Registry section

Papers

Research, connected to implementations and reproductions.Filter and compare in Browse →

• A Multimodal Foundation Agent for Financial Trading (FinAgent)Paper

  Presents FinAgent, a multimodal trading agent that processes numerical, textual, and visual market data with dual-level reflection and diversified memory retrieval.

  2024KDD 2024arXiv:2402.18485ai: advanced
  added 2026-08-15 · external
• AI-Trader: Benchmarking Autonomous Agents in Real-Time Financial MarketsPaper

  A live, data-uncontaminated benchmark that runs six mainstream LLMs as autonomous trading agents across US stocks, A-shares and crypto at several trading frequencies.

  2025arXiv preprintarXiv:2512.10971
  added 2026-08-17 · MIT · external
• AlphaAgent: LLM-Driven Alpha Mining with Regularized Exploration to Counteract Alpha DecayPaper

  Mines decay-resistant alpha factors with three LLM agents under originality, alignment, and complexity regularizers, backtesting the results through Qlib.

  2025KDD 2025arXiv:2502.16789ai: advancedsetup: advanced
  added 2026-08-17 · MIT · external
• Building Diversified Portfolios that Outperform Out of Sample (Hierarchical Risk Parity)Paper

  López de Prado's 2016 paper introducing Hierarchical Risk Parity, a clustering-based allocation algorithm that requires no inversion of the covariance matrix.

  2016The Journal of Portfolio Management 42(4), 59–69
  added 2026-08-17 · proprietary · external
• Can LLM-based Financial Investing Strategies Outperform the Market in Long Run?Paper

  Re-backtests open-source LLM investing agents over 2004-2024 and 100+ S&P 500 symbols, and finds their reported advantages largely disappear once common biases are controlled.

  2026KDD 2026arXiv:2505.07078
  added 2026-08-17 · Apache-2.0 · external
• CryptoTrade: A Reflective LLM-based Agent to Guide Zero-shot Cryptocurrency TradingPaper

  An LLM trading agent that fuses on-chain and off-chain signals with a reflection step to make daily cryptocurrency trading decisions.

  2024EMNLP 2024 (main conference)arXiv:2407.09546
  added 2026-08-17 · CC-BY-NC-SA-4.0 · external
• Deep reinforcement learning for automated stock trading: an ensemble strategyPaper

  Ensembles PPO, A2C, and DDPG into a single Dow-30 trading policy selected by rolling Sharpe ratio, benchmarked against the DJIA and a minimum-variance portfolio.

  2020ICAIF '20 (First ACM International Conference on AI in Finance)arXiv:2511.12120ai: advanced
  added 2026-08-17 · proprietary · external
• FinCon: A Synthesized LLM Multi-Agent System with Conceptual Verbal Reinforcement for Enhanced Financial Decision MakingPaper

  An LLM multi-agent trading system with a manager-analyst hierarchy and a self-critiquing risk-control loop; the code has never been released.

  2024NeurIPS 2024arXiv:2407.06567ai: advanced
  added 2026-08-17 · external
• FinMem: A Performance-Enhanced LLM Trading Agent with Layered Memory and Character DesignPaper

  Proposes an LLM trading agent whose layered memory and configurable character profile let it adapt to new market information over multiple time horizons.

  2023ICLR 2024 Workshop on LLM AgentsarXiv:2311.13743
  added 2026-08-15 · external
• FinRL-Meta: Market Environments and Benchmarks for Data-Driven Financial Reinforcement LearningPaper

  Presents FinRL-Meta, a library that processes real market data into gym-style trading environments and reproduces published DRL trading strategies as benchmarks.

  2022NeurIPS 2022 Datasets and Benchmarks TrackarXiv:2211.03107ai: advanced
  added 2026-08-17 · MIT · external
• FinRL: A Deep Reinforcement Learning Library for Automated Stock Trading in Quantitative FinancePaper

  The originating FinRL paper — a three-layer DRL trading pipeline of market environments, agents, and applications, presented at the NeurIPS 2020 Deep RL Workshop.

  2020Deep Reinforcement Learning Workshop, NeurIPS 2020 (non-archival)arXiv:2011.09607ai: advanced
  added 2026-08-17 · external
• FLAG-TRADER: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial TradingPaper

  Fine-tunes an LLM as the policy network of a reinforcement-learning trading agent; published in Findings of ACL 2025, with no code, weights, or data released.

  2025Findings of the Association for Computational Linguistics: ACL 2025arXiv:2502.11433ai: advanced
  added 2026-08-17 · external
• FreqAI: generalizing adaptive modeling for chaotic time-series market forecastsPaper

  Software paper for FreqAI, the adaptive machine-learning module inside the Freqtrade bot, which retrains models during live deployment and feeds forecasts to entry/exit logic.

  2022Journal of Open Source Software 7(80):4864
  added 2026-08-17 · GPL-3.0 · external
• From Deep Learning to LLMs: A Survey of AI in Quantitative InvestmentPaper

  Surveys AI in quantitative investment from hand-crafted factors through deep learning to LLM-based autonomous agents, organized around the alpha strategy pipeline.

  2025arXivarXiv:2503.21422
  added 2026-08-17 · external
• High-frequency trading in a limit order bookPaper

  The canonical stochastic-control model for market making, deriving an inventory-skewed reservation price and the optimal bid-ask spread around it.

  2008Quantitative Finance 8(3), 217-224trading: advanced
  added 2026-08-17 · proprietary · external
• Large Language Model Agent in Financial Trading: A SurveyPaperbeginner-friendly

  Surveys the research landscape of LLM-based agents applied to financial trading, covering architectures, data inputs, backtesting results, and open challenges.

  2024arXivarXiv:2408.06361
  added 2026-08-15 · external
• Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models (SEP)Paper

  Fine-tunes an LLM with a self-reflective agent and PPO to produce explainable stock predictions without human annotation, then applies the same loop to portfolio weight generation.

  2024WWW 2024arXiv:2402.03659ai: advanced
  added 2026-08-17 · external
• MarS: a Financial Market Simulation Engine Powered by Generative Foundation ModelPaper

  An ICLR 2025 poster introducing the Large Market Model, an order-level generative foundation model, and MarS, the market simulation engine built on it.

  2025ICLR 2025 (poster)arXiv:2409.07486trading: advancedai: advanced
  added 2026-08-17 · MIT · external
• Qlib: An AI-oriented Quantitative Investment PlatformPaper

  The Microsoft Research preprint introducing Qlib, cited by the Qlib repository itself as the project's reference paper.

  2020arXiv preprintarXiv:2009.11189
  added 2026-08-17 · unknown · external
• QuantHarness: Price-Driven Multi-Agent LLMs for High-Frequency TradingPaper

  Proposes a four-agent LLM system for short-horizon technical trading from price data alone, released with an MIT-licensed implementation and a 1,600-file K-line benchmark.

  2025arXivarXiv:2509.09995
  added 2026-08-17 · MIT · external
• R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint OptimizationPaper

  Describes RD-Agent(Q), a multi-agent research-and-development loop that jointly optimizes quant factors and models, shipped as the fin_quant scenario of Microsoft's RD-Agent.

  2025NeurIPS 2025 Datasets and Benchmarks Track (poster)arXiv:2505.15155trading: advancedai: advanced
  added 2026-08-17 · MIT · external
• Stock Movement Prediction from Tweets and Historical Prices (StockNet)Paper

  Introduces a deep generative model that predicts stock movement from tweets and prices jointly, together with the 88-stock StockNet dataset the authors collected for it.

  2018ACL 2018ai: advanced
  added 2026-08-17 · MIT · external
• StockBench: Can LLM Agents Trade Stocks Profitably In Real-world Markets?Paper

  Runs LLM agents as daily traders over 20 top-weighted Dow Jones stocks for 82 trading days in 2025, scored on return, drawdown and Sortino against a passive buy-and-hold baseline.

  2025arXivarXiv:2510.02209
  added 2026-08-17 · Apache-2.0 · external
• Time Travel is Cheating: Going Live with DeepFund for Real-Time Fund Investment BenchmarkingPaper

  Presents DeepFund, a live benchmark that scores LLM fund-management agents on market data published after each model's pretraining cutoff.

  2025NeurIPS 2025 Datasets and Benchmarks Track (poster)arXiv:2505.11065
  added 2026-08-17 · MIT · external
• TRA: Learning Multiple Stock Trading Patterns with Temporal Routing Adaptor and Optimal TransportPaper

  Introduces the Temporal Routing Adaptor, a router-plus-multi-predictor module trained with optimal-transport assignment so one backbone can model several trading patterns.

  2021KDD 2021arXiv:2106.12950ai: advancedsetup: advanced
  added 2026-08-17 · MIT · external
• Trading with the Momentum Transformer: An Intelligent and Interpretable ArchitecturePaper

  Proposes an attention-LSTM hybrid that learns position sizing for a time-series momentum portfolio of liquid futures contracts.

  2021arXivarXiv:2112.08534trading: advancedai: advanced
  added 2026-08-17 · MIT · external
• TradingAgents: Multi-Agents LLM Financial Trading FrameworkPaper

  Introduces a multi-agent LLM trading framework whose specialized roles mirror the structure of a professional trading firm.

  2024arXiv (oral, Multi-Agent AI in the Real World workshop)arXiv:2412.20138
  added 2026-08-15 · external
• Universal Trading for Order Execution with Oracle Policy DistillationPaper

  Introduces Oracle Policy Distillation for reinforcement-learning order execution; the resulting OPDS method ships as a runnable workflow in Microsoft Qlib.

  2021AAAI 2021arXiv:2103.10860trading: advancedai: advancedsetup: advanced
  added 2026-08-17 · MIT · external
• When Agents Trade: Live Multi-Market Trading Benchmark for LLM Agents (AMA)Paper

  A live benchmark that ran four LLM trading-agent architectures across five model backbones on two equities and two cryptocurrencies for two months of 2025, scored against buy-and-hold.

  2025arXivarXiv:2510.11695
  added 2026-08-17 · external

Missing something? Add it — no YAML required.