Pith. sign in

REVIEW 3 major objections 5 minor 98 references

Electricity Market Predictability: Virtues of Machine Learning and Links to the Macroeconomy

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Machine learning predicts Singapore power prices up to 52% out of sample.

desk verdict A genuinely ambitious Singapore electricity forecasting study undone by a likely look-ahead bias in the feature pipeline; worth a careful rework, not acceptance as-is. read the letter →

arxiv 2507.07477 v1 pith:Y6QCCYDI submitted 2025-07-10 econ.GN q-fin.EC

classification econ.GNq-fin.EC
keywords machinelearningelectricitypriceforecastingout-of-samplepredictabilityensemblemethodsSingaporewholesalemarketmacro-financelinksfeatureimportanceregime-dependent
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that machine learning models, fed with a rich set of 619 daily features from Singapore's wholesale electricity market and the macroeconomy, can predict next-day electricity returns out of sample with economically large accuracy. The headline numbers are positive and statistically significant out-of-sample R-squared values for all models, with the best three individual models (GLM, XGB+H, LGBM+H) around 43-44% and a correlation-penalized ensemble reaching about 52%. If true, this means the USEP, the market's settlement price, is substantially forecastable, enough to generate sizable gains for a mean-variance investor, and that predictability is not uniform: it clusters in expansion periods, volatile markets, and high geopolitical risk. The paper also argues the results confirm three 'virtues' of ML—capturing nonlinearity, using complexity in high dimensions, and exploiting $\ell^2$-regularization and bagging when factors are weak.

What carries the argument

The load-bearing mechanism is the correlation-penalized ensemble weighting scheme, which minimizes the validation-period MSE plus a penalty $\lambda \sum_{m,m'} w_m w_{m'} \rho_{m,m'}$ on pairwise prediction correlations, so that redundant highly-correlated models are downweighted. Around this sits the 619-feature design: 35 in-market variables, 7 domestic macro, 9 international macro, 560 macro times in-market interaction terms, plus the lagged price and weekday dummies. The evaluation uses a recursive expanding-window scheme with three out-of-sample $R^2$ benchmarks (lag price, AR(1), historical mean/zero) and DM/CW tests.

What would settle it

Recompute the full prediction pipeline with each day's features standardized using only data up to that day (e.g., trailing or expanding normalization), and compare the resulting R-squared values with the reported ones; if the correlation-penalized ensemble's $R^2_{OOS,mean}$ drops substantially below 51.92% or another model overtakes it, the reported predictability is inflated by look-ahead leakage. A single out-of-sample month where the gap exceeds the DM test's margin would settle the question.

Watch

Extended reading notes

Core claim

The central discovery is that daily Singapore electricity returns ($r_{t+1}$ from USEP log prices) are predictable out of sample once one combines in-market stakeholder data (supply, demand, regulation), domestic and international macro variables, and their interactions. Using a recursive expanding-window scheme, all 15 individual models and 4 ensembles deliver positive out-of-sample $R^2$ against lag-price, AR(1), historical-mean and zero benchmarks; the top individual models reach $R^2_{OOS,mean} \approx 43\%$ and the correlation-penalized ensemble (Ensemble wp) reaches 51.92%. The paper traces this predictability to three ML virtues—nonlinearity capture, complexity, and $\ell^2$/bagging in weak-factor settings—and shows it translates into economic value: a mean-variance investor would accept a risk-free rate of up to 77.15% in lieu of an XGB(+H)-based risky portfolio. Furthermore, predictability is heterogeneous across macro regimes—concentrated in expansions (high night-light intensity), volatile electricity markets, and extreme geopolitical risk periods—while feature attribution points to a supply-side-driven market with strong regulatory influence.

Load-bearing premise

The central results stand or fall on the claim that each day's feature values are constructed only from information available on that day; if the monthly (0,1) standardization uses the full calendar month's mean and standard deviation, then the out-of-sample predictions peek at future data within the very month being forecast.

Editorial extensions

If this is right

  • Singapore's USEP is forecastable enough for practical market operations, so suppliers, retailers, and regulators could use these models for production planning, hedging, and market-monitoring decisions.
  • The correlation-penalized ensemble beats the best individual models and stays strong before and after the 2021 energy crisis, offering a forecast-combination recipe for other high-correlation settings.
  • Predictability is not constant: it concentrates in expansions and volatile regimes, so a regime-aware forecasting system should expect better performance in those states.
  • The feature-importance results imply that the market is supply-side driven and heavily regulated even post-liberalization, which matters for policy evaluation in Singapore and comparable deregulated markets.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the correlation-penalized ensemble should transfer to any forecast-combination problem with highly correlated model outputs, such as equity premium or inflation forecasting, where average pairwise correlation is typically high.
  • Editorial inference: if the monthly-standardization leakage is confirmed, the absolute R-squared numbers should be read as upper bounds, and the ranking across models and regimes may shift once strict causal standardization is enforced.
  • Editorial inference: the dominance of interaction terms suggests that separate markets will need their own stakeholder-level interaction features, so the 52% figure is unlikely to generalize to markets without similarly rich in-market data.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents a large-scale comparative machine learning study for forecasting daily Singapore electricity price returns (USEP log-returns) using 619 features that combine in-market stakeholder data, domestic macro indicators, international macro factors, and interaction terms. The main text reports out-of-sample R² values for 15 individual models and 4 ensemble methods over 2018-2023, claiming that all models achieve positive and statistically significant OOS R² at the 1% level, with top individual models (GLM, XGB(+H), LGBM(+H)) reaching roughly 43% R²_OOS,mean and a correlation-penalized ensemble reaching 51.92%. The paper further reports regime-dependent predictability (bullish/volatile/high-growth periods, geopolitical risk), a trend-based decomposition of OOS performance, a group-level feature importance measure, and utility gains under a mean-variance framework.

Significance. If the empirical claims held, this would be a substantial contribution: the dataset is unusually rich in market-stakeholder variables, the model coverage is broad, the proposed correlation-penalized ensemble is novel and plausibly useful, and the macro-regime analysis is policy-relevant. The paper also connects to recent asset-pricing ML literature (Gu et al., Kelly et al., Shen and Xiu) and provides a large set of robustness checks. However, the manuscript does not provide code or replication data, and the evaluation pipeline contains a critical look-ahead issue in feature standardization that directly contaminates the headline OOS R² values and all results built on them. The verification burden is therefore high, and the current reported magnitudes are not credible without a causal re-estimation.

major comments (3)
  1. [Appendix A.3.2, Section 3.2] The monthly (0,1) standardization described in Appendix A.3.2 computes each feature's monthly mean and standard deviation over the entire calendar month. For an out-of-sample day t in month T, the standardization uses the month's mean and standard deviation, which are calculated using all days of that month, including days after t. This violates the paper's own information-set restriction in Section 3.2 that predictions at time t must not use information unavailable at t. Because the feature set includes the lagged USEP, the monthly moments embed future price realizations, mechanically inflating the OOS R² values in Table 1, the DM/CW test outcomes, the model rankings, the regime comparisons in Table 6, and the utility results in Section 4.4. The entire OOS evaluation must be re-run with a causal standardization scheme (e.g., expanding-window or trailing moments) before the paper's central claims can be assessed.
  2. [Section 4.2.1, Table 6] The macro-regime analysis defines states using full-sample terciles: for example, bearish/bullish market states are based on the full-sample (2003-2023) distribution of daily returns, volatile/tranquil states on full-sample monthly variance, and high/low night-light and GPR states on full-sample distributions. Since the OOS period (2018-2023) is included in the full sample, regime membership for a given OOS day depends on future observations, importing look-ahead information into the state-specific R² calculations. State definitions must be constructed using only in-sample data (or with a recursive/expanding information set) to provide a valid decomposition of OOS predictability.
  3. [Section 4.1.1, Table 1] The text states that 'all R²_OOS values are positive and statistically significant at the 1% level according to the one-sided Diebold-Mariano (DM) test,' but Table 1 contradicts this: for example, OLS's R²_OOS,AR(1) is reported as 14.16***(), with no DM significance inside the parentheses, and many entries in the R²_OOS,lagprice and R²_OOS,AR(1) columns similarly lack DM stars. If the parentheses denote DM significance, the claim is false for those metrics; if the notation means something else, it must be clarified. This inconsistency is load-bearing because the paper's first listed contribution is the universal 1% significance of all R²_OOS values.
minor comments (5)
  1. [Abstract] The abstract states that 'Simulation also supports the first virtue' but does not cite the simulation appendix (Appendix C.3); consider adding a reference for readability.
  2. [Section 3.1] The text says the validation set is 'fixed' in the recursive scheme, but Appendix B.1 shows the validation period rolling forward with each OOS month; the wording should be corrected to 'rolling validation window.'
  3. [Equation (17)] The correlation penalty term ρ_{m,m'} is used in Equation (17) before it is explicitly defined; define it as the Pearson correlation between predictions of models m and m' immediately after the equation.
  4. [General] The manuscript contains no data or code availability statement; given the NDA data note, a clear statement about what can be shared (even synthetic or de-identified data) would help future verification.
  5. [General] Several cross-references and appendix labels appear inconsistent (e.g., Figure A6 is referenced in the text but the figure numbering in the appendix is not fully clear); a careful pass for referencing errors is needed.

Circularity Check

0 steps flagged · score 2.0 of 10

No material circularity: the core OOS comparisons are externally benchmarked; the flagged monthly standardization is a look-ahead leakage risk, not a definitional reduction, and the only self-citation is a non-load-bearing data source.

full rationale

The central derivation chain is not circular. The headline R2_OOS results are computed by comparing genuinely produced model forecasts against external benchmarks (lag price, AR(1), historical mean, zero, and OLS) with DM/CW/MCS/GW tests; those benchmarks are not functions of the model outputs, so the performance comparisons are independent evidence. The 'virtues of ML' arguments are supported by external theoretical results (Kelly et al. 2024; Shen and Xiu 2024), by a Monte Carlo simulation with known DGPs, and by a panel regression of complexity on OOS R2, not by restating the hypotheses. The only self-citation of a coauthor is Foo et al. (2023), used as the source of monthly Singapore macro indices; this is a data citation and is not load-bearing for the forecasting derivation. Two reader-flagged concerns are statistical artifacts rather than circularity: (i) since 560 of 619 features are interaction terms, top-20 feature-importance lists will be interaction-heavy by feature-set composition, but the importance values are still obtained by retraining models and are not defined as feature counts; (ii) state-specific R2 normalizes by state-specific variance, so volatile-state comparisons are variance-scaled, but that is the standard definition of R2, not a self-referential reduction. The Appendix A.3.2 monthly (0,1) standardization is a genuine look-ahead risk: applying a full OOS month's mean and standard deviation to each day's feature vector uses days after t and violates the Section 3.2 information-set restriction, potentially inflating every reported R2_OOS value. This is a serious correctness/evaluation-leakage flaw, but it is not a case where the prediction is definitionally equal to its input; it does not make the derivation circular. The EMC confidentiality note means no external code/data check can resolve the leak, so the OOS magnitudes should be treated with caution on correctness grounds, not circularity grounds.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central empirical claims depend on a large set of tuned hyperparameters, on full-sample macro-state terciles, and on the assumption that monthly standardization is causally valid. There are no invented physical entities. The heaviest burdens are the look-ahead in standardization and the state-splitting procedure.

free parameters (3)
  • Grid-search hyperparameters for 15 models = Selected by validation loss
    L1/L2 penalties, number of components, spline nodes, tree depth, learning rate, and NN architectures are tuned on the validation window; standard model-selection choices, not the paper's target parameters.
  • Correlation penalty lambda in Ensemble wp = Selected on validation window
    The hyperparameter in Eq. (17) controls the trade-off between accuracy and diversity; its value is data-driven and not externally fixed.
  • Macro-regime tercile cutoffs = Bottom, middle, and top thirds of full-sample distributions
    Choices define which days count as bullish, volatile, high night light, or high geopolitical risk; because they are computed on the full 2003-2023 sample, they include the out-of-sample period and are not external inputs.
assumptions (5)
  • domain assumption The target return decomposes as rt+1 = mu_{t+1} + epsilon_{t+1} with E[epsilon_{t+1} | F_t] = 0
    Section 3.2, Eqs. (1)-(2); this is the basis for treating the conditional mean as the forecast target.
  • domain assumption Temporal stability of the feature-target relationship over the expanding window
    Section 3.1, footnote 6; necessary for models trained on distant history to remain predictive in the out-of-sample period.
  • ad hoc to paper Monthly standardization using current-month moments is compatible with causal out-of-sample prediction
    Appendix A.3.2; for days inside out-of-sample months, the month's mean and standard deviation include future days, violating the information-set restriction in Section 3.2.
  • ad hoc to paper Macro states can be formed from full-sample terciles without contaminating out-of-sample evaluation
    Section 4.2.1 and footnote 22; states use 2003-2023 realizations including the 2018-2023 test window, so regime comparisons are not purely out-of-sample.
  • domain assumption USEP return mean is near zero, so mean and zero benchmarks are interchangeable
    Section 4.1.1; supported by summary statistics but used to justify the equivalence of R2_OOS,mean and R2_OOS,zero.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Electricity Market Predictability: Virtues of Machine Learning and Links to the Macroeconomy." pith.science (2026). https://pith.science/paper/Y6QCCYDI

@misc{pith2026250707477,
  author       = {Pith},
  title        = {Pith review of: Electricity Market Predictability: Virtues of Machine Learning and Links to the Macroeconomy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y6QCCYDI}},
  note         = {Machine review of arXiv:2507.07477}
}
read the original abstract

With stakeholder-level in-market data, we conduct a comparative analysis of machine learning (ML) for forecasting electricity prices in Singapore, spanning 15 individual models and 4 ensemble approaches. Our empirical findings justify the three virtues of ML models: (1) the virtue of capturing non-linearity, (2) the complexity (Kelly et al., 2024) and (3) the l2-norm and bagging techniques in a weak factor environment (Shen and Xiu, 2024). Simulation also supports the first virtue. Penalizing prediction correlation improves ensemble performance when individual models are highly correlated. The predictability can be translated into sizable economic gains under the mean-variance framework. We also reveal significant patterns of time-series heterogeneous predictability across macro regimes: predictability is clustered in expansion, volatile market and extreme geopolitical risk periods. Our feature importance results agree with the complex dynamics of Singapore's electricity market after de regulation, yet highlight its relatively supply-driven nature with the continued presence of strong regulatory influences.

Figures

Figures reproduced from arXiv: 2507.07477 by the authors.

Figure 1
Figure 1. Example of the Market Clearing Process From EMA (2009), electricity prices in this figure can be negative due to oversupply. Negative prices mean that power companies need to ”pay” users or grid operators to consume the excess electricity. This phenomenon reflects the unique characteristics of the electricity market, as electricity is difficult to store, and excess supply can drive prices into negative territory. 47… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

98 extracted references · 69 canonical work pages

  1. [1]

    K., Saini, L

    Aggarwal, S. K., Saini, L. M., and Kumar, A. (2009). Electricity price forecasting in deregulated markets: A review and evaluation. International Journal of Electrical Power & Energy Systems , 31(1):13--22

  2. [2]

    Avramov, D., Cheng, S., and Metzker, L. (2023). Machine learning vs. economic restrictions: Evidence from stock return predictability. Management Science , 69(5):2587--2619

  3. [3]

    and Ng, S

    Bai, J. and Ng, S. (2008). Forecasting economic time series using targeted predictors. Journal of Econometrics , 146(2):304--317

  4. [4]

    and Perron, P

    Bai, J. and Perron, P. (1998). Estimating and testing linear models with multiple structural changes. Econometrica , pages 47--78

  5. [5]

    Bates, J. M. and Granger, C. W. (1969). The combination of forecasts. Journal of the operational research society , 20(4):451--468

  6. [6]

    and Barqu \' n, J

    Batlle, C. and Barqu \' n, J. (2005). A strategic production costing model for electricity market price analysis. IEEE Transactions on Power Systems , 20(1):67--74

  7. [7]

    Bento, P., Pombo, J., Calado, M., and Mariano, S. (2018). A bat optimized neural network and wavelet transform approach for short-term price forecasting. Applied energy , 210:88--97

  8. [8]

    Bianchi, D., B \"u chner, M., and Tamoni, A. (2021). Bond risk premiums with machine learning. The Review of Financial Studies , 34(2):1046--1089

Show all 98 references
  1. [9]

    W., Lisi, F., and Nan, F

    Bordignon, S., Bunn, D. W., Lisi, F., and Nan, F. (2013). Combining day-ahead forecasts for british electricity prices. Energy Economics , 35:88--103

  2. [10]

    Breiman, L. (2001). Random forests. Machine learning , 45:5--32

  3. [11]

    Breiman, L. (2002). Manual on setting up, using, and understanding random forests v3. 1. Statistics Department University of California Berkeley, CA, USA , 1(58):3--42

  4. [12]

    Cakici, N., Fieberg, C., Neumaier, T., Poddig, T., and Zaremba, A. (2024). Pockets of predictability: A replication. Journal of Finance, forthcoming

  5. [13]

    and Iacoviello, M

    Caldara, D. and Iacoviello, M. (2022). Measuring geopolitical risk. American Economic Review , 112(4):1194--1225

  6. [14]

    Y., Lo, A

    Campbell, J. Y., Lo, A. W., MacKinlay, A. C., and Whitelaw, R. F. (1998). The econometrics of financial markets. Macroeconomic Dynamics , 2(4):559--562

  7. [15]

    Campbell, J. Y. and Shiller, R. J. (1988). The dividend-price ratio and expectations of future dividends and discount factors. The review of financial studies , 1(3):195--228

  8. [16]

    Campbell, J. Y. and Thompson, S. B. (2008). Predicting excess stock returns out of sample: Can anything beat the historical average? The Review of Financial Studies , 21(4):1509--1531

  9. [17]

    Chen, J., Tang, G., Zhou, G., and Zhu, W. (2023). Chatgpt, stock market predictability and links to the macroeconomy. Olin Business School Center for Finance & Accounting Research Paper , (2023/18)

  10. [18]

    Chen, J., Tang, G., Zhou, G., and Zhu, W. (2025). Chatgpt and deepseek: Can they predict the stock market and macroeconomy? arXiv preprint arXiv:2502.10008

  11. [19]

    T., and Xiu, D

    Chen, Y., Kelly, B. T., and Xiu, D. (2022). Expected returns and large language models. Available at SSRN 4416687

  12. [20]

    Clark, T. E. and West, K. D. (2007). Approximately normal tests for equal predictive accuracy in nested models. Journal of econometrics , 138(1):291--311

  13. [21]

    Cochrane, J. H. and Piazzesi, M. (2005). Bond risk premia. American economic review , 95(1):138--160

  14. [22]

    J., Contreras, J., Esp \' nola, R., and Plazas, M

    Conejo, A. J., Contreras, J., Esp \' nola, R., and Plazas, M. A. (2005). Forecasting electricity prices for a day-ahead pool-based electric energy market. International journal of forecasting , 21(3):435--462

  15. [23]

    W., Feng, G., He, J., and Wang, Y

    Cong, L. W., Feng, G., He, J., and Wang, Y. (2024). Mosaics of predictability. Available at SSRN 4853767

  16. [24]

    W., Tang, K., Wang, J., and Zhang, Y

    Cong, L. W., Tang, K., Wang, J., and Zhang, Y. (2021). Alphaportfolio: Direct construction through deep reinforcement learning and interpretable ai. Available at SSRN 3554486

  17. [25]

    Cybenko, G. (1989). Approximation by superpositions of a sigmoidal function. Mathematics of control, signals and systems , 2(4):303--314

  18. [26]

    and Halling, M

    Dangl, T. and Halling, M. (2012). Predictive regressions with time-varying coefficients. Journal of Financial Economics , 106(1):157--181

  19. [27]

    Dickey, D. A. and Fuller, W. A. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American statistical association , 74(366a):427--431

  20. [28]

    Dickinson, J. (1973). Some statistical results in the combination of forecasts. Journal of the Operational Research Society , 24(2):253--260

  21. [29]

    Dickinson, J. (1975). Some comments on the combination of forecasts. Journal of the Operational Research Society , 26(1):205--210

  22. [30]

    Diebold, F. X. (2015). Comparing predictive accuracy, twenty years later: A personal perspective on the use and abuse of diebold--mariano tests. Journal of Business & Economic Statistics , 33(1):1--1

  23. [31]

    Diebold, F. X. and Mariano, R. S. (1995). Comparing predictive accuracy. Journal of Business & Economic Statistics , 13(3)

  24. [32]

    Dietterich, T. G. (2000). Ensemble methods in machine learning. In International workshop on multiple classifier systems , pages 1--15. Springer

  25. [33]

    Dimopoulos, Y., Bourret, P., and Lek, S. (1995). Use of some sensitivity criteria for choosing networks with good generalization ability. Neural Processing Letters , 2:1--4

  26. [34]

    Du, W. (2023). Pypots: A python toolbox for data mining on partially-observed time series. arXiv preprint arXiv:2305.18811

  27. [35]

    Dudek, G. (2016). Multilayer perceptron for gefcom2014 probabilistic electricity price forecasting. International Journal of Forecasting , 32(3):1057--1060

  28. [36]

    Introduction to the national electricity market of singapore

    EMA (2009). Introduction to the national electricity market of singapore. Energy Market Authority, Singapore

  29. [37]

    Nems market report 2018

    EMC (2018). Nems market report 2018. Energy Market Company, Singapore

  30. [38]

    Nems market report 2023

    EMC (2023). Nems market report 2023. Energy Market Company, Singapore

  31. [39]

    Electricity prices

    Energy Market Authority of Singapore (2024). Electricity prices. Accessed: 2024-11-13

  32. [40]

    Fama, E. F. and Schwert, G. W. (1977). Asset returns and inflation. Journal of financial economics , 5(2):115--146

  33. [41]

    E., Schmidt, L., and Timmermann, A

    Farmer, L. E., Schmidt, L., and Timmermann, A. (2023). Pockets of predictability. The Journal of Finance , 78(3):1279--1341

  34. [42]

    Filippou, I., Li, J., Liu, X., and Taylor, M. P. (2024a). Short-horizon currency expectations. Available at SSRN 4832426

  35. [43]

    Filippou, I., Rapach, D., and Thimsen, C. (2024b). Cryptocurrency return predictability: A machine-learning analysis. Available at SSRN 3914414

  36. [44]

    Y., Tan, J

    Foo, B., Koh, D. Y., Tan, J. P., and Wang, W. (2023). Forecasting singapore’s economy using statistical learning and factor models. The Singapore Economic Review , 68(02):319--353

  37. [45]

    Ghasemi, A., Shayeghi, H., Moradzadeh, M., and Nooshyar, M. (2016). A novel hybrid algorithm for electricity price and load forecasting in smart grids with demand-side management. Applied energy , 177:40--59

  38. [46]

    G., and Rafizadeh, N

    Ghoddusi, H., Creamer, G. G., and Rafizadeh, N. (2019). Machine learning in energy economics and finance: A review. Energy Economics , 81:709--727

  39. [47]

    and White, H

    Giacomini, R. and White, H. (2006). Tests of conditional predictive ability. Econometrica , 74(6):1545--1578

  40. [48]

    Goulet Coulombe, P., Leroux, M., Stevanovic, D., and Surprenant, S. (2022). How is machine learning useful for macroeconomic forecasting? Journal of Applied Econometrics , 37(5):920--964

  41. [49]

    Goyal, A., Welch, I., and Zafirov, A. (2024). A comprehensive 2022 look at the empirical performance of equity premium prediction. The Review of Financial Studies , 37(11):3490--3557

  42. [50]

    R., and Zhang, X

    Green, J., Hand, J. R., and Zhang, X. F. (2017). The characteristics that provide independent information about average us monthly stock returns. The Review of Financial Studies , 30(12):4389--4436

  43. [51]

    Gu, S., Kelly, B., and Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies , 33(5):2223--2273

  44. [52]

    Hamilton, J. D. (1994). Time Series Analysis . Princeton University Press

  45. [53]

    Hansen, L. K. and Salamon, P. (1990). Neural network ensembles. IEEE transactions on pattern analysis and machine intelligence , 12(10):993--1001

  46. [54]

    R., Lunde, A., and Nason, J

    Hansen, P. R., Lunde, A., and Nason, J. M. (2011). The model confidence set. Econometrica , 79(2):453--497

  47. [55]

    R., Liu, Y., and Zhu, H

    Harvey, C. R., Liu, Y., and Zhu, H. (2016). … and the cross-section of expected returns. The Review of Financial Studies , 29(1):5--68

  48. [56]

    J., Martin, J

    Henkel, S. J., Martin, J. S., and Nardari, F. (2011). Time-varying short-horizon predictability. Journal of financial economics , 99(3):560--580

  49. [57]

    Hong, Y., Jiang, F., Meng, L., and Xue, B. (2024). Forecasting inflation using economic narratives. Journal of Business & Economic Statistics , (just-accepted):1--27

  50. [58]

    Hornik, K., Stinchcombe, M., and White, H. (1989). Multilayer feedforward networks are universal approximators. Neural networks , 2(5):359--366

  51. [59]

    and Wang, C

    Hu, T. and Wang, C. (2022). The impact of optimally dispatched energy storage devices on electricity price volatility. International Journal of Electrical Power & Energy Systems , 137:107810

  52. [60]

    Huang, D., Jiang, F., Tu, J., and Zhou, G. (2015). Investor sentiment aligned: A powerful predictor of stock returns. The Review of Financial Studies , 28(3):791--837

  53. [61]

    Ioffe, S. (2015). Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167

  54. [62]

    and Ma, T

    Jagannathan, R. and Ma, T. (2003). Risk reduction in large portfolios: Why imposing the wrong constraints helps. The journal of finance , 58(4):1651--1683

  55. [63]

    Johnsen, T. A. (2001). Demand, generation and price in the norwegian market for electric power. Energy Economics , 23(3):227--251

  56. [64]

    and Ziel, F

    Kath, C. and Ziel, F. (2018). The value of forecasts: Quantifying the economic gains of accurate quarter-hourly electricity price forecasts. Energy Economics , 76:411--423

  57. [65]

    Keles, D., Genoese, M., M \"o st, D., and Fichtner, W. (2012). Comparison of extended mean-reversion and time series models for electricity spot price simulation considering negative prices. Energy economics , 34(4):1012--1032

  58. [66]

    Kelly, B., Malamud, S., and Zhou, K. (2024). The virtue of complexity in return prediction. The Journal of Finance , 79(1):459--503

  59. [67]

    T., Pruitt, S., and Su, Y

    Kelly, B. T., Pruitt, S., and Su, Y. (2019). Characteristics are covariances: A unified model of risk and return. Journal of Financial Economics , 134(3):501--524

  60. [68]

    Killick, R., Fearnhead, P., and Eckley, I. A. (2012). Optimal detection of changepoints with a linear computational cost. Journal of the American Statistical Association , 107(500):1590--1598

  61. [69]

    Kingma, D. P. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

  62. [70]

    Knittel, C. R. and Roberts, M. R. (2005). An empirical examination of restructured electricity prices. Energy economics , 27(5):791--817

  63. [71]

    C., Schmidt, P., and Shin, Y

    Kwiatkowski, D., Phillips, P. C., Schmidt, P., and Shin, Y. (1992). Testing the null hypothesis of stationarity against the alternative of a unit root: How sure are we that economic time series have a unit root? Journal of econometrics , 54(1-3):159--178

  64. [72]

    Lo, A. W. and MacKinlay, A. C. (1988). Stock market prices do not follow random walks: Evidence from a simple specification test. The review of financial studies , 1(1):41--66

  65. [73]

    Mirakyan, A., Meyer-Renschhausen, M., and Koch, A. (2017). Composite forecasting approach, application for next-day electricity price forecasting. Energy Economics , 66:228--237

  66. [74]

    Nagel, S. (2012). Evaporating liquidity. The Review of Financial Studies , 25(7):2005--2039

  67. [75]

    Nowotarski, J., Raviv, E., Tr \"u ck, S., and Weron, R. (2014). An empirical comparison of alternative schemes for combining electricity spot price forecasts. Energy Economics , 46:395--412

  68. [76]

    G., Onumaegbu, C., Wilberforce, T., Ramadan, M., Abdelkareem, M

    Olabi, A. G., Onumaegbu, C., Wilberforce, T., Ramadan, M., Abdelkareem, M. A., and Al-Alami, A. H. (2021). Critical review of energy storage systems. Energy , 214:118987

  69. [77]

    Purchase options for business consumers

    Open Electricity Market (2024). Purchase options for business consumers. Accessed: 2024-11-10

  70. [78]

    Palm, F. C. and Zellner, A. (1992). To combine or not to combine? issues of combining forecasts. Journal of Forecasting , 11(8):687--701

  71. [79]

    Panapakidis, I. P. and Dagoumas, A. S. (2016). Day-ahead electricity price forecasting via the application of artificial neural network based models. Applied Energy , 172:132--151

  72. [80]

    Pettenuzzo, D., Timmermann, A., and Valkanov, R. (2014). Forecasting stock returns under economic constraints. Journal of Financial Economics , 114(3):517--553

  73. [81]

    Pettitt, A. N. (1979). A non-parametric approach to the change-point problem. Journal of the Royal Statistical Society: Series C (Applied Statistics) , 28(2):126--135

  74. [82]

    E., Strauss, J

    Rapach, D. E., Strauss, J. K., and Zhou, G. (2010). Out-of-sample equity premium prediction: Combination forecasts and links to the real economy. The Review of Financial Studies , 23(2):821--862

  75. [83]

    Ruibal, C. M. and Mazumdar, M. (2008). Forecasting the mean and the variance of electricity prices in deregulated markets. IEEE Transactions on Power Systems , 23(1):25--32

  76. [84]

    Sai, W., Pan, Z., Liu, S., Jiao, Z., Zhong, Z., Miao, B., and Chan, S. H. (2023). Event-driven forecasting of wholesale electricity price and frequency regulation price using machine learning algorithms. Applied Energy , 352:121989

  77. [85]

    Shapley, L. S. (1953). A value for n-person games. Contribution to the Theory of Games , 2

  78. [86]

    and Xiu, D

    Shen, Z. and Xiu, D. (2024). Can machines learn weak signals? University of Chicago, Becker Friedman Institute for Economics Working Paper , (2024-29)

  79. [87]

    and Wallis, K

    Smith, J. and Wallis, K. F. (2009). A simple explanation of the forecast combination puzzle. Oxford bulletin of economics and statistics , 71(3):331--355

  80. [88]

    F., Yu, J., and Yuan, Y

    Stambaugh, R. F., Yu, J., and Yuan, Y. (2012). The short of it: Investor sentiment and anomalies. Journal of financial economics , 104(2):288--302

  81. [89]

    Stock, J. H. and Watson, M. W. (2004). Combination forecasts of output growth in a seven-country data set. Journal of forecasting , 23(6):405--430

  82. [90]

    Stock, J. H. and Watson, M. W. (2006). Forecasting with many predictors. Handbook of economic forecasting , 1:515--554

  83. [91]

    Timmermann, A. (2006). Forecast combinations. Handbook of economic forecasting , 1:135--196

  84. [92]

    Van Der Stelt, S., AlSkaif, T., and Van Sark, W. (2018). Techno-economic analysis of household and community energy storage for residential prosumers with smart appliances. Applied Energy , 209:266--276

  85. [93]

    Wang, D., Luo, H., Grunder, O., Lin, Y., and Guo, H. (2017). Multi-step ahead electricity price forecasting using a hybrid model based on two-layer decomposition technique and bp neural network optimized by firefly algorithm. Applied Energy , 190:390--407

  86. [94]

    and Goyal, A

    Welch, I. and Goyal, A. (2008). A comprehensive look at the empirical performance of equity premium prediction. The Review of Financial Studies , 21(4):1455--1508

  87. [95]

    Weron, R. (2006). Modeling and forecasting electricity loads and prices: A statistical approach . John Wiley & Sons

  88. [96]

    Weron, R. (2014). Electricity price forecasting: A review of the state-of-the-art with a look into the future. International journal of forecasting , 30(4):1030--1081

  89. [97]

    and Zhou, G

    Yuan, M. and Zhou, G. (2024). Why na ve 1/n diversification is not so na ve, and how to beat it? Journal of Financial and Quantitative Analysis, forthcoming. This supplementary material to the paper contains five sections. In Section A, we present

  90. [98]

    and Steinert, R

    Ziel, F. and Steinert, R. (2016). Electricity price forecasting using sale and purchase curves: The x-model. Energy Economics , 59:435--454

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.