Paper: arXiv 2406.19405
Authors: Andrei Renatovich Batyrov
Abstract
There are several approaches to modeling and forecasting time series as applied to prices of commodities and financial assets. One of the approaches is to model the price as a non-stationary time series process with heteroscedastic volatility (variance of price). The goal of the research is to generate probabilistic forecasts of day-ahead electricity prices in a spot marker employing stochastic volatility models. A typical stochastic volatility model - that treats the volatility as a latent stochastic process in discrete time - is explored first. Then the research focuses on enriching the baseline model by introducing several exogenous regressors. A better fitting model - as compared to the baseline model - is derived as a result of the research. Out-of-sample forecasts confirm the applicability and robustness of the enriched model. This model may be used in financial derivative instruments for hedging the risk associated with electricity trading. Keywords: Electricity spot prices forecasting, Stochastic volatility, Exogenous regressors, Autoregression, Bayesian inference, Stan
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced Bayesian inference and stochastic volatility modeling with mathematical derivations (e.g., posterior predictive distributions), while also demonstrating strong empirical rigor through out-of-sample forecasting, cross-validation, and implementation in Stan with real data.
Research Flowchart
flowchart TD
Goal["Research Goal<br>Forecast Day-Ahead Electricity Spot Prices<br>Using Probabilistic Methods"]
subgraph Methodology ["Methodology: Stochastic Volatility Models"]
direction LR
Baseline["Baseline Model<br>SV with Latent Volatility"]
Enriched["Enriched Model<br>SV + Exogenous Regressors"]
end
Data["Data/Inputs<br>Electricity Spot Prices<br>Exogenous Variables<br>Weather, Demand"]
Compute["Computational Process<br>Bayesian Inference<br>Stan Implementation<br>MCMC Sampling"]
Outcome["Key Findings/Outcomes<br>1. Enriched Model Outperforms Baseline<br>2. Out-of-Sample Forecast Validation<br>3. Applicability for Risk Hedging<br>in Financial Derivatives"]
Goal --> Data
Data --> Baseline
Baseline --> Enriched
Enriched --> Compute
Compute --> Outcome
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 proposes an enriched stochastic volatility model with exogenous regressors that fits better than the baseline model.
Source sentence
A better fitting model - as compared to the baseline model - is derived as a result of the research.
Assumes 4
- The baseline model assumes a constant mean price (¯y) while allowing stochastic volatility. model
Source sentence
we will keep the assumption of constant mean, yet stochastic volatility at the same time.
- Electricity price series are non‑stationary. statistical
Source sentence
We cannot reject the null hypothesis at the significance level of 0.05. The consumer price is a non-stationary process
- Data are observed at an hourly market frequency. market
Source sentence
The time point t is subject to desired market frequency. In our research we examine hourly profiles distributed over 24 hours (days)
- Day‑of‑week effects may be included as exogenous regressors. model
Source sentence
Observe that the PACF has spikes exactly every 7 days (1 week) which further supports the correctness of possible inclusion of the day of the week as an exogenous regressor.
Data
- Hourly price zone 1 (peak hour) dataset market-prices public
Source sentence
For modeling, we will load the hourly data for the price zone 1 for the period of 01.05.2023 – 30.04.2024 (one year back from now) for the peak hour.
- Cross‑validation dataset (both zones, peak and off‑peak) market-prices public
Source sentence
For model cross-validation, we will load and examine both price zones for both peak and off‑peak hours with cross‑validation over a sliding window for the period of 23.06.2014 – 30.04.2024 (approx. ten years back from now).
- Forecasting dataset (one week ahead) market-prices public
Source sentence
For forecasting, we will generate predictions for both price zones for both peak and off‑peak hours for the period of 01.05.2024 – 07.05.2024 (one week ahead).
Costs
Not applicable
Finds 4
- SV X model reduces MAE by 7.91% on average compared to the SV Baseline model. performance
Source sentence
SV X’s MAE is 7.91% less (= better) than SV Baseline’s MAE on average
- Mann‑Whitney U tests show the improvement of SV X over SV Baseline is statistically significant (p‑values 0.0225 for MAE, 0.0439 for RMSE). statistical
Source sentence
Metric Statistic p-value Result at α MAE 11785 0.0225 ∆x > 0 RMSE 11575 0.0439 ∆x > 0
- Cross‑validation results indicate SV X outperforms SV Baseline for both price zones and hours, except price zone 1 off‑peak where SV Baseline is best. predictive-accuracy
Source sentence
The forecasting results show that for the Price zone 1 + Off‑peak hour the better performance is provided by the SV Baseline model, while for the rest three scenarios the best performer is the SV X model.
- The electricity price series are non‑stationary, justifying stochastic volatility modeling. statistical
Source sentence
We cannot reject the null hypothesis at the significance level of 0.05. The consumer price is a non‑stationary process
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 43 related papers). Coverage is partial, so this is neither replication nor refutation.
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.
Cited References (28)
- Stochastic Volatility: Likelihood Inference and Comparison with ARCH Models (1998) — link
- Probabilistic electricity price forecasting with Bayesian stochastic volatility models (2019) — link
- Daily Indices and Volumes of The Day-ahead Market — link
- Time-Series Models. Stan User’s Guide — link
- Bayesian Stochastic Volatility Model — link
- Complete Guide on Time Series Analysis in Python — link
- Modeling Energy Price Dynamics: GARCH versus Stochastic Volatility (2016) — link
- Electricity price forecasting: A review of the state-of-the-art with a look into the future (2014) — link
and 20 more
- Sometimes you really want to re-invent the wheel: EasyML (2024) — link
- Master’s Thesis Materials. GitHub Repository (2024) — link
- scikit-learn User’s Guide — link
- An Introduction to Statistical Learning with Applications in Python (2023) — link
- Critical Values For Cointegration Tests (2010) — link
- Reliable Prognosis — link
- A First Course in Machine Learning (2012)
- statsmodels User’s Guide — link
- SciPy documentation — link
- NumPy documentation — link
- pandas documentation — link
- Matplotlib User’s Guide — link
- seaborn User’s Guide — link
- Python — link
- Jupyter Notebook — link
- Visual Studio Code — link
- Mathematical Statistics With Applications (2008)
- What is a container? — link
- PyStan — link
- Interpretable Machine Learning. A Guide for Making Black Box Models Explainable (2024) — link
0 of 28 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.