pith. sign in

arxiv: 2502.18834 · v2 · submitted 2025-02-26 · 💻 cs.CE · cs.LG

FinTSB: A Comprehensive and Practical Benchmark for Financial Time Series Forecasting

Yifan Hu , Yuante Li , Peiyuan Liu , Yuxia Zhu , Naiqi Li , Tao Dai , Shu-Tao Xia , Dawei Cheng
show 1 more author
Changjun Jiang
This is my paper

Pith reviewed 2026-05-23 02:47 UTC · model grok-4.3

classification 💻 cs.CE cs.LG
keywords financial time seriesforecasting benchmarkstock movement patternsevaluation standardizationtrading constraintstime series forecastingmarket simulation
0
0 comments X

The pith

FinTSB benchmark categorizes stock movements into four parts, standardizes metrics in three dimensions, and models trading constraints to fix evaluation gaps.

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

Financial time series forecasting evaluations often miss the range of real market behaviors, use inconsistent metrics that prevent fair comparisons, and ignore fees and regulations that make reported results unrealistic. The paper introduces FinTSB to close these gaps by splitting movement patterns into four categories with data preprocessing, unifying metrics across three dimensions in a shared pipeline, and simulating constraints such as transaction fees. If correct, this produces comparable performance numbers across methods from different backbones and yields practical guidance on which models work under specific market conditions. A reader would care because prior results become more reliable for deciding whether a forecasting approach can support actual investment decisions.

Core claim

FinTSB is a benchmark that increases variety by categorizing movement patterns into four specific parts, tokenizing and preprocessing data, and assessing quality via sequence characteristics; eliminates biases by standardizing metrics across three dimensions and providing a lightweight pipeline for methods from various backbones; and models regulatory constraints including transaction fees to simulate real trading scenarios, enabling extensive experiments that highlight insights for model selection under varying conditions.

What carries the argument

FinTSB benchmark, which applies four-part movement pattern categorization, three-dimension metric standardization, and regulatory constraint modeling to produce practical evaluations.

If this is right

  • Methods built on different backbones can be compared directly without biases from varying evaluation settings.
  • Reported performance numbers will incorporate transaction fees and other constraints rather than remaining inflated.
  • Experiments on the benchmark supply concrete guidance for choosing models suited to particular market conditions.
  • Researchers obtain a single platform and pipeline for consistent improvement and testing of forecasting approaches.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Widespread use could redirect research effort toward methods that remain effective once fees and constraints are included.
  • The same structure of pattern categorization plus constraint modeling might transfer to forecasting tasks in regulated domains such as energy or logistics.
  • If new market regimes appear that do not fit the four categories, the taxonomy itself could be empirically extended using additional sequence data.

Load-bearing premise

Dividing all observed stock movement patterns into four specific categories is sufficient to cover the diversity present in dynamic financial markets.

What would settle it

A collection of real financial time series sequences whose movement patterns fall outside the four defined categories or produce inconsistent method rankings even after the proposed standardization is applied.

Figures

Figures reproduced from arXiv: 2502.18834 by Changjun Jiang, Dawei Cheng, Naiqi Li, Peiyuan Liu, Shu-Tao Xia, Tao Dai, Yifan Hu, Yuante Li, Yuxia Zhu.

Figure 2
Figure 2. Figure 2: In the field of financial time series forecasting, exist [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Visualization of financial time series data with dif [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Hexbin plots illustrating the normalized density [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The pipeline of FinTSB with four integral modules. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Cumulative return of typical methods over the whole year of 2024 in the CSI 300. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Inference efficiency comparison under FinTSB. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
read the original abstract

Financial time series (FinTS) record the behavior of human-brain-augmented decision-making, capturing valuable historical information that can be leveraged for profitable investment strategies. Not surprisingly, this area has attracted considerable attention from researchers, who have proposed a wide range of methods based on various backbones. However, the evaluation of the area often exhibits three systemic limitations: 1. Failure to account for the full spectrum of stock movement patterns observed in dynamic financial markets. (Diversity Gap), 2. The absence of unified assessment protocols undermines the validity of cross-study performance comparisons. (Standardization Deficit), and 3. Neglect of critical market structure factors, resulting in inflated performance metrics that lack practical applicability. (Real-World Mismatch). Addressing these limitations, we propose FinTSB, a comprehensive and practical benchmark for financial time series forecasting (FinTSF). To increase the variety, we categorize movement patterns into four specific parts, tokenize and pre-process the data, and assess the data quality based on some sequence characteristics. To eliminate biases due to different evaluation settings, we standardize the metrics across three dimensions and build a user-friendly, lightweight pipeline incorporating methods from various backbones. To accurately simulate real-world trading scenarios and facilitate practical implementation, we extensively model various regulatory constraints, including transaction fees, among others. Finally, we conduct extensive experiments on FinTSB, highlighting key insights to guide model selection under varying market conditions. Overall, FinTSB provides researchers with a novel and comprehensive platform for improving and evaluating FinTSF methods. The code is available at https://github.com/TongjiFinLab/FinTSB.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces FinTSB, a benchmark for financial time series forecasting (FinTSF) that targets three limitations in prior work: the diversity gap (failure to cover full spectrum of stock movement patterns), standardization deficit (inconsistent evaluation protocols), and real-world mismatch (neglect of market constraints like transaction fees). It addresses these by categorizing movement patterns into four parts with tokenization and sequence-based quality assessment, standardizing metrics across three dimensions via a lightweight pipeline incorporating multiple backbones, modeling regulatory constraints, and running experiments to derive model-selection insights under varying conditions. The code is released at https://github.com/TongjiFinLab/FinTSB.

Significance. If the four-category categorization, standardized pipeline, and constraint modeling are shown to be comprehensive and bias-free, FinTSB could provide a reproducible platform that enables fairer cross-study comparisons and more practical FinTSF evaluations. The open-source code is a clear strength that supports community adoption and verification.

major comments (3)
  1. [Abstract, §3] Abstract and §3 (Data Construction): The central claim that categorizing movement patterns into four specific parts, followed by tokenization and sequence-characteristic assessment, closes the diversity gap is not supported by quantitative verification. No table or analysis demonstrates that these four categories capture regime shifts, fat tails, or microstructure events (e.g., flash crashes) beyond the source datasets, leaving the 'full spectrum' assertion untested.
  2. [§4] §4 (Standardization) and experiments section: The claim that standardizing metrics across three dimensions eliminates evaluation biases lacks explicit definition of those dimensions or before/after comparison showing reduced variance in cross-model rankings. Without such evidence, the pipeline's ability to support valid cross-study comparisons remains unverified.
  3. [Experiments] Experiments section: The reported insights on model selection under varying market conditions are presented without ablation showing that results change when the four-category split or constraint modeling is removed, making it unclear whether the benchmark itself drives the observed differences versus prior ad-hoc setups.
minor comments (2)
  1. [§4] Notation for the three standardization dimensions should be introduced with explicit equations or pseudocode in §4 to improve clarity.
  2. [Figures] Figure captions for any sequence-characteristic plots should include the exact statistics used (e.g., autocorrelation, volatility) rather than generic labels.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify areas for improvement in the manuscript. We address each major comment point by point below, indicating planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Abstract, §3] Abstract and §3 (Data Construction): The central claim that categorizing movement patterns into four specific parts, followed by tokenization and sequence-characteristic assessment, closes the diversity gap is not supported by quantitative verification. No table or analysis demonstrates that these four categories capture regime shifts, fat tails, or microstructure events (e.g., flash crashes) beyond the source datasets, leaving the 'full spectrum' assertion untested.

    Authors: We agree that the manuscript would benefit from explicit quantitative verification. The four categories were empirically derived from observed patterns in the source datasets to span trends, volatility, reversals, and stability. We will add a new analysis subsection and table in §3 that reports statistics on regime shifts (via change-point detection), fat tails (kurtosis and tail indices), and coverage of microstructure events across categories, including examples from the data. revision: yes

  2. Referee: [§4] §4 (Standardization) and experiments section: The claim that standardizing metrics across three dimensions eliminates evaluation biases lacks explicit definition of those dimensions or before/after comparison showing reduced variance in cross-model rankings. Without such evidence, the pipeline's ability to support valid cross-study comparisons remains unverified.

    Authors: The three dimensions are explicitly the choice of evaluation metrics, the standardization of experimental protocols (e.g., splits and windows), and the inclusion of diverse backbones; however, we acknowledge the need for clearer exposition and supporting evidence. We will revise §4 to provide formal definitions of the dimensions and add a before/after comparison table quantifying the reduction in ranking variance across models. revision: yes

  3. Referee: [Experiments] Experiments section: The reported insights on model selection under varying market conditions are presented without ablation showing that results change when the four-category split or constraint modeling is removed, making it unclear whether the benchmark itself drives the observed differences versus prior ad-hoc setups.

    Authors: We agree that the absence of explicit ablations leaves the unique contribution of the benchmark components less clear. The current experiments already contrast the standardized pipeline against typical ad-hoc practices in the literature. We will add a discussion subsection in the experiments section that qualitatively contrasts the obtained insights with those from prior non-standardized setups and note the role of the categorization and constraints; a quantitative ablation removing these elements is beyond the scope of the present work but will be flagged for future extension. revision: partial

Circularity Check

0 steps flagged

No circularity: benchmark assembled from explicit design rules without reduction to fitted values or self-citations

full rationale

The paper constructs FinTSB via explicit categorization rules (four movement patterns), standardization of metrics across three dimensions, and incorporation of external regulatory constraints such as transaction fees. No equations, parameter fitting, predictions, or uniqueness theorems appear in the provided text. The central claims rest on data-processing choices and pipeline design rather than any quantity defined in terms of itself or derived from self-citation chains. This is a standard benchmark paper whose contributions are self-contained against external data and constraints.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The benchmark rests on the domain assumption that four movement pattern categories capture market diversity and on standard practices for metric standardization and constraint modeling; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption Financial markets exhibit four distinct movement pattern categories that together cover the full spectrum of observed behaviors.
    Invoked to address the Diversity Gap by categorizing, tokenizing, and pre-processing data.

pith-pipeline@v0.9.0 · 5857 in / 1160 out tokens · 31161 ms · 2026-05-23T02:47:15.778419+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

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

  1. From Feedback Loops to Policy Updates: Reinforcement Fine-Tuning for LLM-Based Alpha Factor Discovery

    cs.CE 2026-05 unverdicted novelty 7.0

    QuantEvolver applies reinforcement fine-tuning to evolve an LLM policy for generating executable alpha factor expressions, yielding higher-quality and more complementary factors than prompt-based baselines on market b...

  2. FinSTaR: Towards Financial Reasoning with Time Series Reasoning Models

    cs.AI 2026-05 conditional novelty 7.0

    FinSTaR reaches 78.9% accuracy on a new financial time series reasoning benchmark by applying Compute-in-CoT for deterministic assessments and Scenario-Aware CoT for stochastic predictions.

  3. From Observations to States: Latent Time Series Forecasting

    cs.LG 2026-01 conditional novelty 7.0

    LatentTSF improves time series forecasting accuracy and representation quality by shifting prediction from observation space to a learned latent state space via autoencoding.

  4. TelecomTS: A Multi-Modal Observability Dataset for Time Series and Language Analysis

    cs.AI 2025-10 conditional novelty 7.0

    TelecomTS is a new observability dataset from 5G networks that preserves absolute scale and supports multi-modal tasks, showing that current time series and language models struggle with abrupt noisy dynamics.

  5. FinDocMRE: A Benchmark for Document-Level Financial Multimodal Reasoning Evaluation

    cs.CE 2026-05 unverdicted novelty 6.0

    FinDocMRE is a new multi-image document-level benchmark spanning 12 financial domains and 5 task types, showing that 11 tested LMMs all score below 65 overall with particular weaknesses in numerical estimation and cro...

  6. GCGNet: Graph-Consistent Generative Network for Time Series Forecasting with Exogenous Variables

    cs.LG 2026-03 unverdicted novelty 6.0

    GCGNet uses a variational generator, graph structure aligner, and graph refiner to jointly capture temporal and channel correlations in time series forecasting with exogenous variables, outperforming baselines on 12 r...

  7. Strat-LLM: Stratified Strategy Alignment for LLM-based Stock Trading with Real-time Multi-Source Signals

    cs.AI 2026-05 unverdicted novelty 5.0

    Strat-LLM demonstrates that LLM trading performance varies by reasoning mode and model scale, with strict alignment reducing drawdowns in downtrends and deep reasoning avoiding small-gain traps.

Reference graph

Works this paper leans on

89 extracted references · 89 canonical work pages · cited by 7 Pith papers · 9 internal anchors

  1. [1]

    Hervé Abdi and Lynne J Williams. 2010. Principal component analysis. Wiley interdisciplinary reviews: computational statistics 2, 4 (2010), 433–459

    work page 2010

  2. [2]

    Mad- dix, Michael W

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Syndar Rangapuram, Sebas- tian Pineda Arango, Shubham Kapoor, Jasper Zschiegner, Danielle C. Mad- dix, Michael W. Mahoney, Kari Torkkola, Andrew Gordon Wilson, Michael Bohlke-Schneider, and Yuyang Wang. 2024. Chronos: Learning the Language ...

    work page 2024

  3. [3]

    Shaojie Bai, J Zico Kolter, and Vladlen Koltun. 2018. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint arXiv:1803.01271 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  4. [4]

    G. E. P. Box, Gwilym M. Jenkins, and John F. Macgregor. 1968. Some Recent Advances in Forecasting and Control.Journal of The Royal Statistical Society Series C-applied Statistics 17 (1968), 158–179. https://api.semanticscholar.org/CorpusID: 227323668

    work page 1968

  5. [5]

    Salvatore Carta, Anselmo Ferreira, Alessandro Sebastian Podda, Diego Reforgiato Recupero, and Antonio Sanna. 2021. Multi-DQN: An ensemble of Deep Q-learning agents for stock market forecasting. Expert Systems with Applications (2021)

    work page 2021

  6. [6]

    Tianqi Chen and Carlos Guestrin. 2016. XGBoost: A Scalable Tree Boosting System. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (San Francisco, California, USA) (KDD ’16). Association for Computing Machinery, New York, NY, USA, 785–794. https: //doi.org/10.1145/2939672.2939785

    work page doi:10.1145/2939672.2939785 2016

  7. [7]

    Weijun Chen, Shun Li, Xipu Yu, Heyuan Wang, Wei Chen, and Tengjiao Wang

    work page

  8. [8]

    In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24 , Kate Larson (Ed.)

    Automatic De-Biased Temporal-Relational Modeling for Stock Invest- ment Recommendation. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24 , Kate Larson (Ed.). International Joint Conferences on Artificial Intelligence Organization, 1999–2008. https: //doi.org/10.24963/ijcai.2024/221 Main Track

    work page doi:10.24963/ijcai.2024/221 1999

  9. [9]

    Dawei Cheng, Ye Liu, Zhibin Niu, and Liqing Zhang. 2018. Modeling Similarities Among Multi-Dimensional Financial Time Series. IEEE Access 6 (2018), 43404– 43413. https://doi.org/10.1109/ACCESS.2018.2862908

    work page doi:10.1109/access.2018.2862908 2018

  10. [10]

    Dawei Cheng, Fangzhou Yang, Sheng Xiang, and Jin Liu. 2022. Financial time series forecasting with multi-modality graph neural network. Pattern Recognition 121 (2022), 108218. https://doi.org/10.1016/j.patcog.2021.108218

    work page doi:10.1016/j.patcog.2021.108218 2022

  11. [11]

    Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, and Yoshua Bengio. 2014. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling. arXiv preprint arXiv:1412.3555 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  12. [12]

    Tao Dai, Beiliang Wu, Peiyuan Liu, Naiqi Li, Jigang Bao, Yong Jiang, and Shu-Tao Xia. 2024. Periodicity Decoupling Framework for Long-term Series Forecasting. International Conference on Learning Representations (2024)

    work page 2024

  13. [13]

    Divyanshu Daiya, Monika Yadav, and Harshit Singh Rao. 2024. Diffstock: Proba- bilistic Relational Stock Market Predictions Using Diffusion Models. In ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    work page 2024

  14. [14]

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Im- ageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition . 248–255. https://doi.org/10.1109/CVPR. 2009.5206848

    work page doi:10.1109/cvpr 2009

  15. [15]

    Shangkun Deng, Xiaoru Huang, Yingke Zhu, Zhihao Su, Zhe Fu, and Tatsuro Shimada. 2023. Stock index direction forecasting using an explainable eXtreme Gradient Boosting and investor sentiments. The North American Journal of Economics and Finance (2023)

    work page 2023

  16. [16]

    Yitong Duan, Lei Wang, Qizhong Zhang, and Jian Li. 2022. FactorVAE: A Proba- bilistic Dynamic Factor Model Based on Variational Autoencoder for Predicting Cross-Sectional Stock Returns. Proceedings of the AAAI Conference on Artificial Intelligence (2022)

    work page 2022

  17. [17]

    Yuan Gao, Haokun Chen, Xiang Wang, Zhicai Wang, Xue Wang, Jinyang Gao, and Bolin Ding. 2024. DiffsFormer: A Diffusion Transformer on Stock Factor Augmentation. arXiv preprint arXiv:2402.06656 (2024)

    work page arXiv 2024

  18. [18]

    Georg M. Goerg. 2013. Forecastable component analysis. In International Confer- ence on Machine Learning (ICML’13). JMLR.org

    work page 2013

  19. [19]

    Albert Gu and Tri Dao. 2024. Mamba: Linear-Time Sequence Modeling with Selective State Spaces. In First Conference on Language Modeling . https: //openreview.net/forum?id=tEYskw1VY2

    work page 2024

  20. [20]

    Tuomas Haarnoja, Aurick Zhou, Kristian Hartikainen, George Tucker, Sehoon Ha, Jie Tan, Vikash Kumar, Henry Zhu, Abhishek Gupta, Pieter Abbeel, and Sergey Levine. 2019. Soft Actor-Critic Algorithms and Applications. arXiv preprint arXiv:1812.05905 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  21. [21]

    David Hirshleifer, Lin Peng, and Qiguang Wang. 2024. News Diffusion in Social Networks and Stock Market Reactions. The Review of Financial Studies (2024), hhae025. https://doi.org/10.1093/rfs/hhae025

    work page doi:10.1093/rfs/hhae025 2024

  22. [22]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising diffusion probabilistic models. Advances in neural information processing systems (2020)

    work page 2020

  23. [23]

    Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. Neural Computation (1997)

    work page 1997

  24. [24]

    Yifan Hu, Peiyuan Liu, Yuante Li, Dawei Cheng, Naiqi Li, Tao Dai, Jigang Bao, and Xia Shu-Tao. 2025. FinMamba: Market-Aware Graph Enhanced Multi-Level Mamba for Stock Movement Prediction. arXiv preprint arXiv:2502.06707 (2025)

    work page arXiv 2025

  25. [25]

    Yifan Hu, Peiyuan Liu, Peng Zhu, Dawei Cheng, and Tao Dai. 2024. Adap- tive Multi-Scale Decomposition Framework for Time Series Forecasting. arXiv preprint arXiv:2406.03751 (2024)

    work page arXiv 2024

  26. [26]

    Yifan Hu, Guibin Zhang, Peiyuan Liu, Disen Lan, Naiqi Li, Dawei Cheng, Tao Dai, Shu-Tao Xia, and Shirui Pan. 2025. TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecasting. arXiv preprint arXiv:2501.13041 (2025)

    work page arXiv 2025

  27. [27]

    Yu-Hao Huang, Chang Xu, Yang Liu, Weiqing Liu, Wu-Jun Li, and Jiang Bian

    work page

  28. [28]

    arXiv preprint arXiv:2408.12991 (2024)

    Controllable Financial Market Generation with Diffusion Guided Meta Agent. arXiv preprint arXiv:2408.12991 (2024)

    work page arXiv 2024

  29. [29]

    Pavan Kumar Illa, Balakesavareddy Parvathala, and Anand Kumar Sharma. 2022. Stock price prediction methodology using random forest algorithm and support vector machine. Materials Today: Proceedings (2022)

    work page 2022

  30. [30]

    Narasimhan Jegadeesh and Sheridan Titman. 1993. Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance 48 (1993), 65–91. https://api.semanticscholar.org/CorpusID:13713547

    work page 1993

  31. [31]

    Jihyeong Jeon, Jiwon Park, Chanhee Park, and U Kang. 2024. FreQuant: A Reinforcement-Learning based Adaptive Portfolio Optimization with Multi- frequency Decomposition. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24) . Association for Computing Machinery

    work page 2024

  32. [32]

    Guolin Ke, Qi Meng, Thomas Finley, Taifeng Wang, Wei Chen, Weidong Ma, Qiwei Ye, and Tie-Yan Liu. 2017. LightGBM: a highly efficient gradient boosting decision tree. In Proceedings of the 31st International Conference on Neural Information Processing Systems (Long Beach, California, USA) (NIPS’17). Curran Associates Inc., Red Hook, NY, USA

    work page 2017

  33. [33]

    Thomas N Kipf and Max Welling. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  34. [34]

    Suchow, and Khaldoun Khashanah

    Haohang Li, Yangyang Yu, Zhi Chen, Yuechen Jiang, Yang Li, Denghui Zhang, Rong Liu, Jordan W. Suchow, and Khaldoun Khashanah. 2024. FinMem: A Performance-Enhanced LLM Trading Agent with Layered Memory and Char- acter Design. In ICLR 2024 Workshop on Large Language Model (LLM) Agents . https://openreview.net/forum?id=sstfVOwbiG

    work page 2024

  35. [35]

    Shuqi Li, Yuebo Sun, Yuxin Lin, Xin Gao, Shuo Shang, and Rui Yan. 2024. Causal- Stock: Deep End-to-end Causal Discovery for News-driven Multi-stock Move- ment Prediction. In The Thirty-eighth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=5BXXoJh0Vr

    work page 2024

  36. [36]

    Tong Li, Zhaoyang Liu, Yanyan Shen, Xue Wang, Haokun Chen, and Sen Huang

    work page

  37. [37]

    Proceedings of the AAAI Conference on Artificial Intelligence 38 (2024)

    MASTER: Market-Guided Stock Transformer for Stock Price Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence 38 (2024)

    work page 2024

  38. [38]

    Yang Li, Yangyang Yu, Haohang Li, Zhi Chen, and Khaldoun Khashanah. 2023. TradingGPT: Multi-Agent System with Layered Memory and Distinct Characters for Enhanced Financial Trading Performance. arXiv preprint arXiv:2309.03736 (2023)

    work page arXiv 2023

  39. [39]

    Continuous control with deep reinforcement learning

    Timothy P. Lillicrap, Jonathan J. Hunt, Alexander Pritzel, Nicolas Heess, Tom Erez, Yuval Tassa, David Silver, and Daan Wierstra. 2019. Continuous control with deep reinforcement learning. arXiv preprint arXiv:1509.02971 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  40. [40]

    Shengsheng Lin, Weiwei Lin, Wentai Wu, Feiyu Zhao, Ruichao Mo, and Haotong Zhang. 2023. Segrnn: Segment recurrent neural network for long-term time series forecasting. arXiv preprint arXiv:2308.11200 (2023)

    work page arXiv 2023

  41. [41]

    Microsoft COCO: Common Objects in Context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, Lubomir Bourdev, Ross Girshick, James Hays, Pietro Perona, Deva Ramanan, C. Lawrence Zitnick, and Piotr Dollár. 2015. Microsoft COCO: Common Objects in Context. arXiv preprint arXiv:1405.0312 (2015)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  42. [42]

    Yuling Lin, Haixiang Guo, and Jinglu Hu. 2013. An SVM-based approach for stock market trend prediction. In The 2013 International Joint Conference on Neural Networks (IJCNN)

    work page 2013

  43. [43]

    Peiyuan Liu, Hang Guo, Tao Dai, Naiqi Li, Jigang Bao, Xudong Ren, Yong Jiang, and Shu-Tao Xia. 2024. CALF: Aligning LLMs for Time Series Forecasting via Cross-modal Fine-Tuning. arXiv preprint arXiv:2403.07300 (2024)

    work page arXiv 2024

  44. [44]

    Peiyuan Liu, Beiliang Wu, Yifan Hu, Naiqi Li, Tao Dai, Jigang Bao, and Shu-Tao Xia. 2024. TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting. arXiv preprint arXiv:2410.04442 (2024)

    work page arXiv 2024

  45. [45]

    Xiao-Yang Liu, Hongyang Yang, Jiechao Gao, and Christina Dan Wang. 2021. FinRL: Deep reinforcement learning framework to automate trading in quantita- tive finance. ACM International Conference on AI in Finance (ICAIF) (2021)

    work page 2021

  46. [46]

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024. iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. In International Conference on Learning Representations

    work page 2024

  47. [47]

    Yong Liu, Haoran Zhang, Chenyu Li, Xiangdong Huang, Jianmin Wang, and Mingsheng Long. 2024. Timer: Generative Pre-trained Transformers Are Large Time Series Models. In Forty-first International Conference on Machine Learning . https://openreview.net/forum?id=bYRYb7DMNo

    work page 2024

  48. [48]

    Henrik Madsen. 2007. Time series analysis. Chapman & Hall

    work page 2007

  49. [49]

    Rizwan Mushtaq. 2011. Augmented dickey fuller test. (2011). Conference, Date, Preprint. Yifan Hu, Yuante Li, Peiyuan Liu, Yuxia Zhu, Naiqi Li, Tao Dai, Shu-tao Xia, Dawei Cheng and Changjun Jiang

    work page 2011

  50. [50]

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2023. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. In International Conference on Learning Representations

    work page 2023

  51. [51]

    Hui Niu, Siyuan Li, Jiahao Zheng, Zhouchi Lin, Bo An, Jian Li, and Jian Guo

    work page

  52. [52]

    In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24

    IMM: An Imitative Reinforcement Learning Approach with Predictive Representation Learning for Automatic Market Making. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24 . International Joint Conferences on Artificial Intelligence Organization

    work page

  53. [53]

    Kaiming Pan, Yifan Hu, Li Han, Haoyu Sun, Dawei Cheng, and Yuqi Liang. 2024. Cross-contextual Sequential Optimization via Deep Reinforcement Learning for Algorithmic Trading. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management (Boise, ID, USA) (CIKM ’24). Association for Computing Machinery, New York, NY, USA, 4811–4818

    work page 2024

  54. [54]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. 2019. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems 32 (2019)

    work page 2019

  55. [55]

    Poterba and Lawrence H

    James M. Poterba and Lawrence H. Summers. 1988. Mean reversion in stock prices: Evidence and Implications. Journal of Financial Economics (1988)

    work page 1988

  56. [56]

    Cottrell

    Yao Qin, Dongjin Song, Haifeng Chen, Wei Cheng, Guofei Jiang, and Garrison W. Cottrell. 2017. A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI-17. 2627–2633. https://doi.org/10.24963/ijcai.2017/ 366

    work page doi:10.24963/ijcai.2017/ 2017

  57. [57]

    Jensen, Zhenli Sheng, and Bin Yang

    Xiangfei Qiu, Jilin Hu, Lekui Zhou, Xingjian Wu, Junyang Du, Buang Zhang, Chenjuan Guo, Aoying Zhou, Christian S. Jensen, Zhenli Sheng, and Bin Yang

    work page

  58. [58]

    TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods. Proc. VLDB Endow. (2024)

    work page 2024

  59. [59]

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov

    work page

  60. [60]

    Proximal Policy Optimization Algorithms

    Proximal Policy Optimization Algorithms. arXiv preprint arXiv:1707.06347 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  61. [61]

    Xiaoming Shi, Shiyu Wang, Yuqi Nie, Dianqi Li, Zhou Ye, Qingsong Wen, and Ming Jin. 2025. Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts. In The Thirteenth International Conference on Learning Repre- sentations. https://openreview.net/forum?id=e1wDDFmlVu

    work page 2025

  62. [62]

    Farzan Soleymani and Eric Paquet. 2021. Deep graph convolutional reinforcement learning for financial portfolio management – DeepPocket. Expert Syst. Appl. (2021)

    work page 2021

  63. [63]

    Jiaming Song, Chenlin Meng, and Stefano Ermon. 2020. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  64. [64]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems (2017)

    work page 2017

  65. [65]

    Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, Yoshua Bengio, et al. 2017. Graph attention networks. stat (2017)

    work page 2017

  66. [66]

    Heyuan Wang, Tengjiao Wang, Shun Li, Jiayi Zheng, Shijie Guan, and Wei Chen

    work page

  67. [67]

    In Proceedings of the Thirty-First International Joint Conference on Artificial Intelli- gence, IJCAI-22, Lud De Raedt (Ed.)

    Adaptive Long-Short Pattern Transformer for Stock Investment Selection. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelli- gence, IJCAI-22, Lud De Raedt (Ed.). International Joint Conferences on Artificial Intelligence Organization, 3970–3977. https://doi.org/10.24963/ijcai.2022/551 Main Track

    work page doi:10.24963/ijcai.2022/551 2022

  68. [68]

    Jingyuan Wang, Yang Zhang, Ke Tang, Junjie Wu, and Zhang Xiong. 2019. Al- phaStock: A Buying-Winners-and-Selling-Losers Investment Strategy using In- terpretable Deep Reinforcement Attention Networks. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

    work page 2019

  69. [69]

    Zhang, and JUN ZHOU

    Shiyu Wang, Haixu Wu, Xiaoming Shi, Tengge Hu, Huakun Luo, Lintao Ma, James Y. Zhang, and JUN ZHOU. 2024. TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting. In International Conference on Learning Representations. https://openreview.net/forum?id=7oLshfEIC2

    work page 2024

  70. [70]

    Yuxuan Wang, Haixu Wu, Jiaxiang Dong, Yong Liu, Mingsheng Long, and Jianmin Wang. 2024. Deep Time Series Models: A Comprehensive Survey and Benchmark. arXiv preprint arXiv:2407.13278 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  71. [71]

    Zhicheng Wang, Biwei Huang, Shikui Tu, Kun Zhang, and Lei Xu. 2021. Deep- Trader: A Deep Reinforcement Learning Approach for Risk-Return Balanced Portfolio Management with Market Conditions Embedding. Proceedings of the AAAI Conference on Artificial Intelligence (2021)

    work page 2021

  72. [72]

    Hongjie Xia, Huijie Ao, Long Li, Yu Liu, Sen Liu, Guangnan Ye, and Hongfeng Chai. 2024. CI-STHPAN: Pre-trained Attention Network for Stock Selection with Channel-Independent Spatio-Temporal Hypergraph. Proceedings of the AAAI Conference on Artificial Intelligence 38, 8 (Mar. 2024), 9187–9195. https: //doi.org/10.1609/aaai.v38i8.28770

    work page doi:10.1609/aaai.v38i8.28770 2024

  73. [73]

    Haochong Xia, Shuo Sun, Xinrun Wang, and Bo An. 2024. Market-GAN: Adding Control to Financial Market Data Generation with Semantic Context.Proceedings of the AAAI Conference on Artificial Intelligence (2024)

    work page 2024

  74. [74]

    Quanzhou Xiang, Zhan Chen, Qi Sun, and Rujun Jiang. 2024. RSAP-DFM: Regime- Shifting Adaptive Posterior Dynamic Factor Model for Stock Returns Prediction. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24. International Joint Conferences on Artificial Intelligence Organization

    work page 2024

  75. [75]

    Sheng Xiang, Dawei Cheng, Chencheng Shang, Ying Zhang, and Yuqi Liang. 2022. Temporal and Heterogeneous Graph Neural Network for Financial Time Series Prediction. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management

    work page 2022

  76. [76]

    Rong Xing, Rui Cheng, Jiwen Huang, Qing Li, and Jingmei Zhao. 2024. Learning to Understand the Vague Graph for Stock Prediction With Momentum Spillovers. IEEE Trans. on Knowl. and Data Eng. (April 2024), 1698–1712. https://doi.org/10. 1109/TKDE.2023.3310592

    work page arXiv 2024

  77. [77]

    Wenbo Yan and Ying Tan. 2024. TCGPN: Temporal-Correlation Graph Pre-trained Network for Stock Forecasting. arXiv preprint arXiv:2407.18519 (2024)

    work page arXiv 2024

  78. [78]

    Xiao Yang, Weiqing Liu, Dong Zhou, Jiang Bian, and Tie-Yan Liu. 2020. Qlib: An AI-oriented Quantitative Investment Platform. arXiv preprint arXiv:2009.11189 (2020)

    work page arXiv 2020

  79. [79]

    Suchow, Zhenyu Cui, Rong Liu, Zhaozhuo Xu, Denghui Zhang, Koduvayur Subbalakshmi, GUOJUN XIONG, Yueru He, Jimin Huang, Dong Li, and Qianqian Xie

    Yangyang Yu, Zhiyuan Yao, Haohang Li, Zhiyang Deng, Yuechen Jiang, Yupeng Cao, Zhi Chen, Jordan W. Suchow, Zhenyu Cui, Rong Liu, Zhaozhuo Xu, Denghui Zhang, Koduvayur Subbalakshmi, GUOJUN XIONG, Yueru He, Jimin Huang, Dong Li, and Qianqian Xie. 2024. FinCon: A Synthesized LLM Multi-Agent System with Conceptual Verbal Reinforcement for Enhanced Financial D...

    work page 2024

  80. [80]

    Liang Zeng, Lei Wang, Hui Niu, Ruchen Zhang, Ling Wang, and Jian Li. 2024. Trade When Opportunity Comes: Price Movement Forecasting via Locality-Aware Attention and Iterative Refinement Labeling. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24 . International Joint Conferences on Artificial Intelligen...

    work page 2024

Showing first 80 references.