Pith. sign in

REVIEW 7 minor 63 references

Coherent forecast combination for linearly constrained multiple time series

T0 review · 0 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves a closed-form minimum-variance rule that combines multiple experts' forecasts while enforcing linear constraints, in one step.

desk verdict A correct and useful unification of forecast combination and reconciliation for unbalanced panels, with honest empirics; the known-W caveat is real, explicit, and handled in standard ways. read the letter →

arxiv 2412.03429 v1 pith:BEJ3RT6O submitted 2024-12-04 stat.ME stat.APstat.CO

Daniele Girolimetto , Tommaso Di Fonzo This is my paper · ORCID
classification stat.MEstat.APstat.CO MSC 62M1062H12
keywords coherentforecastcombinationreconciliationlinearlyconstrainedtimeseriesmultipleexpertsunbalancedpanelsminimummeansquareerrorhierarchicalelectricitygenerationforecasting
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 claims that forecast combination and forecast reconciliation for linearly constrained multiple time series should be solved as one optimization, not two. It proves that the best unbiased linear forecast that both pools base forecasts from several experts and forces the final vector to respect constraints like national-accounting or hierarchy additivity has a closed form, even when different experts cover different variables. If true, the result gives users an exact error-covariance formula and a single set of weights, replacing ad-hoc sequential combination-then-reconciliation or reconciliation-then-combination recipes. The authors report simulation and daily Australian electricity-generation evidence that the simultaneous solution improves over base forecasts, single-expert reconciliation, and simpler sequential combinations.

What carries the argument

The carrying mechanism is a two-stage linear map whose two pieces share one covariance matrix. First, the base forecasts are combined by the MMSE multi-task weight $\boldsymbol\Omega=\mathbf W^{-1}\mathbf K\mathbf W_c$, producing an incoherent combined vector with covariance $\mathbf W_c$. Then the oblique projector $\mathbf M=\mathbf I_n-\mathbf W_c\mathbf C^\top(\mathbf C\mathbf W_c\mathbf C^\top)^{-1}\mathbf C$ maps that vector onto the coherence subspace $\{\mathbf y:\mathbf C\mathbf y=\mathbf 0\}$. Because the projector is built from the same $\mathbf W_c$ that governs the combination, the reconciliation step is calibrated to the combination's own error distribution, which is what delivers the closed-form variance $\mathbf M\mathbf W_c$.

What would settle it

Generate replicate panels where the true base-forecast covariance $\mathbf W$ is known, estimate the coherent and competing combinations, and check whether the closed-form MMSE vector has the smallest mean squared error among all unbiased linear coherent estimators when data are drawn from the assumed model; any systematic violation would falsify Theorem 1, while using an estimated $\mathbf W$ would test the practical claim rather than the theoretical one.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central result is Theorem 1. Given $p$ unbiased forecast vectors $\hat{\mathbf y}_j$ assembled as $\hat{\mathbf y}=\mathbf K\mathbf y+\boldsymbol\varepsilon$ with known positive-definite error covariance $\mathbf W$, the minimum mean square error linear coherent combination is the solution to $\min_{\mathbf y}(\hat{\mathbf y}-\mathbf K\mathbf y)^\top\mathbf W^{-1}(\hat{\mathbf y}-\mathbf K\mathbf y)$ subject to $\mathbf C\mathbf y=\mathbf 0$, and it is $\hat{\mathbf y}_c=\mathbf M\boldsymbol\Omega^\top\hat{\mathbf y}$, with $\mathbf M=\mathbf I_n-\mathbf W_c\mathbf C^\top(\mathbf C\mathbf W_c\mathbf C^\top)^{-1}\mathbf C$, $\boldsymbol\Omega=\mathbf W^{-1}\mathbf K\mathbf W_c$, and $\mathbf W_c=(\mathbf K^\top\mathbf W^{-1}\mathbf K)^{-1}$. Its error covariance is $\mathbf M\mathbf W_c$, and for each expert the coherent forecast is never worse in covariance order than the combined or base forecast. The formula covers unbalanced panels by encoding which variables each expert forecasts in selection matrices $\mathbf L_j$.

Load-bearing premise

The load-bearing premise is that the base forecasts are unbiased and that their error covariance matrix $\mathbf W$ is known; in practice $\mathbf W$ is estimated from in-sample residuals, and estimation error in $\mathbf W$ is outside the theoretical guarantees.

Editorial extensions

If this is right

  • Any set of linear equality constraints—hierarchy sums, GDP accounting identities, energy balances—can be imposed at the same time as the forecasts are pooled, so the user never has to reconcile a pre-combined forecast or combine pre-reconciled forecasts.
  • The error-covariance ranking $L_j\hat{\mathbf W}_cL_j^\top \preceq L_j\mathbf W_cL_j^\top \preceq \mathbf W_j$ is a direct corollary: under known $\mathbf W$, adding coherence does not inflate forecast-error covariance relative to either the combination or the base forecasts.
  • Unbalanced panels are handled by selection matrices $L_j$, so the method applies when experts forecast different subsets of variables, a case the simple sequential reconciliation-then-combination approach cannot cover.
  • When errors are uncorrelated across variables, the combination step reduces to the classical inverse-variance single-variable weights, while the projection step still lets every variable's forecasts influence the others.
  • In the reported simulation and in the Australian electricity-generation experiment, the simultaneous optimal coherent combination frequently attains the best average relative MAE and MSE among the compared approaches.

Reading between the lines

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

  • A natural extension is to treat $\mathbf W$ as uncertain and shrink the coherent weights themselves rather than only the covariance estimate; the paper does not derive such a correction, but its exact formula for $\hat{\mathbf W}_c$ would allow a delta-method or bootstrap assessment of weight-instability losses.
  • The same projection structure could be reused for probabilistic forecasting by reconciling quantile or simulation draws through the linear map $\mathbf M\boldsymbol\Omega^\top$, though the paper only treats point forecasts.
  • Because the formula separates combination and projection, it can be adapted to settings where some experts are trusted only for sub-vectors, such as nowcasting GDP with mixed-frequency data sources; this would be a testable extension of the unbalanced-panel machinery.
  • The empirical advantage over equal weights may shrink when the covariance estimate is noisy, so a fair stress-test is to compare the method on short training samples where shrinkage intensity is high.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 7 minor

Summary. The paper develops a framework for coherent forecast combination in linearly constrained multiple time series, handling unbalanced panels where experts may forecast different subsets of variables. The authors formulate the problem as a constrained quadratic program (17) and derive closed-form expressions for the optimal coherent combined forecast vector and its error covariance matrix in Theorem 1 and Corollary 1. They also present equivalent structural and by-variable formulations, discuss shrinkage-based estimators for the covariance matrix W, and validate the method through simulations and an application to Australian electricity generation data.

Significance. The main theoretical result is clean and useful: it unifies linear forecast combination and reconciliation in a simultaneous GLS-based framework with an explicit projection interpretation. I verified the algebra in Appendix A; Theorem 1 and Corollary 1 are correct under the stated assumptions of unbiased base forecasts and known positive-definite W. The paper is also commendable for providing reproducible implementations via R packages and a GitHub repository, and for a thorough empirical comparison including DM and MCS tests. The principal caveat, clearly acknowledged by the authors, is that all optimality claims are conditional on a known W; the feasible version with estimated W is evaluated empirically but its theoretical properties are not analyzed. This is a limitation rather than a flaw, and it does not undermine the central derivation.

minor comments (7)
  1. [Appendix D.3 (proof of Theorem 2)] The first-order condition system displayed in the proof of Theorem 2 states 'GS - Inb = 0' as the second equation, but the constraint being imposed is 'GKS = Inb' as written in (D.14). The subsequent algebra uses the correct constraint, so this appears to be a typographical error; please correct it to avoid confusion.
  2. [Section 5.2, equation (29)] In the shrinkage formula (29), the Hadamard product 'In ⊙ cW' requires the identity matrix to have the same dimension as cW, which is m × m. As written, the dimensions do not conform. The same issue appears in the formulas for cWj,shr and bΣi,shr; please use I_m, I_nj, and I_pi respectively.
  3. [Section 3.1, expression (12)] The index range in (12) reads 'i = 1, . . . , nj, i = 1, . . . , pi', which is ambiguous; the second range should refer to the expert index j, i.e., 'j = 1, . . . , p' (or 'j = 1, . . . , pi' as appropriate).
  4. [Appendix F, Figure F.5 caption] The caption states that the figure uses 'MSE (first row) and MAE (second row)', but the panel labels shown in the figure indicate MAE in the first row and MSE in the second row; please align the caption with the actual panels.
  5. [Section 5.2, multi-step covariance assumption] The assumption 'cWh ∝ cW' for h > 1 is introduced without comment. Since the electricity application uses horizons 1–7, it would be helpful to state explicitly that this is a common working approximation in forecast reconciliation (e.g., Wickramasuriya et al., 2019) and to note in Section 8 that the proportionality constant is not estimated.
  6. [Section 6, discussion of Table 6] The sentence 'as the number of residuals used to estimate the error covariance matrices grows, the quality of forecast improves' could be made more precise by tying it to the AvgRelMAE metric and the specific N values in Table 6.
  7. [Section 8, limitations] The known-W limitation is acknowledged, but since the simulations and application use feasible estimators, a brief statement about how estimation error in W might affect the ranking of methods would help practitioners; this is not required for the theoretical result but would improve practical guidance.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: Theorem 1 is a direct constrained-GLS solution, and the known-W and unbiasedness assumptions are explicit inputs, not fitted predictions.

full rationale

The paper's central claim is Theorem 1 (Section 4): the solution of min_y (by - Ky)^T W^-1 (by - Ky) subject to Cy = 0 is eyc = M Omega^T by. Appendix A derives this by solving the KKT system of the constrained quadratic program; the matrices M = I - Wc C^T (C Wc C^T)^-1 C, Omega = W^-1 K Wc, and Wc = (K^T W^-1 K)^-1 follow algebraically from the first-order conditions, and Corollary 1's covariance formula fWc = M Wc follows from eyc - y = M Omega^T epsilon. The unbiasedness proof uses only Omega^T K = I and C mu = 0. Thus the theorem does not reduce to its own inputs: W is an assumed covariance, not a fitted parameter, and no reported forecast is a renaming of fitted values. Section 8 explicitly acknowledges the two assumptions of unbiased base forecasts and known error covariance, and Section 5.2 gives the standard feasible plug-in version based on in-sample residuals with shrinkage. The simulation and electricity application evaluate this feasible version out-of-sample against equal-weight, MinT reconciliation, and sequential combination benchmarks, so the central claim has independent empirical content. The only self-citation is Girolimetto and Di Fonzo (2024c) for the structural-like representation y = Sb of general linearly constrained time series; this is a parameter-free linear algebra statement that can be verified directly from the zero-constraints matrix C, and the main proof works directly with C, so the citation is not load-bearing. The use of the authors' R packages FoReco and FoCo2 is implementation tooling rather than evidence. No circular step was identified.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

All non-derived inputs are listed; none are fitted to make the main theorem work. The main practical parameter is the forecast error covariance matrix W, which is assumed known in the theorem but must be estimated in applications. The structural-form equivalence comes from prior work by the same authors. No invented entities.

free parameters (2)
  • Forecast error covariance matrix W and its estimator variants = Estimated from in-sample forecast errors; shrunk and block-diagonal versions in Eq. (29)
    The weights Omega = W^-1 K Wc depend on W. The theory assumes W known; all applications estimate it from residuals, with shrinkage intensity fitted to data. Performance differs by estimator (occbe vs occbv vs occshr vs occwls).
  • Shrinkage intensities lambda_j and nu_i = Estimated via Schafer-Strimmer shrinkage (Section 5.2)
    Used to construct feasible cWshr and block-diagonal shrunk matrices; these data-driven constants affect the weights and the empirical rankings.
assumptions (5)
  • domain assumption Base forecasts are unbiased: E(by_j^i) = E(y_i), with zero-mean errors epsilon_j^i.
    Equation (12) states the unbiasedness condition; it is used for E(yc) = mu and for the optimality within the class of unbiased linear estimators. Violations are simulated in Setting 6 but not modeled.
  • domain assumption The error covariance matrix W is known and positive definite.
    Sections 4 and 8 state that the MMSE result assumes W known. In practice W is estimated, so finite-sample optimality is not guaranteed.
  • domain assumption The constraint matrix C has full row rank and constraints are consistent and non-redundant.
    Footnoted in Section 2.1; needed for C Wc C^T to be invertible and for the projection M to be well defined.
  • domain assumption Every general linearly constrained time series admits a structural representation y = S b.
    Invoked in Section 2.1 and Appendix D, taken from the authors' prior result Girolimetto and Di Fonzo (2024c). This is a load-bearing self-citation for the generality claim, though the zero-constrained form does not need it.
  • ad hoc to paper For horizons h > 1, the multi-step error covariance W_h is proportional to the one-step W.
    Section 5.2 assumes W_h proportional to W following Wickramasuriya et al. (2019); this is used for the h = 1 to 7 electricity results without testing the proportionality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Coherent forecast combination for linearly constrained multiple time series." pith.science (2026). https://pith.science/paper/BEJ3RT6O

@misc{pith2026241203429,
  author       = {Pith},
  title        = {Pith review of: Coherent forecast combination for linearly constrained multiple time series},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BEJ3RT6O}},
  note         = {Machine review of arXiv:2412.03429}
}
read the original abstract

Linearly constrained multiple time series may be encountered in many practical contexts, such as the National Accounts (e.g., GDP disaggregated by Income, Expenditure and Output), and multilevel frameworks where the variables are organized according to hierarchies or groupings, like the total energy consumption of a country disaggregated by region and energy sources. In these cases, when multiple incoherent base forecasts for each individual variable are available, a forecast combination-and-reconciliation approach, that we call coherent forecast combination, may be used to improve the accuracy of the base forecasts and achieve coherence in the final result. In this paper, we develop an optimization-based technique that combines multiple unbiased base forecasts while assuring the constraints valid for the series. We present closed form expressions for the coherent combined forecast vector and its error covariance matrix in the general case where a different number of forecasts is available for each variable. We also discuss practical issues related to the covariance matrix that is part of the optimal solution. Through simulations and a forecasting experiment on the daily Australian electricity generation hierarchical time series, we show that the proposed methodology, in addition to adhering to sound statistical principles, may yield in significant improvement on base forecasts, single-task combination and single-expert reconciliation approaches as well.

Figures

Figures reproduced from arXiv: 2412.03429 by the authors.

Figure 1
Figure 1. Two examples of linearly constrained time series. Left: a three-level genuine hierarchical structure. Right: a general linearly constrained multiple time series consisting of two simple hierarchies sharing the same top-level series. with nu = 3, nb = 4, ut = [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Australian daily electricity generation hierarchy (left) and the corresponding linear combination matrix (right), mapping 15 bottom variables into 8 upper variables. Red color denotes bottom variables that - when aggregated - are subtracted instead of added. A ∈ {−1, 0, 1} 8×15 for the whole dataset of 23 time series ( [PITH_FULL_IMAGE:figures/full_fig_p023_2.png] view at source ↗
Figure 3
Figure 3. Australian daily electricity generation time series. approaches and five coherent combination procedures (see [PITH_FULL_IMAGE:figures/full_fig_p024_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Pairwise DM-test results for the Australian electricity generation dataset, evaluated using absolute loss (top panels) and quadratic loss (bottom panel) across different forecast horizons. The left panel corresponds to forecast horizon h = 1, while the right panel is f…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 30 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry bracket.state close.brackets = "]" * 'skip if add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry bracket.state close.brackets = "]" * 'skip if add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not...

    work page

  3. [3]

    & Narita, F

    Ando, S. & Narita, F. (2024). https://doi.org/10.3390/forecast6020025 An alternative proof of Minimum Trace reconciliation . Forecasting , 6 (2), 456--461

    work page doi:10.3390/forecast6020025 2024

  4. [4]

    A., & Hyndman, R

    Athanasopoulos, G., Ahmed, R. A., & Hyndman, R. (2009). https://doi.org/10.1016/j.ijforecast.2008.07.004 Hierarchical forecasts for Australian domestic tourism . International Journal of Forecasting , 25 (1), 146--166

    doi:10.1016/j.ijforecast.2008.07.004 2009

  5. [5]

    J., Kourentzes, N., & Panagiotelis, A

    Athanasopoulos, G., Hyndman, R. J., Kourentzes, N., & Panagiotelis, A. (2024). https://doi.org/10.1016/j.ijforecast.2023.10.010 Forecast reconciliation: A review . International Journal of Forecasting , 40 (2), 430--456

    doi:10.1016/j.ijforecast.2023.10.010 2024

  6. [6]

    Baltagi, B. H. (2021). Econometric Analysis of Panel Data . New York: Springer

    work page 2021

  7. [7]

    & Granger, C

    Bates, J. & Granger, C. W. (1969). The combination of forecasts https://doi.org/10.1057/jors.1969.103. Operations Research Quarterly , 20 (4), 451--468

    doi:10.1057/jors.1969.103 1969

  8. [8]

    Bańbura, M., Giannone, D., & Reichlin, L. (2010). Large bayesian vector auto regressions https://doi.org/10.1002/jae.1137. Journal of Applied Econometrics , 25 (1), 71--92

    doi:10.1002/jae.1137 2010

Show all 63 references
  1. [9]

    W., & Hyndman, R

    Ben Taieb, S., Taylor, J. W., & Hyndman, R. J. (2021). Hierarchical probabilistic forecasting of electricity demand with smart meter data https://doi.org/10.1080/01621459.2020.1736081. Journal of the American Statistical Association , 116 (533), 27--43

    2021

  2. [10]

    Box, G. E. P. & Jenkins, G. M. (1976). Time series analysis: Forecasting and control . San Francisco: Holden-Day

    1976

  3. [11]

    Byron, R. P. (1978). The estimation of large social account matrices https://doi.org/10.2307/2344807. Journal of the Royal Statistical Society. Series A , 141 (3), 359--367

    1978 doi

  4. [12]

    Byron, R. P. (1979). Corrigenda: The estimation of large social account matrices https://doi.org/10.2307/2982515. Journal of the Royal Statistical Society. Series A , 142 (3), 405

    1979 doi

  5. [13]

    & Timmermann, A

    Capistr\'an, C. & Timmermann, A. (2009). Forecast combination with entry and exit of experts https://doi.org/10.1198/jbes.2009.07211. Journal of Business & Economic Statistics , 27 (4), 428--440

    2009 arXiv

  6. [14]

    R., Vasnev, A

    Claeskens, G., Magnus, J. R., Vasnev, A. L., & Wang, W. (2016). The forecast combination puzzle: A simple theoretical explanation https://doi.org/10.1016/j.ijforecast.2015.12.005. International Journal of Forecasting , 32 (3), 754--762

    2016 doi

  7. [15]

    Clemen, R. T. (1989). https://doi.org/10.1016/0169-2070(89)90012-5 Combining forecasts: A review and annotated bibliography . International Journal of Forecasting , 5 (4), 559--583

    1989 doi

  8. [16]

    Cleveland, R. (1990). STL : A seasonal-trend decomposition procedure based on loess. Journal of Official Statistics , 6 (1), 3--73

    1990

  9. [17]

    Conflitti, C., De Mol, C., & Giannone, D. (2015). Optimal combination of survey forecasts https://doi.org/10.1016/j.ijforecast.2015.03.009. International Journal of Forecasting , 31 (4), 1096--1103

    2015 doi

  10. [18]

    & Fildes, R

    Davydenko, A. & Fildes, R. (2013). Measuring forecasting accuracy: The case of judgmental adjustments to sku-level demand forecasts https://doi.org/10.1016/j.ijforecast.2012.09.002. International Journal of Forecasting , 29 (3), 510--522

    2013 doi

  11. [19]

    M., Hyndman, R

    De Livera, A. M., Hyndman, R. J., & Snyder, R. D. (2011). Forecasting time series with complex seasonal patterns using exponential smoothing https://doi.org/10.1198/jasa.2011.tm09771. Journal of the American Statistical Association , 106 (496), 1513--1527

    2011 doi

  12. [20]

    & Girolimetto, D

    Di Fonzo, T. & Girolimetto, D. (2023). Cross-temporal forecast reconciliation: Optimal combination method and heuristic alternatives https://doi.org/10.1016/j.ijforecast.2021.08.004. International Journal of Forecasting , 39 (1), 39--57

    2023 doi

  13. [21]

    & Girolimetto, D

    Di Fonzo, T. & Girolimetto, D. (2024). Forecast combination based forecast reconciliation https://doi.org/10.1016/j.ijforecast.2022.07.001. International Journal of Forecasting , 40 (2), 490--514

    2024 doi

  14. [22]

    Diebold, F. X. & Mariano, R. S. (1995). Comparing predictive accuracy. Journal of Business & Economic Statistics , 13 (3), 253--263

    1995

  15. [23]

    Fleming, P. J. & Wallace, J. J. (1986). How not to lie with statistics: The correct way to summarize benchmark results https://doi.org/10.1145/5666.5673. Communications of the ACM , 29 (3), 218--221

    1986

  16. [24]

    T., Covey, R., Martin, G

    Frazier, D. T., Covey, R., Martin, G. M., & Poskitt, D. (2023). Solving the forecast combination puzzle https://doi.org/10.48550/arXiv.2308.05263. Monash Econometrics and Business Statistics Working Papers 18/23, Monash University, Department of Econometrics and Business Statistics

    2023 doi arXiv

  17. [25]

    Girolimetto, D., Athanasopoulos, G., Di Fonzo, T., & Hyndman, R. J. (2024). Cross-temporal probabilistic forecast reconciliation: Methodological and practical issues https://doi.org/10.1016/j.ijforecast.2023.10.003. International Journal of Forecasting , 40 (3), 1134--1151

    2024 doi

  18. [26]

    & Di Fonzo, T

    Girolimetto, D. & Di Fonzo, T. (2024a). https://doi.org/10.32614/cran.package.foreco FoReco : Forecast reconciliation . R package

    2024 doi

  19. [27]

    & Di Fonzo, T

    Girolimetto, D. & Di Fonzo, T. (2024b). Foco2: Coherent forecast combination https://github.com/danigiro/FoCo2. R package

    2024

  20. [28]

    & Di Fonzo, T

    Girolimetto, D. & Di Fonzo, T. (2024c). Point and probabilistic forecast reconciliation for general linearly constrained multiple time series https://doi.org/10.1007/s10260-023-00738-6. Statistical Methods & Applications , 33 (2), 581--607

    2024 doi

  21. [29]

    Goehry, B., Goude, Y., Massart, P., & Poggi, J.-M. (2020). Aggregation of multi-scale experts for bottom-up load forecasting https://doi.org/10.1109/TSG.2019.2945088. IEEE Transactions on Smart Grid , 11 (3), 1895--1904

    2020

  22. [30]

    R., Lunde, A., & Nason, J

    Hansen, P. R., Lunde, A., & Nason, J. M. (2011). https://doi.org/10.3982/ECTA5771 The M odel C onfidence S et . Econometrica , 79 (2), 453--497

    2011 doi

  23. [31]

    Hollyman, R., Petropoulos, F., & Tipping, M. E. (2021). Understanding forecast reconciliation https://doi.org/10.1016/j.ejor.2021.01.017. European Journal of Operational Research , 294 (1), 149--160

    2021 doi

  24. [32]

    J., Ahmed, R

    Hyndman, R. J., Ahmed, R. A., Athanasopoulos, G., & Shang, H. L. (2011). https://doi.org/10.1016/j.csda.2011.03.006 Optimal combination forecasts for hierarchical time series . Computational Statistics and Data Analysis , 55 (9), 2579--2589

    2011 doi

  25. [33]

    J., et al

    Hyndman, R. J., et al. (2023). forecast: Forecasting functions for time series and linear models https://doi.org/10.32614/CRAN.package.forecast. R package

    2023 doi

  26. [34]

    Hyndman, R. J. & Khandakar, Y. (2008). https://doi.org/10.18637/jss.v027.i03 Automatic time series forecasting: The forecast package for R . Journal of Statistical Software , 27 , 1--22

    2008 doi

  27. [35]

    J., Lee, A

    Hyndman, R. J., Lee, A. J., & Wang, E. (2016). Fast computation of reconciled forecasts for hierarchical and grouped time series https://doi.org/10.1016/j.csda.2015.11.007. Computational Statistics & Data Analysis , 97 , 16--32

    2016 doi

  28. [36]

    Kapetanios, G., Labhard, V., & Price, S. (2008). https://www.jstor.org/stable/27638959 Forecasting using bayesian and information-theoretic model averaging: An application to UK inflation . Journal of Business & Economic Statistics , 26 (1), 33--41

    2008

  29. [37]

    & Olkin, I

    Keller, T. & Olkin, I. (2004). Combining correlated unbiased estimators of the mean of a normal distribution. Lecture Notes - Monograph Series , 45 , 218--227

    2004

  30. [38]

    J., Franses, P

    Koning, A. J., Franses, P. H., Hibon, M., & Stekler, H. O. (2005). https://doi.org/10.1016/j.ijforecast.2004.10.003 The M3 competition: Statistical tests of the results . International Journal of Forecasting , 21 (3), 397--409

    2005 doi

  31. [39]

    & Athanasopoulos, G

    Kourentzes, N. & Athanasopoulos, G. (2019). https://doi.org/10.1016/j.annals.2019.02.001 Cross-temporal coherent forecasts for A ustralian tourism . Annals of Tourism Research , 75 , 393--409

    2019 doi

  32. [40]

    & Rochet, P

    Lavancier, F. & Rochet, P. (2016). A general procedure to combine estimators https://doi.org/10.1016/j.csda.2015.08.001. Computational Statistics and Data Analysis , 94 , 175--192

    2016 doi

  33. [41]

    Liu, L., Hao, X., & Wang, Y. (2024). Solving the forecast combination puzzle using double shrinkages https://doi.org/10.1111/obes.12590. Oxford Bulletin of Economics & Statistics , 86 (3), 714--741

    2024 doi

  34. [42]

    Magnus, J. R. & Neudecker, H. (2019). Matrix differential calculus with applications in statistics and econometrics . Hoboken, NJ: Wiley, third edition

    2019

  35. [43]

    Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2022). https://doi.org/10.1016/j.ijforecast.2021.11.013 M5 accuracy competition: Results, findings, and conclusions . International Journal of Forecasting , 38 (4), 1346--1364

    2022 doi

  36. [44]

    Mohamed, R. A. (2023). Enhancing forecast accuracy using combination methods for the hierarchical time series approach https://doi.org/10.1371/journal.pone.0287897. PLoS One , 18 (7), e0287897

    2023 doi

  37. [45]

    & Granger, C

    Newbold, P. & Granger, C. W. (1974). Experience with forecasting univariate time series and the combination of forecasts https://doi.org/10.2307/2344546. Journal of the Royal Statistical Society, A , 137 (2), 131--146

    1974 doi

  38. [46]

    Panagiotelis, A., Gamakumara, P., Athanasopoulos, G., & Hyndman, R. J. (2021). https://doi.org/10.1016/j.ijforecast.2020.06.004 Forecast reconciliation: A geometric view with new insights on bias correction . International Journal of Forecasting , 37 (1), 343--359

    2021 doi

  39. [47]

    Panagiotelis, A., Gamakumara, P., Athanasopoulos, G., & Hyndman, R. J. (2023). Probabilistic forecast reconciliation: Properties, evaluation and score optimisation https://doi.org/10.1016/j.ejor.2022.07.040. European Journal of Operational Research , 36 (2), 693--706

    2023 doi

  40. [48]

    Porrill, J. (1988). Optimal combination and constraints for geometrical sensor data https://doi.org/10.1177/027836498800700606. The International Journal of Robotics Research , 7 (6)

    1988 doi

  41. [49]

    A., Cheng, G., & Yang, Y

    Qian, W., Rolling, C. A., Cheng, G., & Yang, Y. (2019). On the forecast combination puzzle. Econometrics , 7 (3), 39

    2019

  42. [50]

    Rao, C. R. (2000). Statistical proofs of some matrix inequalities https://doi.org/10.1016/S0024-3795(99)00276-1. Linear Algebra and its Applications , 321 (1--3), 307--320

    2000 doi

  43. [51]

    & Hyndman, R

    Rostami-Tabar, B. & Hyndman, R. J. (2024). Hierarchical time series forecasting in emergency medical services https://doi.org/10.1177/10946705241232169. Journal of Service Research , in press

    2024 doi

  44. [52]

    & Strimmer, K

    Sch \" a fer, J. & Strimmer, K. (2005). https://doi.org/10.2202/1544-6115.1175 A Shrinkage Approach to Large-Scale Covariance Matrix Estimation and Implications for Functional Genomics . Statistical Applications in Genetics and Molecular Biology , 4 (1)

    2005

  45. [53]

    Seber, G. (1984). Multivariate observations . New York: Wiley

    1984

  46. [54]

    & Wallis, K

    Smith, J. & Wallis, K. (2009). A simple explanation of the forecast combination puzzle https://doi.org/10.1111/j.1468-0084.2008.00541.x. Oxford Bulletin of Economics & Statistics , 71 (3), 331--355

    2009

  47. [55]

    Spiliotis, E., Petropoulos, F., & Assimakopoulos, V. (2019). Improving the forecasting performance of temporal hierarchies https://doi.org/10.1371/journal.pone.0223422. PLoS One , 14 (10), e0223422

    2019 doi

  48. [56]

    G., & Meade, J

    Stone, R., Champernowne, D. G., & Meade, J. E. (1942). The precision of national income estimates https://doi.org/10.2307/2967664. The Review of Economic Studies , 9 (2), 111--125

    1942 doi

  49. [57]

    & Deng, Z.-L

    Sun, S.-L. & Deng, Z.-L. (2004). https://doi.org/10.1016/j.automatica.2004.01.014 Multi-sensor optimal information fusion K alman filter . Automatica , 40 (6), 1017--1023

    2004 doi

  50. [58]

    Thompson, R., Qian, Y., & Vasnev, A. L. (2024). Flexible global forecast combinations https://doi.org/10.1016/j.omega.2024.103073. Omega , 126 , 103073

    2024

  51. [59]

    Timmermann, A. (2006). Forecast combinations . In G. Elliott, C. C. Granger, & A. Timmermann (Eds.), Handbook of Economic Forecasting (pp.\ 135--196). New York: Elsevier

    2006

  52. [60]

    J., Li, F., & Kang, Y

    Wang, X., Hyndman, R. J., Li, F., & Kang, Y. (2024). Forecast combinations: an over 50-year review https://doi.org/10.1016/j.ijforecast.2022.11.005. International Journal of Forecasting , 39 (4), 1518--1547

    2024 doi

  53. [61]

    L., Athanasopoulos, G., & Hyndman, R

    Wickramasuriya, S. L., Athanasopoulos, G., & Hyndman, R. J. (2019). Optimal forecast reconciliation for hierarchical and grouped time series through trace minimization https://doi.org/10.1080/01621459.2018.1448825. Journal of the American Statistical Association , 114 (526), 804--819

    2019

  54. [62]

    & Zhang, A

    Yang, D. & Zhang, A. N. (2019). Impact of information sharing and forecast combination on fast-moving-consumer-goods demand forecast accuracy https://doi.org/10.3390/info10080260. Information , 10 (8), 1167--1182

    2019 doi

  55. [63]

    Zhang, B., Panagiotelis, A., & Li, H. (2024). Constructing hierarchical time series through clustering: Is there an optimal way for forecasting? https://doi.org/10.1016/j.ijforecast.2024.10.002 International Journal of Forecasting , in press

    2024 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.