arXiv subjects

Arkadiusz Lipiecki

Publications and source records attributed to Arkadiusz Lipiecki.

10 recordsLinked to original sources

arXiv2026

Foundation models for electricity price forecasting and battery arbitrage: Can they replace market-specific forecasting models?

Foundation models promise accurate forecasts with little or no task-specific training, but whether they can replace models designed specifically for electricity price forecasting remains unclear. We compare nine variants from five foundation model families, evaluated in zero-shot mode, with two state-of-the-art electricity price forecasting benchmarks in Germany, Poland, and Spain over 2021-2025. Their performance is assessed in terms of point and probabilistic forecasting accuracy, as well as economic value in battery energy storage arbitrage. Only the TabPFN models consistently and significantly outperform the benchmarks across all three markets and all statistical measures. However, this statistical dominance does not translate directly into economic dominance: TabPFN performs best under unlimited bids and riskier quantile-based strategies, whereas the Distributional Deep Neural Network benchmark is more profitable when risk tolerance is lower. Thus, foundation models cannot universally replace market-specific models, and their value depends on both model architecture and the decision problem.

cs.LG↗

arXiv2026

Probabilistic forecasting via post-processing prediction errors: In- or out-of-sample?

Many forecasting systems produce point forecasts even when decisions require information about uncertainty. We investigate whether post-processing methods can systematically improve upon traditional Gaussian predictive distributions constructed from in-sample residuals. We propose a hybrid framework that combines forecast error post-processing with model-specific scaling of forecast uncertainty across horizons. For a comprehensive evaluation, we apply historical simulation, conformal prediction, quantile regression, and GARCH-based post-processing to point forecasts generated by Theta, exponential smoothing, and ARIMA models. Using 14,407 monthly series from the M4 competition and forecast horizons of 1 to 12 months, we evaluate performance using the continuous ranked probability score and rank-based statistical comparisons. Averaged across horizons, all post-processing variants improve upon the benchmark predictive distributions, with gains of up to 4.6%. In-sample calibration outperforms its out-of-sample counterpart in 11 of the 12 model-method combinations, although the preferred post-processing method depends on the base model and forecast horizon. The advantage of in-sample calibration generally increases at longer horizons. Our results show that organisations can extend existing point-forecasting systems to provide useful uncertainty quantification without computationally intensive repeated model re-estimation.

stat.AP↗

arXiv2026

Competition between private and expressed opinions in binary choice: the $α$-EPO $q$-voter model

People often express opinions that differ from their privately held views, a phenomenon known in economy as preference falsification. Expressed-private opinion (EPO) models capture this by assigning each agent two dynamical variables: a private (internal) and an expressed (external) opinion. Within the nonlinear $q$-voter model, two EPO variants have been studied so far: with and without self-anticonformity. In both formulations, agents update private and expressed binary opinions, one after another and at the same rate, which has led to two update schemes studied previously: AT (act then think), in which an agent first updates its expressed and then its private opinion, and TA (think then act), in which the order is reversed. To eliminate this ad hoc distinction and quantify the interplay between private and expressed opinions, we introduce the $α$-EPO $q$-voter model with asynchronous updating -- in each elementary step, an agent updates its private opinion with probability $α$ or its expressed opinion with complementary probability $1-α$. We derive mean-field theory and, for the first time for EPO $q$-voter dynamics, a pair approximation, and validate them with Monte Carlo simulations on artificial and real organizational networks. Comparing the two model variants, we show that the collective outcome controlled by $α$ strongly depends on self-anticonformity: with self-anticonformity the results are robust to $α$, whereas without it $α$ shifts the agreement-disagreement threshold and can change the type of phase transition. In the mean-field limit this change occurs only for $q=3$, but the pair approximation reveals an additional low-connectivity regime in which both $α$ and the average degree $k$ control the emergence and width of hysteresis also for larger influence groups.

physics.soc-ph↗

arXiv2026

Statistical and economic evaluation of forecasts in electricity markets: beyond RMSE and MAE

Electricity price forecasts are typically evaluated using accuracy measures such as RMSE and MAE, although these metrics often fail to reflect their economic value in operational decisions. This paper investigates which statistical properties of electricity price forecasts are most relevant for economic performance, using battery energy storage system (BESS) arbitrage as an application. We assess prediction quality along four dimensions: forecast accuracy, intraday error dispersion, association between predicted and realized prices, and the ability to identify daily price extrema. We construct a comprehensive pool of 192 hourly day-ahead electricity price forecasts and use it to evaluate the relationship between proposed quality measures and profits generated for two representative BESS configurations. The results show that traditional accuracy metrics are only weakly correlated with BESS income. At the same time, dispersion- and association-based measures better capture a forecast's economic value by reflecting its ability to reproduce daily price patterns. These findings demonstrate that incorporating complementary evaluation criteria may improve forecast selection and enhance the economic performance of BESS.

q-fin.CP↗

arXiv2026

Stealing Accuracy: Predicting Day-ahead Electricity Prices with Temporal Hierarchy Forecasting (THieF)

We introduce the concept of temporal hierarchy forecasting (THieF) in predicting day-ahead electricity prices and show that reconciling forecasts for hourly products and 2- to 24-hour blocks can significantly (up to 13%) improve accuracy at all levels. These results remain consistent throughout a challenging 4-year test period (2021-2024) in the German and Spanish power markets and across model architectures, including linear regression, shallow feedforward neural networks, gradient-boosted decision trees, and a state-of-the-art, pretrained transformer. Given that (i) trading of block products is becoming more common and (ii) the computational cost of reconciliation is comparable to that of predicting hourly prices alone, we recommend using it in daily forecasting practice.

q-fin.ST↗

arXiv2025

When heterogeneity drives hysteresis: Anticonformity in the multistate $q$-voter model on networks

Discontinuous phase transitions are closely linked to tipping points, critical mass effects, and hysteresis, phenomena that have been confirmed empirically and recognized as highly important in social systems. The multistate $q$-voter model, an agent-based approach to simulate discrete decision-making and opinion dynamics, is particularly relevant in this context. Previous studies of the $q$-voter model with anticonformity on complete graphs uncovered a counterintuitive result. Changing the model formulation from the annealed (homogeneous agents with varying behavior) to quenched (heterogeneous agents with fixed behavior) produces discontinuous phase transitions. This is contrary to the common expectation that quenched heterogeneity smooths transitions. To test whether this effect is merely a mean-field artifact, we extend the analysis to random graphs. Using pair approximation and Monte Carlo simulations, we show that the phenomenon persists beyond the complete graph, specifically on random graphs and Barabási-Albert scale-free networks. The novelty of our work is twofold: (i) we demonstrate for the first time that replacing the annealed with the quenched approach can change the type of phase transitions from continuous to discontinuous not only on complete graphs but also on sparser networks, and (ii) we provide pair-approximation results for the multistate $q$-voter model with competing conformity and anticonformity mechanisms, covering both quenched and annealed cases, which had previously been studied only in binary models.

physics.soc-ph↗

arXiv2025

Isotonic Quantile Regression Averaging for uncertainty quantification of electricity price forecasts

Quantifying the uncertainty of forecasting models is essential to assess and mitigate the risks associated with data-driven decisions, especially in volatile domains such as electricity markets. Machine learning methods can provide highly accurate electricity price forecasts, critical for informing the decisions of market participants. However, these models often lack uncertainty estimates, which limits the ability of decision makers to avoid unnecessary risks. In this paper, we propose a novel method for generating probabilistic forecasts from ensembles of point forecasts, called Isotonic Quantile Regression Averaging (iQRA). Building on the established framework of Quantile Regression Averaging (QRA), we introduce stochastic order constraints to improve forecast accuracy, reliability, and computational costs. In an extensive forecasting study of the German day-ahead electricity market, we show that iQRA consistently outperforms state-of-the-art postprocessing methods in terms of both reliability and sharpness. It produces well-calibrated prediction intervals across multiple confidence levels, providing superior reliability to all benchmark methods, particularly coverage-based conformal prediction. In addition, isotonic regularization decreases the complexity of the quantile regression problem and offers a hyperparameter-free approach to variable selection.

cs.LG↗

arXiv2024

Postprocessing of point predictions for probabilistic forecasting of day-ahead electricity prices: The benefits of using isotonic distributional regression

Operational decisions relying on predictive distributions of electricity prices can result in significantly higher profits compared to those based solely on point forecasts. However, the majority of models developed in both academic and industrial settings provide only point predictions. To address this, we examine three postprocessing methods for converting point forecasts of day-ahead electricity prices into probabilistic ones: Quantile Regression Averaging, Conformal Prediction, and the recently introduced Isotonic Distributional Regression. We find that while the latter demonstrates the most varied behavior, it contributes the most to the ensemble of the three predictive distributions, as measured by Shapley values. Remarkably, the performance of the combination is superior to that of state-of-the-art Distributional Deep Neural Networks over two 4.5-year test periods from the German and Spanish power markets, spanning the COVID pandemic and the war in Ukraine.

q-fin.ST↗

arXiv2022

Polarization in the three-state $q$-voter model with anticonformity and bounded confidence

Engaging with dissenting views, fostering productive disagreements or strategic anticonformity can benefit organizations as it challenges the status quo. The question arises, however, whether such strategic anticonformity ultimately leads to social polarization, which is not a desirable phenomenon. We address this question within an agent-based model of discrete choices. Using the way of modeling social responses in continuous opinion models, we propose a three-state $q$-voter model with anticonformity and bounded confidence. We analyze the model on a complete graph using the mean-field approach and Monte Carlo simulations. We show that strong polarization appears only for a small probability of anticonformity, which means that conformity combined with homophily enhances polarization. Our findings agree with results obtained previously in the group discussion experiment and within various continuous opinion models.

physics.soc-ph↗

arXiv2021

Discontinuous phase transitions in the q-voter model with generalized anticonformity on random graphs

We study the binary $q$-voter model with generalized anticonformity on random Erdős-Rényi graphs. In such a model, two types of social responses, conformity and anticonformity, occur with complementary probabilities and the size of the source of influence $q_c$ in case of conformity is independent from the size of the source of influence $q_a$ in case of anticonformity. For $q_c=q_a=q$ the model reduces to the original $q$-voter model with anticonformity. Previously, such a generalized model was studied only on the complete graph, which corresponds to the mean-field approach. It was shown that it can display discontinuous phase transitions for $q_c \ge q_a + Δq$, where $Δq=4$ for $q_a \le 3$ and $Δq=3$ for $q_a>3$. In this paper, we pose the question if discontinuous phase transitions survive on random graphs with an average node degree $\langle k\rangle \le 150$ observed empirically in social networks. Using the pair approximation, as well as Monte Carlo simulations, we show that discontinuous phase transitions indeed can survive, even for relatively small values of $\langle k\rangle$. Moreover, we show that for $q_a < q_c - 1$ pair approximation results overlap the Monte Carlo ones. On the other hand, for $q_a \ge q_c - 1$ pair approximation gives qualitatively wrong results indicating discontinuous phase transitions neither observed in the simulations nor within the mean-field approach. Finally, we report an intriguing result showing that the difference between the spinodals obtained within the pair approximation and the mean-field approach follows a power law with respect to $\langle k\rangle$, as long as the pair approximation indicates correctly the type of the phase transition.

physics.soc-ph↗