Paper: arXiv 2310.07692
Authors: Aurélien Alfonsi, Nerea Vadillo
Abstract
This paper develops a coupled model for day-ahead electricity prices and average daily temperature which allows to model quanto weather and energy derivatives. These products have gained on popularity as they enable to hedge against both volumetric and price risks. Electricity day-ahead prices and average daily temperatures are modelled through non homogeneous Ornstein-Uhlenbeck processes driven by a Brownian motion and a Normal Inverse Gaussian Lévy process, which allows to include dependence between them. A Conditional Least Square method is developed to estimate the different parameters of the model and used on real data. Then, explicit and semi-explicit formulas are obtained for derivatives including quanto options and compared with Monte Carlo simulations. Last, we develop explicit formulas to hedge statically single and double sided quanto options by a portfolio of electricity options and temperature options (CDD or HDD).
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 6.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper presents advanced stochastic calculus including non-homogeneous Ornstein-Uhlenbeck processes with Normal Inverse Gaussian Lévy drivers, explicit derivative pricing formulas, and static hedging strategies, reflecting high mathematical complexity. It demonstrates empirical rigor through parameter estimation via Conditional Least Squares on real electricity and temperature data, goodness-of-fit tests, and comparisons with Monte Carlo simulations.
Research Flowchart
flowchart TD A["Research Goal<br>Model & Price Quanto<br>Temperature-Electricity Derivatives"] --> B["Model Development<br>Non-Homogeneous OU + NIG Lévy Process<br>for Coupled Prices & Temperature"] B --> C["Parameter Estimation<br>Conditional Least Squares on Real Data"] C --> D["Valuation & Computation<br>Explicit/Semi-Explicit Formulas &<br>Monte Carlo Validation"] D --> E["Risk Management<br>Static Hedging Strategies<br>via Energy & Weather Options"] E --> F["Key Outcomes<br>Accurate Pricing & Efficient<br>Risk Management for Quanto Products"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper develops a coupled model for day‑ahead electricity prices and average daily temperature, provides estimation methods, derives explicit pricing formulas for quanto options and proposes static hedging strategies.
Source sentence
This paper develops a coupled model for day-ahead electricity prices and average daily temperature which allows to model quanto weather and energy derivatives.
Assumes 6
- Electricity log‑spot prices follow a non‑homogeneous Ornstein‑Uhlenbeck process with Normal Inverse Gaussian jumps. model
Source sentence
Electricity day‑ahead prices and average daily temperatures are modelled through non homogeneous Ornstein‑Uhlenbeck processes driven by a Brownian motion and a Normal Inverse Gaussian Lévy process, which allows to include dependence between them.
- Daily average temperature follows a mean‑reverting Gaussian Ornstein‑Uhlenbeck process with Brownian noise. model
Source sentence
... The parameter κT corresponds to the speed of mean‑reversion toward the trend component. ... W_T is a Brownian motion independent of L_X, and σ_T > 0 to the standard deviation of the noise.
- The dependence between price and temperature is introduced via a linear term λσ_T dW_T in the price dynamics. model
Source sentence
Last, the parameter λ ∈ R allows for some dependence between both processes.
- Temperature derivatives are not traded in liquid markets, so risk‑neutral pricing is not applicable; analysis is performed under the historical (real‑world) probability measure. market
Source sentence
Temperature and energy quantos do not exist in any open market. Therefore, risk‑neutral pricing is arguable and we stick to analyse the risk under the historical probability.
Show 2 more
- Parameters of the marginal processes (κ_X, μ_X(·), κ_T, μ_T(·), σ_T) can be estimated separately before estimating λ and the NIG parameters. statistical
Source sentence
Let us recall that the parameters of the Temperature diffusion κ_T, μ_T(·) and σ_T can be estimated as in Section 3, as well κ_X and μ_X(·) by Proposition B.1. We assume these parameters estimated, and focus first on the estimation of λ.
- The residuals of the fitted model are approximately NIG‑distributed, justifying the use of the NIG Lévy process. statistical
Source sentence
... The first motivation to keep this model is a rather good fitting of the residuals as can been seen in Figure 3.
Data
- French day‑ahead electricity market data market-prices unknown
Source sentence
χ2 tests comparing the simulated and observed joint distributions confirm the goodness of fit of the combined model for both French and Northern Italian datasets.
- Northern Italian day‑ahead electricity market data market-prices unknown
Source sentence
χ2 tests comparing the simulated and observed joint distributions confirm the goodness of fit of the combined model for both French and Northern Italian datasets.
- Daily average temperature at Charles de Gaulle and Milano Linate stations macro unknown
Source sentence
We mainly suggest using a simple mean‑reverting Gaussian model as in Benth and Benth [11] to model the daily average temperature for Charles de Gaulle and Milano Linate weather station data.
Costs
Not applicable
Finds 4
- Static hedging using the coupled model reduces the portfolio PnL variance dramatically (e.g., standard deviation from 2,197 to 391 in January). performance
Source sentence
the PnL of the portfolio is reduced on average from −3, 358 to 0.208 in January
- When λ = 0 (independent dynamics) the average PnL is strongly negative (e.g., −93.072 in January), whereas with λ < 0 the average PnL is near zero. performance
Source sentence
In January (resp. May), the average PnL of the portfolio with λ is 0.208 (resp. −0.113) instead of −93.072 (resp. −12.283) for λ = 0 for January.
- χ2 independence tests reject independence for the French dataset (p‑value 1.764e‑06) but not consistently for the Italian dataset, motivating a dependent joint model. statistical
Source sentence
... the French dataset clearly rejects the independence hypotheses while the North Italian dataset only rejects the independence hypothesis for 16 categories.
- The coupled model reproduces well the empirical features of both electricity and temperature data, as shown by goodness‑of‑fit tests on joint distributions. theoretical
Source sentence
χ2 tests comparing the simulated and observed joint distributions confirm the goodness of fit of the combined model for both French and Northern Italian datasets.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 16 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.