Paper: arXiv 2411.12747

Authors: Junhua Liu

Abstract

Financial AI empowers sophisticated approaches to financial market forecasting, portfolio optimization, and automated trading. This survey provides a systematic analysis of these developments across three primary dimensions: predictive models that capture complex market dynamics, decision-making frameworks that optimize trading and investment strategies, and knowledge augmentation systems that leverage unstructured financial information. We examine significant innovations including foundation models for financial time series, graph-based architectures for market relationship modeling, and hierarchical frameworks for portfolio optimization. Analysis reveals crucial trade-offs between model sophistication and practical constraints, particularly in high-frequency trading applications. We identify critical gaps and open challenges between theoretical advances and industrial implementation, outlining open challenges and opportunities for improving both model performance and practical applicability.

Complexity vs Empirical Score

  • Math Complexity: 8.0/10
  • Empirical Rigor: 3.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper is rich in mathematical formalisms, presenting detailed notations and problem formulations for predictive and decision-making tasks. However, as a survey of existing works, it focuses on theoretical analysis and architectural reviews rather than providing new empirical results, backtests, or implementation code.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Analyze Financial AI Architectures &<br>Challenges"] --> B{"Key Methodology"}

  B --> C["1. Systematic Analysis of Three Dimensions"]
  C --> D["Predictive Models<br>Market Dynamics"]
  C --> E["Decision Frameworks<br>Portfolio Optimization"]
  C --> F["Knowledge Augmentation<br>Unstructured Data"]

  B --> G["Data/Inputs<br>Financial Time Series, Market Graphs,<br>Unstructured Financial Info"]

  G --> H["Computational Processes<br>Foundation Models, Graph-Based<br>Architectures, Hierarchical Frameworks"]

  H --> I{"Key Findings & Outcomes"}
  
  I --> J["Innovations<br>Foundation Models, Graph Architecture,<br>Hierarchical Optimization"]
  I --> K["Trade-offs<br>Sophistication vs. Practical Constraints<br>(e.g., High-Frequency Trading)"]
  I --> L["Open Challenges<br>Theoretical Advances vs.<br>Industrial Implementation"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper provides a systematic analysis of financial AI developments across predictive models, decision-making frameworks, and knowledge augmentation systems.

Source sentence
This survey provides a systematic analysis of these developments across three primary dimensions: predictive models that capture complex market dynamics, decision-making frameworks that optimize trading and investment strategies, and knowledge augmentation systems that leverage unstructured financial information.

Assumes

No explicit assumptions found in the text.

Data

No dataset named in the text.

Costs

Not applicable The paper reports no trading or execution result; the cost the extractor found is a model parameter.

The paper-text extraction found a cost parameter inside a model (Incorporates turnover sparsity regularization to manage transaction costs.); the classifier found no reported trading result, so costs are not applicable.

Finds 5

  • Analysis reveals crucial trade‑offs between model sophistication and practical constraints, particularly in high‑frequency trading applications. performance
    Source sentence
    Analysis reveals crucial trade-offs between model sophistication and practical constraints, particularly in high-frequency trading applications.
  • The survey identifies critical gaps between theoretical advances and industrial implementation, outlining open challenges and opportunities. other
    Source sentence
    We identify critical gaps and open challenges between theoretical advances and industrial implementation, outlining open challenges and opportunities for improving both model performance and practical applicability.
  • Few surveyed papers provide comprehensive evidence of successful industrial deployment, with most evaluations relying on historical backtesting. other
    Source sentence
    While works like [Han et al., 2023] report production latency metrics, and [Qin et al., 2024; Zong et al., 2024] demonstrate strong backtesting performance, none of the surveyed papers provide comprehensive evidence of suc-cessful industrial deployment.
  • Architectural approaches demonstrate clear trade‑offs between modeling sophistication and computational efficiency. other
    Source sentence
    The architectural approaches demonstrate clear trade-offs be-tween modeling sophistication and computational efficiency.
  • Production deployment remains a significant challenge, with limited real‑world implementation and regulatory compliance demonstrated. other
    Source sentence
    Production deployment remains a significant challenge, with few studies demonstrating successful real-world implementation or com-prehensive compliance with regulatory requirements.

Checked by others

No other paper in the archive has been found to support or contradict these findings. Coverage is partial, so this is neither replication nor refutation.

Code

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (69)

  1. Market-GAN: Adding Control to Financial Market Data Generation with Semantic Context — → 2309.07708
  2. Extracting Financial Events from Raw Texts via Matrix Chunking (2024)
  3. Fnspid: A comprehensive financial news dataset in time series — → 2402.06698
  4. FinReport: Explainable Stock Earnings Forecasting via News Factor Analyzing Model (2024)
  5. AlphaFin: Benchmarking Financial Analysis with Retrieval-Augmented Stock-Chain Framework (2024) — link
  6. When AI Meets Finance (StockAgent): Large Language Model-based Stock Trading in Simulated Real-world Environments — → 2407.18957
  7. Prompting for Numerical Sequences: A Case Study on Market Comment Generation (2024) — link
  8. Large Language Models as Financial Data Annotators: A Study on Effectiveness and Efficiency (2024) — link
and 61 more
  1. Stockemotions: Discover investor emotions for financial sentiment analysis and multivariate time series — → 2301.09279
  2. Temporal Relational Reasoning of Large Language Models for Detecting Stock Portfolio Crashes — → 2410.17266
  3. Econagent: large language model-empowered agents for simulating macroeconomic activities (2024)
  4. A Survey of Large Language Models for Financial Applications: Progress, Prospects and Challenges — → 2406.11903
  5. Large language models in finance: A survey — → 2311.10723
  6. A survey of large language models in finance (finllms) (2024) — link
  7. Large Language Model Agent in Financial Trading: A Survey — → 2408.06361
  8. A Scoping Review of ChatGPT Research in Accounting and Finance — → 2412.05731
  9. Revolutionizing Finance with LLMs: An Overview of Applications and Insights (2024) — link
  10. Gsm-symbolic: Understanding the limitations of mathematical reasoning in large language models (2024) — link
  11. Can large language models reason and plan? (2024)
  12. Can large language models reason and plan? (2024)
  13. Language agents with reinforcement learning for strategic play in the werewolf game (2023) — link
  14. Reflexion: Language agents with verbal reinforcement learning (2024)
  15. Large Language Model as a Policy Teacher for Training Reinforcement Learning Agents (2024)
  16. LLM-based Multi-Agent Reinforcement Learning: Current and Future Directions (2024) — link
  17. Trend-Heuristic Reinforcement Learning Framework for News-Oriented Stock Portfolio Management (2024)
  18. Instruct-fingpt: Financial sentiment analysis by instruction tuning of general-purpose large language models — → 2306.12659
  19. Developing a Multi-agent and Self-adaptive Framework with Deep Reinforcement Learning for Dynamic Portfolio Risk Management — → 2402.00515
  20. Reinforcement Learning with Maskable Stock Representation for Portfolio Management in Customizable Stock Pools — → 2311.10801
  21. CrisisBERT: A Robust Transformer for Crisis Classification and Contextual Crisis Embedding (2021) — link
  22. FreQuant: A Reinforcement-Learning based Adaptive Portfolio Optimization with Multi-frequency Decomposition (2024)
  23. Mitigating Extremal Risks: A Network-Based Portfolio Strategy — → 2409.12208
  24. Risk-Managed Sparse Index Tracking Via Market Graph Clustering (2024)
  25. Automate Strategy Finding with LLM in Quant investment — → 2409.06289
  26. MDGNN: Multi-Relational Dynamic Graph Neural Network for Comprehensive and Dynamic Stock Investment Prediction — → 2402.06633
  27. MASTER: Market-Guided Stock Transformer for Stock Price Forecasting (2024)
  28. Double-Path Adaptive-correlation Spatial-Temporal Inverted Transformer for Stock Time Series Forecasting (2024) — link
  29. Stock movement prediction based on bi-typed hybrid-relational market knowledge graph via dual attention networks (2022)
  30. Doubleadapt: A meta-learning approach to incremental learning for stock trend forecasting — → 2306.09862
  31. Spatial-Temporal Graph Representation Learning for Tactical Networks Future State Prediction (2024) — link
  32. Long short-term memory (1997)
  33. Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation (2014) — link
  34. Attention is All you Need (2017)
  35. GARCH-Informed Neural Networks for Volatility Prediction in Financial Markets — → 2410.00288
  36. Diffstock: Probabilistic relational stock market predictions using diffusion models — → 2403.14063
  37. From News to Forecast: Integrating Event Analysis in LLM-Based Time Series Forecasting with Reflection (2024)
  38. StockFormer: Learning Hybrid Trading Machines with Predictive Coding. (2023)
  39. MacMic: Executing Iceberg Orders via Hierarchical Reinforcement Learning (2024) — link
  40. IMM: An Imitative Reinforcement Learning Approach with Predictive Representation Learning for Automatic Market Making — → 2308.08918
  41. Earnhft: Efficient hierarchical reinforcement learning for high frequency trading — → 2309.12891
  42. MacroHFT: Memory Augmented Context-aware Reinforcement Learning On High Frequency Trading — → 2406.14537
  43. Efficient Continuous Space Policy Optimization for High-frequency Trading (2023)
  44. Optimizing Trading Strategies in Quantitative Markets using Multi-Agent Reinforcement Learning — → 2303.11959
  45. Hierarchical Reinforced Trader (HRT): A Bi-Level Approach for Optimizing Stock Selection and Execution — → 2410.14927
  46. A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist — → 2402.18485
  47. CI-STHPAN: Pre-trained Attention Network for Stock Selection with Channel-Independent Spatio-Temporal Hypergraph (2024)
  48. Automatic De-Biased Temporal-Relational Modeling for Stock Investment Recommendation (2024) — link
  49. RSAP-DFM: Regime-Shifting Adaptive Posterior Dynamic Factor Model for Stock Returns Prediction (2024) — link
  50. Relational Temporal Graph Convolutional Networks for Ranking-Based Stock Prediction (2023) — link
  51. ECHO-GL: Earnings Calls-Driven Heterogeneous Graph Learning for Stock Movement Prediction (2024)
  52. Multi-Relational Graph Diffusion Neural Network with Parallel Retention for Stock Trends Classification — → 2401.05430
  53. Trade When Opportunity Comes: Price Movement Forecasting via Locality-Aware Attention and Iterative Refinement Labeling (2024) — link
  54. MANA-Net: Mitigating Aggregated Sentiment Homogenization with News Weighting for Enhanced Market Prediction — → 2409.05698
  55. Saliency-Aware Interpolative Augmentation for Multimodal Financial Prediction (2024)
  56. Learning to generate explainable stock predictions using self-reflective large language models — → 2402.03659
  57. Timer: Generative Pre-trained Transformers Are Large Time Series Models (2024)
  58. MOMENT: A Family of Open Time-series Foundation Models (2024)
  59. TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting (2024)
  60. TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis (2023)
  61. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers (2023)

27 of 69 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.