- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes HyperForecast, a long-horizon time series forecasting framework based on hyperbolic geometry. It first decomposes time series into trends, coarse-grained periodic patterns, fine-grained periodic patterns, and residuals using learnable cascaded convolutions. These components are then mapped into Poincaré hyperbolic space, where future values are predicted by estimating geodesic velocities. Hierarchical and temporal consistency losses are introduced to preserve the structural relationships among representations. The framework supports both autoregressive (AR) and non-autoregressive (NAR) forecasting, achieving competitive results on 9 out of 11 benchmark datasets. Ablation studies further demonstrate the benefits of hyperbolic geometry. The key idea is to leverage hyperbolic space to capture multiscale hierarchical relationships and thereby improve long-horizon time series forecasting.
Critical strengths
- The method is based on hyperbolic geometry, combining multiscale decomposition with manifold dynamics to explicitly model hierarchical relationships among components of different frequencies in hyperbolic space, which is interpretable.
- Develop AR/NAR forecasting mechanisms, using hierarchical and temporal consistency losses to preserve representational structure and improve forecasting stability.
- The results are promising. The model achieves SOTA in 9 out of 11 benchmarks.
Critical weaknesses
- Different time series may vary greatly. The assumption that components of different frequencies from a hierarchical structure may not hold for all time series. So the general advantages of leveraging hyperbolic geometry thoery requires further demonstration.
- The NAR approach relies on a global velocity estimate to predict future values, which may limit its ability to model complex nonlinear dynamics and regime shifts.
- The experiments focus primarily on conventional long-horizon forecasting benchmarks, providing insufficient evidence of generalization to more complex and highly non-stationary scenarios. So I suggest authors to include results for more complex, non-stationary datasets to address concerns (1) and (3).
Questions for authors
N/A
5 comments
Response to Reviewer 7GQp (Part 2/3)
W2. The NAR approach relies on a global velocity estimate to predict future values, which may limit its ability to model complex nonlinear dynamics and regime shifts.
Answer: Thank you for the point. NAR estimates one tangent velocity per component (four velocities, one per time scale) as an exponentially weighted average of the displacements in the lookback, so recent displacements dominate, and extrapolates it along a geodesic of the ball. Even this single-velocity path is not a uniform straight line in the ball's coordinates, as its steps are shorter near the boundary inside the ball. The per-horizon gate, the Poincaré FC layer, and the MLP decoder then bend the forecast further. The AR variant is the local alternative, which re-estimates the velocity after every predicted segment. If a single velocity limited the model, AR should gain on NAR as the horizon grows. As per Table 24, NAR is better on 29 of the 44 dataset-horizon pairs, and AR on 13 (2 ties), and the advantage of NAR grows with the horizon, so re-estimating the velocity step by step does not help. A likely reason is that AR feeds its own predictions back, so errors compound along the rollout, while NAR forecasts every segment from the observed lookback.
We agree that a single velocity per component cannot anticipate a regime change that begins after the forecast origin, which no input-only forecaster can, and that an abrupt level shift inside the lookback is the harder case for it. Two parts of the model control how it reacts to a shift. RevIN normalizes each input window by its own mean and standard deviation, so a level shift between the training and test periods is removed before the embedding. Within the lookback, the exponential weights use a fixed decay of 0.9 for all four components, and a smaller decay, or a shorter part of the lookback, makes the velocity follow the displacements after a shift sooner, at the cost of more noise. A regime-aware velocity is a natural extension that we leave for future work (line 523 of revised manuscript), for example by learning the decay per component, so the trend keeps a long memory and the residual a short one, or by combining a short-memory and a long-memory velocity with a gate learned from the lookback.
Response to Reviewer 7GQp (Part 3/3)
W3. The experiments focus primarily on conventional long-horizon forecasting benchmarks, providing insufficient evidence of generalization to more complex and highly non-stationary scenarios. So I suggest authors to include results for more complex, non-stationary datasets to address concerns (1) and (3).
Answer: Thank you for the suggestion. We address this with three analyses.
(i) Non-stationarity of the benchmarks. Task: long-term forecasting on the 11 benchmarks. For each dataset, the table below gives the share of channels that KPSS and ADF tests flag as non-stationary (1.00 = all channels), and the shift of the test-period mean from the training mean in training standard deviations (larger means more drift between training and test):
| Dataset | KPSS non-stationary | ADF non-stationary | Test mean shift (train SD) |
|---|---|---|---|
| Exchange | 1.00 | 0.88 | 1.39 |
| CzeLan | 0.36 | 0.18 | 0.73 |
| Weather | 0.76 | 0.00 | 0.42 |
| PM2.5 | 0.83 | 0.00 | 0.31 |
| Wind | 1.00 | 0.14 | 0.27 |
| Temp | 1.00 | 0.07 | 0.25 |
| METR-LA | 0.47 | 0.00 | 0.21 |
| Solar | 1.00 | 0.00 | 0.20 |
| ZafNoo | 0.91 | 0.09 | 0.17 |
| AQWan | 0.82 | 0.00 | 0.09 |
| AQShunyi | 0.45 | 0.00 | 0.08 |
The benchmarks differ widely in how much they drift between training and test. HyperForecast stays ahead of the best baselines on most of them, including the most drifting ones, so its gains are not limited to stationary data.
(ii) Missing values. Task: forecasting from corrupted inputs. At test time only, random blocks of 24 consecutive steps are removed from each channel until 10% (or 30%) of the steps are missing and filled with the last observed value; all models are trained on clean data and evaluated under the same corruption (H = 96, L = 720, MSE). The table below gives the clean MSE and the MSE under corruption with its relative increase.
| Dataset | Model | Original | 10% missing | 30% missing |
|---|---|---|---|---|
| AQShunyi | HyperForecast | 0.610 | 0.613 (+0.6%) | 0.621 (+1.9%) |
| TimeBase | 0.629 | 0.633 (+0.7%) | 0.643 (+2.2%) | |
| PhaseFormer | 0.644 | 0.648 (+0.7%) | 0.660 (+2.5%) | |
| AQWan | HyperForecast | 0.699 | 0.706 (+1.0%) | 0.708 (+1.3%) |
| TimeBase | 0.726 | 0.730 (+0.5%) | 0.739 (+1.8%) | |
| PhaseFormer | 0.743 | 0.748 (+0.7%) | 0.762 (+2.6%) | |
| Weather | HyperForecast | 0.146 | 0.147 (+0.9%) | 0.150 (+3.3%) |
| TimeBase | 0.145 | 0.146 (+1.1%) | 0.150 (+4.0%) | |
| PhaseFormer | 0.147 | 0.149 (+1.2%) | 0.154 (+4.4%) |
Missing values degrade HyperForecast no more than the baselines. Every model loses at most about 1% of its MSE at 10% missing and a few percent at 30%, and at 30% the loss is smallest for HyperForecast on all three datasets. Its MSE also stays the lowest, or on par with the best baseline.
(iii) More complex benchmarks. Thank you for the suggestion. We evaluated HyperForecast on the Web/CloudOps subsets of GIFT-Eval (Aksu et al., 2024), whose workload traces are highly non-stationary, with frequent spikes and abrupt shifts in intensity. We follow the GIFT-Eval protocol, with its short-term horizon of 48 steps, its train, validation, and test split, and the MSE of the forecast on the original scale. HyperForecast is trained on each dataset with the learning rate chosen on the GIFT-Eval validation window (similar evaluation protocol as PhaseFormer’s experiments on GIFT-Eval), while the foundation models Moirai-2 (Liu et al., 2025), Moirai-base (Woo et al., 2024), and Chronos-base (Ansari et al., 2024) are evaluated zero-shot. The values of the other models are those reported by PhaseFormer.
| Dataset | HyperForecast | PhaseFormer | Moirai-2 | Moirai-base | Chronos-base |
|---|---|---|---|---|---|
| bizitobs_l2c/5T | 16.22 | 39.43 | 20.62 | 22.10 | 24.10 |
| bizitobs_l2c/H | 65.07 | 99.96 | 67.85 | 272.00 | 243.00 |
| bitbrains_rnd/5T | 1,474,743 | 1,606,034 | 1,660,532 | 180,000,000 | 1,800,000 |
| bitbrains_rnd/H | 1,780,993 | 1,970,330 | 2,235,573 | 4,020,000 | 2,210,000 |
| bitbrains_fast_storage/5T | 1,725,546 | 1,887,398 | 2,142,125 | 25,400,000,000 | 1,920,000 |
| bitbrains_fast_storage/H | 2,709,691 | 2,585,180 | 2,942,651 | 604,000,000 | 2,620,000 |
HyperForecast has the lowest MSE on five of the six subsets, including the two bizitobs_l2c subsets where Moirai-2 is the best of the other models, and on bitbrains_fast_storage/H it is close behind PhaseFormer and Chronos-base. Its gains therefore carry over to these volatile, non-stationary workloads as well. In our revised manuscript, we added the three analyses as Appendix L (Tables 21 to 23), referenced in Section 5 (lines 377-379).
References
Aksu, Taha, et al. "Gift-eval: A benchmark for general time series forecasting model evaluation." arXiv preprint arXiv:2410.10393 (2024).
Liu, Chenghao, et al. "Moirai 2.0: When less is more for time series forecasting." arXiv preprint arXiv:2511.11698 (2025).
Woo, Gerald et al. "Unified Training of Universal Time Series Forecasting Transformers." Proceedings of the 41st International Conference on Machine Learning (2024).
Ansari, Abdul Fatir, et al. "Chronos: Learning the Language of Time Series." Transactions on Machine Learning Research (2024)
Response to Reviewer 7GQp (Part 1/3)
Answer: Thank you for the point. We agree that the strength of the frequency hierarchy differs between series and cannot be assumed for every one, so we measured it on the datasets directly, before any model is trained. For each dataset, we cut 24-step segments from the four components of the series (fixed moving-average cascade with the day, week, and two-week kernels), pooled them into one point cloud, and computed its relative Gromov-δ (Euclidean distances, scaled by their 99th percentile). As a reference, the table below also gives the δ of Gaussian noise with the same mean and covariance as the data, which has the same spread but no hierarchy. A lower δ than this reference means the components are more tree-like than their spread alone would produce:
On all eleven datasets, the pooled components are more tree-like than Gaussian noise with the same spread, so the structure is present in the data before any model is applied. The trained embeddings in Table 2 have δ between 0.002 and 0.023, a similar range. So the Poincaré model preserves this structure; it does not create it.
The benefit also comes from the exponential volume growth of the Poincaré ball, which leaves more room at larger radii for the higher-frequency components, whether or not the data is strictly from a tree. The volume of the Poincaré ball grows exponentially with the radius, because its metric factor λx=1−c∥x∥22 grows toward the boundary, so the higher-frequency components at larger radii have more room than in a Euclidean space of the same dimension, whose volume grows only polynomially. The forecast uses the same geometry: the geodesic step expzN(t⋅vˉ) covers the same distance in the ball's own geometry at every step, so in the ball's coordinates the step shrinks at larger radii and the state stays inside the ball for every t, whereas a Euclidean step zN+t⋅vˉ has the same length at every radius. In our revised manuscript, Section 5 (lines 428-430) now points to the δ of the decomposed data in Appendix J, and Section 4.5 (lines 315-319) adds the geometric argument.