Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Foundation Time-Series AI Model for Realized Volatility Forecasting

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A time-series foundation model, fine-tuned on rolling chunks of recent data, delivers one-day-ahead volatility forecasts that statistically beat HAR, ARFIMA, CHAR, and RGARCH benchmarks.

desk verdict A useful first benchmark of TimesFM for realized volatility, but the claimed statistical outperformance rests on a misapplied DM/GW test and needs to be redone. read the letter →

arxiv 2505.11163 v1 pith:7ZYQGWW5 submitted 2025-05-16 q-fin.RM q-fin.ST

Anubha Goel , Puneet Pasricha , Martin Magris , Juho Kanniainen This is my paper · ORCID
classification q-fin.RMq-fin.ST MSC 62M1091B8468T07
keywords volatilityforecastingrealizedvariancefoundationmodelstimeseriesanalysisincrementallearningtransferTimesFMDiebold-Marianotest
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 is trying to establish that a general-purpose time-series foundation model, the TimesFM decoder-only transformer, can be turned into a superior one-day-ahead volatility forecaster by fine-tuning it incrementally on each index's own history. Tested on 21 major global equity indices between 2000 and 2021, the fine-tuned variants produce lower realized-variance forecast errors than the standard econometric benchmarks HAR, ARFIMA, CHAR, and RGARCH across most of six loss functions. The sharpest claim is statistical: under the asymmetric QLIKE loss, the 512-context pretrained and incrementally fine-tuned variants beat every econometric benchmark with Diebold-Mariano and Giacomini-White p-values below 0.10, while the log-transformed 64-context fine-tuned model is never statistically worse than any competitor. A sympathetic reader would care because the result suggests a single pretrained model, updated with lightweight retraining on recent data, could replace bespoke econometric machinery for a core risk-management task, and because the paper finds that zero-shot performance alone is only a reasonable baseline, with the adaptation step carrying the result.

What carries the argument

The load-bearing object is TimesFM v2.0, a decoder-only transformer — a neural network that reads a sequence of past observations and directly emits the next value — pretrained on over 6 billion real-world and synthetic time-series points, which the paper uses to produce one-day-ahead point forecasts of realized variance from a look-back context of 64, 128, or 512 daily observations, with no covariates and no feature engineering. The mechanism that carries the argument is the incremental fine-tuning protocol: the pretrained model is first adapted to the first 50% of an index's data with linear probing, keeping the transformer layers frozen and updating only the core layer, then the resulting checkpoint is fine-tuned again on the next 20% segment, and again on the segment after that, so the model inherits all previous learning while tracking the recent regime; the econometric benchmarks are re-estimated on the same rolling splits. The evidence is evaluated with six loss functions — MSE, MAE, MAPE, MDA, QLIKE, and sMAPE — and the statistical claims ride on Diebold-Mariano and Giacomini-White tests together with Model Confidence Set inclusion rates. The QLIKE loss is the hinge of the strongest claim, since it penalizes under-prediction of variance asymmetrically and is the loss relevant to risk management.

What would settle it

Re-run the comparison at the individual-stock level: for each of the 21 indices, form the daily QLIKE loss differential between $\mathrm{TFM512}_{IL}$ and HAR over the out-of-sample period and apply the Diebold-Mariano test with Newey-West standard errors to that daily series. If few stocks show significance, or if a pooled panel of daily loss differentials yields p-values above 0.10, then the claimed statistical outperformance is an artifact of testing on 21 averaged numbers rather than genuine evidence of superior predictive ability.

Watch

Extended reading notes

Core claim

The paper claims that TimesFM, a decoder-only transformer pretrained on billions of diverse time-series points, can match and often beat the standard econometric models of volatility forecasting after a modest adaptation step. The adaptation is an incremental fine-tuning scheme: the pretrained model is fine-tuned on the first half of each index's realized variance series, then repeatedly re-fine-tuned, checkpoint onward, on each successive block of around 20% of the data, so the model always chases the most recent regime. Out of sample, one day ahead, on 21 global equity indices from 2000 to 2021, the fine-tuned models achieve lower average losses than HAR, CHAR, ARFIMA, and RGARCH under most of the six loss functions examined. The headline statistical result is that the 512-context variants $\mathrm{TFM512}_{PT}$ and $\mathrm{TFM512}_{IL}$ reject equal predictive accuracy against every econometric benchmark under the QLIKE loss at the 10% significance level in the Diebold-Mariano and Giacomini-White tests, which the authors interpret as evidence that the foundation model captures tail-sensitive volatility dynamics better than the classical parametric rivals; the log-transformed, 64-point-context fine-tuned variant $\mathrm{TFM64}^{\log}_{IL}$ is reported as never statistically worse than any other model on any loss function.

Load-bearing premise

The claim of statistical superiority rests on running the Diebold-Mariano and Giacomini-White tests on only 21 numbers, one average forecast loss per stock, even though the tests' theory is built on a long time series of daily loss differentials, so if that cross-sectional application is invalid the statistical outperformance collapses into a statement about average losses only.

Editorial extensions

If this is right

  • A financial institution could deploy one pretrained foundation model, refreshed by incremental fine-tuning on recent realized variance, and obtain one-day-ahead forecasts that are statistically better than HAR, ARFIMA, CHAR, and RGARCH under the risk-relevant QLIKE loss at the 10% significance level.
  • Shorter context lengths, around 64 daily observations, are the better choice after fine-tuning: longer contexts improve the zero-shot model but appear to add noise once the model is adapted, so deployment guidance is to keep the look-back short for fine-tuned variants.
  • Zero-shot foundation-model forecasting alone is not enough for volatility: the paper's comparisons show that static pretraining is only a reasonable baseline, and that incremental adaptation is what produces the outperformance.
  • The log-transformed 64-context fine-tuned variant is the safest pick of the study, being never statistically worse than any competitor under any of the six loss functions, which the paper recommends as the preferred configuration.
  • Because fine-tuning updates only the core layers, keeping the model current is computationally cheap, arguing for rolling foundation models forward in production rather than periodically retraining from scratch.

Reading between the lines

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

  • Because the paper's significance tests run on the cross-section of 21 average losses per stock, the defensible reading of the outperformance claim, until a proper time-series implementation is carried out, is that the fine-tuned models hold lower average losses — which the skill-score tables establish directly — rather than proven forecast superiority.
  • The incremental fine-tuning scheme is a moving-window retraining protocol, which suggests the gains come from regime-adaptivity rather than from pretraining per se; a testable extension would refit the simple HAR model on the identical rolling windows to see how much of the gap survives when econometric models are updated just as often.
  • The QLIKE advantage of the 512-context variants may reflect better calibration of the conditional variance, meaning fewer under-predictions, rather than sharper point prediction; checking whether the gain translates into out-of-sample Value-at-Risk coverage rates would separate those mechanisms.
  • A natural extension of the same protocol points at other volatility targets the paper's own benchmarks can distinguish, such as bipower variation, jump components, or implied volatility, since the continuous and discontinuous parts of quadratic variation are known to follow different dynamics.
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

4 major / 6 minor

Summary. The manuscript evaluates Google's TimesFM, a decoder-only time-series foundation model, for one-day-ahead realized variance forecasting on 21 global equity indices from the Oxford-Man Institute Realized Library over 2000-2021. Six TimesFM configurations are compared with HAR, ARFIMA, CHAR, and RGARCH benchmarks (plus log-transformed variants): three context lengths (64, 128, 512) in zero-shot (pretrained) form and after an incremental-learning fine-tuning scheme that iteratively refines the model on expanding windows. Performance is assessed under six loss functions (MSE, MAE, MDA, MAPE, sMAPE, QLIKE), with Diebold-Mariano and Giacomini-White tests, Model Confidence Sets, and skill-score tables. The headline claims are that fine-tuned variants improve forecast accuracy and that TimesFM models 'statistically outperform' the traditional benchmarks, with TFM512PT and TFM512IL highlighted under QLIKE and TFM64logIL identified as the overall best model.

Significance. If the statistical claims could be substantiated, this would be a useful contribution to the emerging literature on foundation models in financial forecasting. The paper's strengths are genuine: the incremental fine-tuning protocol is described in enough detail to reproduce (data splits, hyperparameters, checkpointing), the walk-forward evaluation is applied uniformly to the neural and econometric models, and the descriptive layer is concrete and falsifiable, notably the finding that TFM64logIL attains the lowest average MSE and MAPE and the highest MDA in Table 3. The credibility of the headline result, however, depends entirely on the DM/GW significance tests, and that layer is, as presented, methodologically invalid and numerically inconsistent; the claimed 'statistical outperformance' cannot be taken as established from the manuscript as written.

major comments (4)
  1. [§4 (Evaluation Framework) and Tables 7-12 (captions)] The DM and GW tests are applied to the wrong sampling unit for the asymptotic theory the paper itself spells out. Section 4 defines the loss differential d_t,i,j over the out-of-sample time series t = 1,...,P and states that the DM statistic is asymptotically standard normal under a HAC variance estimator, and the GW statistic chi-square. Every DM/GW table (Tables 7-12) instead reports, in the caption's words, tests applied to the 'cross-section (21 values) of average MSE' (resp. MAE, MDA, MAPE, sMAPE, QLIKE) obtained for the 21 stocks. Replacing the time series of daily loss differentials with 21 per-stock average losses removes the temporal structure on which the null distributions, the HAC estimator, and the P going to infinity asymptotics all rely; with 21 observations the normal and chi-square approximations and the Newey-West variance estimator are not meaningful. Since the abstract and conclusion base the 'statistically outperform' claim precisely on these p-values, the central statistical claim is not established by the reported tests. The same concern applies to the Model Confidence Set summarized in Figure 4, whose implementation (cross-sectional versus time-series loss differentials, bootstrap scheme) is not described.
  2. [Tables 7-12 (p-value patterns) and Table 3] The p-value tables contain patterns that cannot arise from any pairwise comparison of the type described. In Table 7, the RGARCH column is 0.8379 for all 18 non-diagonal rows and the RGARCH row is 0.1621 or 0.1622 for all columns; Table 8 shows the RGARCH row essentially constant at 0.1439-0.1440 and Table 10 at 0.1553-0.1577. These near-constant entries across 19 distinct model pairs are not compatible with pairwise tests on 21 per-stock average losses, which would necessarily differ from one row or column to the next. Moreover, Table 3 reports RGARCH's average MSE (0.00804) as roughly six to seven times larger than every other model's, so a paired cross-sectional test of H1: MSE_RGARCH > MSE_other would be rejected with p-values near zero, not p of about 0.16 (RGARCH row) or p of about 0.84 (RGARCH column). The authors must trace the computation or the data feeding it and regenerate the p-value tables before the statistical claims can be evaluated.
  3. [§5 'Is there any statistically better model?' and Conclusion (Tables 7-12 orientation)] The text reading of the QLIKE results is reversed relative to the tables' stated convention. Table 12's caption defines H0: Qlike_i = Qlike_j versus H1: Qlike_i > Qlike_j with row model i and column model j, so a low p-value means the row model is significantly worse. Under this convention, the evidence that TFM512PT and TFM512IL 'consistently outperform traditional models under the Q-like loss function' (Conclusion, and the analogous passage in Section 5) lives in the econometric rows versus the TimesFM columns, for example Table 12 ARFIMA row versus TFM512PT column (DM p = 0.0493) and HAR row versus TFM512IL column (DM p = 0.0219); the TFM512PT and TFM512IL rows themselves show p-values above 0.94 against every econometric model. As written, the sentence 'TFM512PT and TFM512IL consistently achieve p-values below 0.10 in all pairwise comparisons against econometric models' points the reader to the wrong entries. The MDA tables (Table 9) need a separate sign check for the same reason, since higher MDA indicates better directional accuracy rather than a larger loss, which is the opposite of the 'row is worse when p is low' interpretation used in the captions.
  4. [§5 'Comparative Performance' and 'Identifying the Best Performing Model' (Tables 4-6, Table 3)] The Section 5 narrative reverses the direction of the relative-error tables. Tables 4-6 define each entry as the error of the column model divided by the error of the row benchmark, so values below 1 mean the column model is better. The text instead says 'A model is considered strong if its skill scores exceed 1 in its respective row, meaning it outperforms the benchmark' and later asserts that models 'achieve skill scores exceeding 1 in MSE, MAD, and Qlike, reinforcing their robustness'; both statements are backward relative to the table definitions, and specific claims are contradicted by the tables themselves, for example Table 5 Panel B shows TFM64IL with 0.836 against the ARFIMA row, below 1, and the text's claim that 'TFM64IL achieves skill scores greater than 1 in all six loss functions' is false under either reading of Tables 4-6. This reversal runs through the 'Identifying the Best Performing Model' discussion and needs to be corrected throughout. In addition, the abstract's blanket statement that 'Fine-tuned variants not only improve forecast accuracy' is not supported by Table 3 for the linear IL models, whose average MSE (0.00124-0.00128) exceeds ARFIMA's (0.00120), CHAR's (0.00117), and HAR's (0.00118); the improvement over econometric benchmarks in MSE is specific to the log-transformed variants.
minor comments (6)
  1. [Metrics and captions] Metric terminology is inconsistent: Section 4 defines MAE, Table 3 reports 'MAD' (Mean Absolute Deviation), Tables 4-6 Panel B and Table 8 switch back to MAE, and Table 11's caption refers to 'average MAPE' while the panel reports sMAPE. Standardize the names and the captions.
  2. [Figures and text] Small errors in figures and text: Figure 5's caption says models are evaluated 'across all remaining 18 models' although 19 models are compared; Figure 3's caption says the central line represents the median MSE even though the panels show relative errors for six different loss functions; Section 1 contains the typo 'dicusses'; Section 3 has 'a incremental fine-tuning procedure'; and Section 2's heading 'Literature Review Realized Volatility Forecasting' is missing punctuation.
  3. [§3 Incremental Fine-Tuning] The fine-tuning section says the authors 'adopted the code fine-tuning in the TimesFM library'; for reproducibility they should cite the repository version or commit and state how the hyperparameters and the three context lengths were chosen relative to the out-of-sample period, since the same test window appears to drive both the configuration comparison and the headline results.
  4. [Tables 7-12] Tables 7-12 report only p-values; the per-stock average losses or the loss-differential series that feed the tests should be made available, or the code provided, so that the computations can be verified and the cross-sectional dispersion assessed.
  5. [Introduction] The novelty statement 'to the best of our knowledge, this study is the first to extensively explore the application of time series foundation models for volatility forecasting' should be reconciled with the authors' own reference [27] applying foundation models to VaR forecasting and with the active concurrent literature; the claim should be scoped more carefully.
  6. [Table 2] Table 2 labels the series 'realized volatility' although the data are 5-minute sub-sampled realized variance from the Oxford-Man library; the labels should be aligned with the definitions in the dataset section.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the paper is an empirical benchmark; the only self-citation is motivational and not load-bearing.

full rationale

The paper's central claim is an empirical comparison of one-day-ahead realized volatility forecasts: TimesFM (pretrained and incrementally fine-tuned) is benchmarked against HAR, CHAR, ARFIMA, and RGARCH on 21 indices using six loss functions, with Diebold-Mariano and Giacomini-White tests for statistical significance. No derived quantity is defined in terms of the target claim, and no fitted parameter is renamed as a prediction. The only self-citation is reference [27] (the authors' prior VaR foundation-model paper), used as motivation: 'Recently, [27] applied time-series foundation models to Value-at-Risk (VaR) forecasting...' This citation is not load-bearing; the statistical comparison in the present paper does not depend on any result from [27]. The DM/GW tests are applied to 21 cross-sectional average losses per stock rather than to the time series of daily loss differentials required by the tests' asymptotic theory, and Table 12's row/column conventions appear inconsistent with the text's claim that TFM512PT and TFM512IL have p-values below 0.10 against econometric models. However, this is a statistical validity problem, not circularity: the reported p-values are not constructed to equal the loss averages by definition. No specific circular reduction can be exhibited, so the appropriate finding is no significant circularity, with a minor non-load-bearing self-citation worth score 2.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No new theoretical entities are introduced. The central claim rests on standard realized variance asymptotics, the transferability of TimesFM pretraining, and a nonstandard statistical test setup.

free parameters (3)
  • Context length for TimesFM = 64 (chosen as best)
    The model is evaluated at 64, 128, and 512; the 64-day context is declared best after inspecting out-of-sample results, making it a post hoc selection.
  • Fine-tuning data split proportions = 50% initial train plus validation, then iterative 16%/4% additions
    Chosen by the authors; it determines which periods are out-of-sample and how many incremental rounds occur.
  • Fine-tuning hyperparameters = Learning rate 1e-3 to 1e-4 cosine over 40,000 steps; EMA decay 0.9999; gradient clip 100; patience 5
    Hand-set values with no sensitivity analysis, and the central claim depends on the chosen procedure.
assumptions (4)
  • standard math 5-minute sub-sampled realized variance is a consistent estimator of daily quadratic variation.
    Invoked in the Preliminaries section; a standard asymptotic result, not the target claim.
  • domain assumption TimesFM pretrained weights are a valid starting point for financial volatility series.
    The whole zero-shot evaluation assumes that pretraining on diverse time series transfers to realized variance.
  • ad hoc to paper DM and GW tests applied to 21 cross-sectional averages are valid tests of equal predictive accuracy.
    The paper's statistical significance claims rely on this nonstandard application, which is not justified.
  • domain assumption The expanding-window protocol with previous test data recycled into training yields valid out-of-sample forecasts.
    The incremental learning evaluation assumes this protocol is fair; it is not compared with a purely rolling protocol.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Foundation Time-Series AI Model for Realized Volatility Forecasting." pith.science (2026). https://pith.science/paper/7ZYQGWW5

@misc{pith2026250511163,
  author       = {Pith},
  title        = {Pith review of: Foundation Time-Series AI Model for Realized Volatility Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7ZYQGWW5}},
  note         = {Machine review of arXiv:2505.11163}
}
read the original abstract

Time series foundation models (FMs) have emerged as a popular paradigm for zero-shot multi-domain forecasting. These models are trained on numerous diverse datasets and claim to be effective forecasters across multiple different time series domains, including financial data. In this study, we evaluate the effectiveness of FMs, specifically the TimesFM model, for volatility forecasting, a core task in financial risk management. We first evaluate TimesFM in its pretrained (zero-shot) form, followed by our custom fine-tuning procedure based on incremental learning, and compare the resulting models against standard econometric benchmarks. While the pretrained model provides a reasonable baseline, our findings show that incremental fine-tuning, which allows the model to adapt to new financial return data over time, is essential for learning volatility patterns effectively. Fine-tuned variants not only improve forecast accuracy but also statistically outperform traditional models, as demonstrated through Diebold-Mariano and Giacomini-White tests. These results highlight the potential of foundation models as scalable and adaptive tools for financial forecasting-capable of delivering strong performance in dynamic market environments when paired with targeted fine-tuning strategies.

Figures

Figures reproduced from arXiv: 2505.11163 by the authors.

Figure 1
Figure 1. TimesFM structure Note. From [18]. A Decoder-Only Foundation Model [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Fine-tuning procedure with incremental learning. An iterative fine-tuning [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Visualization of Cross-Sectional Out-of-Sample Relative Error Distribution for six errors considered in the study: (a) MSE, (b) MAE, (c) MDA, (d) MAPE, (e) sMAPE and (f) Qlike. For each error, we construct a boxplot for the one-day-ahead out-of-sample forecast error of models (TFM64IL, TFM512IL, TFM64P T , TFM512P T , TFM64log IL , TFM512log IL , TFM64log P T , TFM512log P T ) relative to CHARlog model. The sample i… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Inclusion rate in the Model Inclusion Set (MCS): We present the results of the Model Confidence Set (MCS). This method identifies a subset of superior models by sequentially testing the null hypothesis of equal predictive accuracy, H0 : MSEmodel i = MSEmodel j . We rep…
Figure 5
Figure 5. Figure 5: We partition the test set into deciles based on the observed daily realized variance. Specifically, the first subsample contains the 10% of observations with the lowest realized variance in the out-of-sample period and is labeled (0.0, 0.1). In this figure, we present …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Forecasting Realized Volatility with Time Series Foundation Models: A Comparison with Econometric Benchmarks

    q-fin.ST 2026-07 accept novelty 6.0 of 10

    Zero-shot time series foundation models largely fail to beat econometric benchmarks for realized volatility forecasting, with only TTM achieving a narrow, calibration-driven edge.

Reference graph

Works this paper leans on

58 extracted references · 46 canonical work pages · cited by 1 Pith paper

  1. [1]

    Andersen, Tim Bollerslev, and Francis X

    Torben G. Andersen, Tim Bollerslev, and Francis X. Diebold. Roughing it up: Including jump components in the measurement, modeling, and forecasting of return volatility. The Review of Economics and Statistics , 89(4):701–720, 11 2007

    work page 2007

  2. [2]

    Chronos: Learning the language of time series

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mer- cado, Huibin Shen, Oleksandr Shchur, Syama Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, et al. Chronos: Learning the language of time series. 2024

    work page 2024

  3. [3]

    Garch based artificial neural networks in forecasting conditional variance of stock returns

    Josip Arneri´ c, Tea Poklepovi´ c, and Zdravka Aljinovi´ c. Garch based artificial neural networks in forecasting conditional variance of stock returns. Croatian Operational Research Review, pages 329–343, 2014

    work page 2014

  4. [4]

    Lassoing the har model: A model selection perspective on realized volatility dynamics

    Francesco Audrino and Simon D Knaus. Lassoing the har model: A model selection perspective on realized volatility dynamics. Econometric Reviews , 35(8-10):1485–1521, 2016

    work page 2016

  5. [5]

    The impact of senti- ment and attention measures on stock market volatility

    Francesco Audrino, Fabio Sigrist, and Daniele Ballinari. The impact of senti- ment and attention measures on stock market volatility. International Journal of Forecasting, 36(2):334–357, 2020

    work page 2020

  6. [6]

    Estimating quadratic variation using realized variance

    Ole E Barndorff-Nielsen and Neil Shephard. Estimating quadratic variation using realized variance. Journal of Applied Econometrics , 17(5):457–477, 2002

    work page 2002

  7. [7]

    Barndorff-Nielsen and Neil Shephard

    Ole E. Barndorff-Nielsen and Neil Shephard. Power and bipower variation with stochastic volatility and jumps. Journal of Financial Econometrics , 2(1):1–37, 2004

    work page 2004

  8. [8]

    Mamba4cast: Efficient zero-shot time series forecasting with state space models

    Sathya Kamesh Bhethanabhotla, Omar Swelam, Julien Siems, David Salinas, and Frank Hutter. Mamba4cast: Efficient zero-shot time series forecasting with state space models. arXiv preprint arXiv:2410.09385 , 2024

    arXiv 2024

Show all 58 references
  1. [9]

    Generalized autoregressive conditional heteroskedasticity

    Tim Bollerslev. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3):307–327, 1986

    1986

  2. [10]

    Patton, and Rogier Quaedvlieg

    Tim Bollerslev, Andrew J. Patton, and Rogier Quaedvlieg. Exploiting the errors: A simple approach for improved volatility forecasting. Journal of Econometrics , 192(1):1–18, 2016

    2016

  3. [11]

    Realized volatility forecasting with neural networks

    Andrea Bucci. Realized volatility forecasting with neural networks. Journal of Financial Econometrics, 18(3):502–531, 2020

    2020

  4. [12]

    Timedit: General- purpose diffusion transformers for time series foundation model

    Defu Cao, Wen Ye, Yizhou Zhang, and Yan Liu. Timedit: General- purpose diffusion transformers for time series foundation model. arXiv preprint arXiv:2409.02322, 2024

    2024 arXiv

  5. [13]

    Building news measures from textual data and an application to volatility forecasting

    Massimiliano Caporin and Francesco Poli. Building news measures from textual data and an application to volatility forecasting. Econometrics, 5(3):35, 2017. 26 Anubha Goel et al

    2017

  6. [14]

    Visionts: Visual masked autoencoders are free-lunch zero-shot time series forecasters

    Mouxiang Chen, Lefei Shen, Zhuo Li, Xiaoyun Joy Wang, Jianling Sun, and Chenghao Liu. Visionts: Visual masked autoencoders are free-lunch zero-shot time series forecasters. arXiv preprint arXiv:2408.17253 , 2024

    2024 arXiv

  7. [15]

    A machine learn- ing approach to volatility forecasting

    Kim Christensen, Mathias Siggaard, and Bezirgen Veliyev. A machine learn- ing approach to volatility forecasting. Journal of Financial Econometrics , 21(5):1680–1727, 2023

    2023

  8. [16]

    A simple approximate long-memory model of realized volatility

    Fulvio Corsi. A simple approximate long-memory model of realized volatility. Journal of Financial Econometrics , 7(2):174–196, 2009

    2009

  9. [17]

    Har volatility modelling with heterogeneous leverage and jumps

    Fulvio Corsi and Roberto Reno. Har volatility modelling with heterogeneous leverage and jumps. Available at SSRN , 1316953, 2009

    2009

  10. [18]

    A decoder- only foundation model for time-series forecasting

    Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou. A decoder- only foundation model for time-series forecasting. In Proceedings of the 41st International Conference on Machine Learning , 2024

    2024

  11. [19]

    Diebold and Robert S

    Francis X. Diebold and Robert S. Mariano. Comparing predictive accuracy. Journal of Business & Economic Statistics , 20(1):134–144, 2002

    2002

  12. [20]

    Tiny time mixers (ttms): Fast pre-trained models for enhanced zero/few-shot forecasting of multivariate time series

    Vijay Ekambaram, Arindam Jati, Pankaj Dayama, Sumanta Mukherjee, Nam Nguyen, Wesley M Gifford, Chandra Reddy, and Jayant Kalagnanam. Tiny time mixers (ttms): Fast pre-trained models for enhanced zero/few-shot forecasting of multivariate time series. Advances in Neural Informat...

    2024

  13. [21]

    Autoregressive conditional heteroscedasticity with estimates of the variance of united kingdom inflation

    Robert F Engle. Autoregressive conditional heteroscedasticity with estimates of the variance of united kingdom inflation. Econometrica, pages 987–1007, 1982

    1982

  14. [22]

    Modeling and predicting the cboe market volatility index

    Marcelo Fernandes, Marcelo C Medeiros, and Marcel Scharth. Modeling and predicting the cboe market volatility index. Journal of Banking & Finance , 40:1–10, 2014

    2014

  15. [23]

    Using lstm and gru neural network methods for traffic flow prediction

    Rui Fu, Zuo Zhang, and Li Li. Using lstm and gru neural network methods for traffic flow prediction. In 2016 31st Youth academic annual conference of Chinese association of automation (YAC) , pages 324–328. IEEE, 2016

    2016

  16. [24]

    Neural network–based financial volatility forecasting: A systematic review

    Wenbo Ge, Pooia Lalbakhsh, Leigh Isai, Artem Lenskiy, and Hanna Suominen. Neural network–based financial volatility forecasting: A systematic review. ACM Computing Surveys (CSUR) , 55(1):1–30, 2022

    2022

  17. [25]

    Tests of conditional predictive ability

    Raffaella Giacomini and Halbert White. Tests of conditional predictive ability. Econometrica, 74(6):1545–1578, 2006

    2006

  18. [26]

    Monash time series forecasting archive.arXiv preprint arXiv:2105.06643, 2021

    Rakshitha Godahewa, Christoph Bergmeir, Geoffrey I Webb, Rob J Hyndman, and Pablo Montero-Manso. Monash time series forecasting archive.arXiv preprint arXiv:2105.06643, 2021

    2021 arXiv

  19. [27]

    Time-series foundation ai model for value-at-risk forecasting

    Anubha Goel, Puneet Pasricha, and Juho Kanniainen. Time-series foundation ai model for value-at-risk forecasting. arXiv preprint arXiv:2410.11773 , 2024

    2024 arXiv

  20. [28]

    Using neural networks for forecasting volatility of s&p 500 index futures prices

    Shaikh A Hamid and Zahid Iqbal. Using neural networks for forecasting volatility of s&p 500 index futures prices. Journal of Business Research, 57(10):1116–1125, 2004

    2004

  21. [29]

    Hansen and Asger Lunde

    Peter R. Hansen and Asger Lunde. A forecast comparison of volatility models: does anything beat a garch (1, 1)? Journal of Applied Econometrics , 20(7):873– 889, 2005

    2005

  22. [30]

    The model confidence set

    Peter R Hansen, Asger Lunde, and James M Nason. The model confidence set. Econometrica, 79(2):453–497, 2011. AI for Volatility Forecasting 27

    2011

  23. [31]

    Realized garch: a joint model for returns and realized measures of volatility

    Peter Reinhard Hansen, Zhuo Huang, and Howard Howan Shek. Realized garch: a joint model for returns and realized measures of volatility. Journal of Applied Econometrics, 27(6):877–906, 2012

    2012

  24. [32]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 16000–16009, 2022

    2022

  25. [33]

    Combining conditional volatility forecasts using neural networks: an application to the ems exchange rates

    Michael Y Hu and Christos Tsoukalas. Combining conditional volatility forecasts using neural networks: an application to the ems exchange rates. Journal of International Financial Markets, Institutions and Money , 9(4):407–422, 1999

    1999

  26. [34]

    Fore- casting realised volatility using arfima and har models

    Marwan Izzeldin, M Kabir Hassan, Vasileios Pappas, and Mike Tsionas. Fore- casting realised volatility using arfima and har models. Quantitative Finance , 19(10):1627–1638, 2019

    2019

  27. [35]

    Volatility fore- cast using hybrid neural network models

    Werner Kristjanpoller, Anton Fadic, and Marcel C Minutolo. Volatility fore- cast using hybrid neural network models. Expert Systems with Applications , 41(5):2437–2442, 2014

    2014

  28. [36]

    Foundation models for time series analysis: A tutorial and survey

    Yuxuan Liang, Haomin Wen, Yuqi Nie, Yushan Jiang, Ming Jin, Dongjin Song, Shirui Pan, and Qingsong Wen. Foundation models for time series analysis: A tutorial and survey. In Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining , pages 6555–6565, 2024

    2024

  29. [37]

    Trading volume and realized volatility forecasting: Evidence from the china stock market

    Min Liu, Wei-Chong Choo, Chi-Chuan Lee, and Chien-Chiang Lee. Trading volume and realized volatility forecasting: Evidence from the china stock market. Journal of Forecasting, 42(1):76–100, 2023

    2023

  30. [38]

    Novel volatility forecasting using deep learning–long short term memory recurrent neural networks

    Yang Liu. Novel volatility forecasting using deep learning–long short term memory recurrent neural networks. Expert Systems with Applications , 132:99–109, 2019

    2019

  31. [39]

    Forecasting of realised volatility with the random forests algorithm

    Chuong Luong and Nikolai Dokuchaev. Forecasting of realised volatility with the random forests algorithm. Journal of Risk and Financial Management , 11(4):61, 2018

    2018

  32. [40]

    The m4 competition: 100,000 time series and 61 forecasting methods

    Spyros Makridakis, Evangelos Spiliotis, and Vassilios Assimakopoulos. The m4 competition: 100,000 time series and 61 forecasting methods. International Journal of Forecasting, 36(1):54–74, 2020

    2020

  33. [41]

    Stock market volatil- ity: Identifying major drivers and the nature of their impact

    Stefan Mittnik, Nikolay Robinzonov, and Martin Spindler. Stock market volatil- ity: Identifying major drivers and the nature of their impact. Journal of Banking & Finance, 58:1–14, 2015

    2015

  34. [42]

    A time series is worth 64 words: Long-term forecasting with transformers

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. A time series is worth 64 words: Long-term forecasting with transformers. arXiv preprint arXiv:2211.14730, 2022

    2022 arXiv

  35. [43]

    N- beats: Neural basis expansion analysis for interpretable time series forecasting

    Boris N Oreshkin, Dmitri Carpov, Nicolas Chapados, and Yoshua Bengio. N- beats: Neural basis expansion analysis for interpretable time series forecasting. arXiv preprint arXiv:1905.10437 , 2019

    1905 arXiv

  36. [44]

    Deep adaptive input normalization for time series forecast- ing

    Nikolaos Passalis, Anastasios Tefas, Juho Kanniainen, Moncef Gabbouj, and Alexandros Iosifidis. Deep adaptive input normalization for time series forecast- ing. IEEE transactions on neural networks and learning systems , 31(9):3760– 3765, 2019

    2019

  37. [45]

    Volatility forecast comparison using imperfect volatility proxies

    Andrew J Patton. Volatility forecast comparison using imperfect volatility proxies. Journal of Econometrics , 160(1):246–256, 2011. 28 Anubha Goel et al

    2011

  38. [46]

    Good volatility, bad volatility: Signed jumps and the persistence of volatility

    Andrew J Patton and Kevin Sheppard. Good volatility, bad volatility: Signed jumps and the persistence of volatility. Review of Economics and Statistics , 97(3):683–697, 2015

    2015

  39. [47]

    A review of general- ized zero-shot learning methods

    Farhad Pourpanah, Moloud Abdar, Yuxuan Luo, Xinlei Zhou, Ran Wang, Chee Peng Lim, Xi-Zhao Wang, and QM Jonathan Wu. A review of general- ized zero-shot learning methods. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(4):4051–4070, 2022

    2022

  40. [48]

    Machine learning for realised volatility forecasting

    Eghbal Rahimikia and Ser-Huang Poon. Machine learning for realised volatility forecasting. Available at SSRN , 3707796, 2020

    2020

  41. [49]

    Lag-llama: Towards foundation models for time series forecasting

    Kashif Rasul, Arjun Ashok, Andrew Robert Williams, Arian Khorasani, George Adamopoulos, Rishika Bhagwatkar, Marin Biloˇ s, Hena Ghonia, Nadhir Hassen, Anderson Schneider, et al. Lag-llama: Towards foundation models for time series forecasting. In R0-FoMo: Robustness of Few-sho...

    2023

  42. [50]

    Deepar: Probabilistic forecasting with autoregressive recurrent networks

    David Salinas, Valentin Flunkert, Jan Gasthaus, and Tim Januschowski. Deepar: Probabilistic forecasting with autoregressive recurrent networks. International Journal of Forecasting, 36(3):1181–1191, 2020

    2020

  43. [51]

    Forecasting stock market volatility using realized garch model: International evidence

    Prateek Sharma et al. Forecasting stock market volatility using realized garch model: International evidence. The Quarterly Review of Economics and Finance , 59:222–230, 2016

    2016

  44. [52]

    Introducing nbeatsx to realized volatility forecasting

    Hugo Gobato Souto and Amir Moradi. Introducing nbeatsx to realized volatility forecasting. Expert Systems with Applications , 242:122802, 2024

    2024

  45. [53]

    Financial returns modelled by the product of two stochastic processes-a study of the daily sugar prices 1961-75

    Stephen John Taylor. Financial returns modelled by the product of two stochastic processes-a study of the daily sugar prices 1961-75. Time series analysis: theory and practice, 1:203–226, 1982

    1961

  46. [54]

    Data-driven neural architecture learning for financial time-series forecasting

    Dat Thanh Tran, Juho Kanniainen, Moncef Gabbouj, and Alexandros Iosi- fidis. Data-driven neural architecture learning for financial time-series forecasting. arXiv preprint arXiv:1903.06751 , 2019

    1903 arXiv

  47. [55]

    Unified training of universal time series forecasting transformers

    Gerald Woo, Chenghao Liu, Akshat Kumar, Caiming Xiong, Silvio Savarese, and Doyen Sahoo. Unified training of universal time series forecasting transformers. 2024

    2024

  48. [56]

    Gpt (generative pre-trained transformer)–a comprehensive review on enabling technologies, potential applications, emerging challenges, and future directions

    Gokul Yenduri, M Ramalingam, G Chemmalar Selvi, Y Supriya, Gautam Sri- vastava, Praveen Kumar Reddy Maddikunta, G Deepti Raj, Rutvij H Jhaveri, B Prabadevi, Weizheng Wang, et al. Gpt (generative pre-trained transformer)–a comprehensive review on enabling technologies, potentia...

    2024

  49. [57]

    Deeplob: Deep convolutional neural networks for limit order books

    Zihao Zhang, Stefan Zohren, and Stephen Roberts. Deeplob: Deep convolutional neural networks for limit order books. IEEE Transactions on Signal Processing , 67(11):3001–3012, 2019

    2019

  50. [58]

    Informer: Beyond efficient transformer for long sequence time- series forecasting

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. Informer: Beyond efficient transformer for long sequence time- series forecasting. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 11106–11115, 202...

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.