–10pm Mon–Fri intervals
Cap Futures ($300 cap)	Quarterly	Q1–Q4	Payoff when 30-min price > $300/MWh
Strip Futures	Annual	Cal+1, Cal+2	Annual average spot

Data is sourced from ASX’s Market Data API (end-of-day) and stored in gold.asx_futures_eod.

Curve Construction

The forward curve is constructed through a bootstrapping process:

Step 1 — Load ASX EOD Data

# Fetch latest ASX futures settlement prices
futures = spark.sql("""
    SELECT
        settlement_date,
        product_code,
        region_id,
        tenor_start,
        tenor_end,
        settlement_price
    FROM energy_copilot.gold.asx_futures_eod
    WHERE settlement_date = CURRENT_DATE()
    ORDER BY region_id, tenor_start
""")

Step 2 — Fill Quarterly Gaps

ASX only quotes specific tenors. A cubic spline interpolates between quarterly anchors to produce a daily forward curve:

from scipy.interpolate import CubicSpline
import numpy as np

def build_forward_curve(anchor_dates, anchor_prices):
    """
    Build a daily forward curve via cubic spline interpolation.
    Anchor points: ASX quarterly settlement prices.
    """
    x = np.array([(d - anchor_dates[0]).days for d in anchor_dates])
    cs = CubicSpline(x, anchor_prices, bc_type='natural')
    daily_x = np.arange(0, x[-1] + 1)
    return cs(daily_x)

Step 3 — Apply Seasonal Shape

Quarterly ASX prices are flat within the quarter. To reflect intra-quarter seasonality (summer peaks, winter heating, spring shoulder), a seasonal shape calibrated from historical 30-minute spot prices is applied:

Daily_Forward = Quarterly_ASX_Price × Shape_Factor(day_of_year, region)

Seasonal shape factors are stored in gold.seasonal_shapes and updated quarterly from the prior 3 years of spot data.

Step 4 — Peak / Off-Peak Decomposition

For products that distinguish peak and off-peak, the curve decomposes:

Off_Peak_Price = (24h_average × 24 - Peak_Price × 15) / 9

Where 15 hours per day are defined as peak (7am–10pm on business days).

Dashboard Pages

Forward Curve Viewer (`/middle-office/forward-curves`)

Interactive chart: base load forward curve for selected region (0–24 months)
Toggle: base load vs peak vs off-peak
Comparison overlay: today vs 1 month ago, 3 months ago
Confidence band: implied volatility bands around the forward curve

Screenshot: Forward curve chart for NSW1 showing 24-month curve with seasonal shape and ±1σ band.

ASX Futures Table (`/middle-office/asx-futures`)

Raw ASX EOD settlement prices by quarter and product
Open interest and daily volume where available
Change from prior day settlement

Regional Curve Characteristics

Region	Summer Premium	Winter Premium	Volatility Skew
NSW1	Moderate (AC load)	Low	Symmetric
QLD1	High (AC + solar)	Low	Right-skewed (rare spikes)
SA1	High (renewable gap risk)	Low	Strong right skew
TAS1	Low	Moderate (hydro)	Left-skewed (hydro glut)
VIC1	Moderate	Low	Moderate right skew

API Endpoints

# Current forward curve for region
GET /api/forward-curves/latest?region=SA1&tenor=24months

# ASX futures EOD prices
GET /api/forward-curves/asx-futures?region=NSW1&product=base

# Historical curve comparison
GET /api/forward-curves/history?region=VIC1&date=2025-02-21

# Seasonal shape factors
GET /api/forward-curves/seasonal-shapes?region=QLD1