README_DailyPowerCurve.md

📈 Future Day-Ahead Power Curve Construction

🧠 Objective

To model a realistic future daily Day-Ahead (DA) power price curve, we use a seasonal shaping model based on normalized historical price patterns, scaled by current forward market expectations (e.g., calendar futures).

---

🔧 Methodology Overview

Figure: Historical day-ahead price volatility with notable spikes due to exceptional events.

However, for constructing a forward curve, the key objective is to extract seasonal and daily patterns and apply them to the current market context.
To compare price dynamics across years meaningfully, we must first put all data on a common scale.

✅ Solution: Normalization

---

Step 1: Normalize Historical Prices

We start by normalizing historical daily DA prices to remove distortions caused by absolute price levels (e.g., 2022 vs. 2020):

• Normalization method:
  Scale each year's prices to a common range (e.g., min-max normalization or dividing by annual average).

• Why not standardize?
  Standardization ((x - mean) / std) distorts the shape of the price curve.
  Normalization preserves the relative seasonal and daily structure.

---

Figure: Historical day-ahead price Normalized.

---

Step 2: Classify Historical Days

To capture seasonality and weekly cycles, we tag each historical day with:

• Month (1–12)
• Week-in-month (1–5)
• Weekday (0 = Monday, …, 6 = Sunday)

This enables alignment of structurally similar days across years — e.g., all first Mondays of January.

---

Step 3: Compute the Average Shape

For each combination of (month, week_in_month, weekday), we compute the average normalized price.

This yields a representative daily shape that captures:

• 🔁 Seasonal variation (monthly and intra-month)
• 📆 Weekday vs. weekend effects
• ✅ Intra-day patterns

Figure: Yearly day-ahead price profile.

---

Step 4: Scale with Forward Prices

We construct future DA power price curves by repeating the normalized daily shape and scaling it using calendar year forward prices:

Example:

CAL = {
    '2026': 62.46,
    '2027': 60.18,
    '2028': 63.79
}

Each year's profile is scaled to match its respective forward price.

---

✅ Final Output

The final result is a daily forward power price curve that:

• Reflects historical seasonality and weekday structure
• Aligns with market expectations from forward prices
• Can be adapted for any horizon (year, quarter, month)

Figure: Forward DA Price curve.

---

📫 Contact

For further information or collaboration inquiries:
📧 hugo.lambert.perso@gmail.com