DEV.co
AI Frameworks · TauricResearch

TradingAgents

TradingAgents is an open-source Python framework that uses multiple LLM-powered agents (analysts, researchers, traders, risk managers) to collaboratively evaluate markets and execute trading decisions. It supports multiple LLM providers and includes backtesting, data connectors, and structured decision logging for research and experimentation.

Source: GitHub — github.com/TauricResearch/TradingAgents
91.7k
GitHub stars
17.7k
Forks
Python
Primary language
Apache-2.0
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
RepositoryTauricResearch/TradingAgents
OwnerTauricResearch
Primary languagePython
LicenseApache-2.0 — OSI-approved
Stars91.7k
Forks17.7k
Open issues279
Latest releasev0.3.1 (2026-07-05)
Last updated2026-07-05
Sourcehttps://github.com/TauricResearch/TradingAgents

What TradingAgents is

Multi-agent framework built on LangGraph with specialized agent roles that process financial data from multiple sources (Alpha Vantage, FRED, Polymarket, sentiment APIs), use structured outputs, and maintain persistent decision logs. Supports GPT, Gemini, Claude, DeepSeek, Qwen, GLM, Grok, and OpenAI-compatible endpoints; includes checkpoint resume and configurable retry budgets.

Quickstart

Get the TradingAgents source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/TauricResearch/TradingAgents.gitcd TradingAgents# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Trading Strategy Research & Backtesting

Test multi-agent trading approaches on historical data with structured decision logging to understand agent behavior and refine strategies without live capital risk.

Financial Data Orchestration

Build pipelines that aggregate diverse data sources (market data, sentiment, fundamentals, macroeconomic indicators) and feed structured analysis to downstream trading systems.

Multi-Model LLM Experimentation

Compare trading performance and decision quality across different LLM providers and model families in a consistent framework without rebuilding infrastructure.

Implementation considerations

  • Extensive API key management: OpenAI, Google, Anthropic, xAI, DeepSeek, Aliyun Dashscope, Zhipu, plus optional data vendors (Alpha Vantage, FRED, Polymarket, sentiment sources). Set via environment variables or config.
  • Data fidelity and access control: Framework includes look-ahead filtering (Alpha Vantage) and ticker path-traversal hardening; verify data contracts and vendor APIs match your jurisdiction and use case.
  • LLM model selection and cost: Multiple provider options (GPT-5.x, Gemini 3.x, Claude 4.x, Grok 4.x, open models via Ollama) with different latency, cost, and quality profiles. Temperature and retry budget are configurable.
  • Checkpoint resume and state management: LangGraph-based persistence allows recovery from interruptions, but verify snapshot semantics and consistency in your backtest/production context.
  • Non-deterministic outputs: LLM reasoning is probabilistic. Use persistent decision logs, structured outputs, and repeated runs to assess stability; not suitable for deterministic financial workflows.

When to avoid it — and what to weigh

  • Live Production Trading — Framework is explicitly designed for research. No guarantees on trading performance, regulatory compliance, or financial safety. LLM outputs are non-deterministic and context-dependent.
  • Regulated Financial Services — No indication of compliance audit, regulatory certifications, or institutional risk controls. Using LLM agents for actual financial decisions requires legal and compliance review.
  • Low-Latency / High-Frequency Trading — Agent reasoning and LLM inference introduce latency; not designed for millisecond-scale trading or HFT workloads.
  • Minimal Operational Overhead Required — Requires managing multiple API keys, LLM provider integrations, data vendor subscriptions, and careful configuration of model parameters and retry budgets.

License & commercial use

Apache License 2.0 (Apache-2.0). Permissive OSI-approved license allowing commercial use, modification, and distribution with attribution and liability disclaimer. No patent protection or warranty.

Apache-2.0 permits commercial use. However, the project is explicitly designed for research and includes a disclaimer that it is not financial, investment, or trading advice. Any commercial application (prop trading firm, asset manager, fintech) must conduct independent legal and compliance review, implement regulatory controls, and accept full liability for trading losses. The license does not convey fitness for regulated financial services.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityModerate
DEV.co fitGood
Assessment confidenceHigh
Security considerations

Multiple considerations apply: (1) API keys for multiple LLM providers and data vendors are required; use environment variables and secrets management, not hardcoded. (2) Ticker path-traversal hardening mentioned in v0.2.5 suggests prior vulnerability; verify latest patches. (3) LLM outputs are not cryptographically signed or audited; agents can generate plausible-sounding but incorrect financial advice. (4) No mention of input validation, prompt injection defenses, or adversarial robustness. (5) Data from sentiment sources (Reddit, StockTwits) and news APIs may be stale, manipulated, or unreliable. (6) Backtested performance does not guarantee live performance or safety.

Alternatives to consider

OpenBB Agent (OpenBB)

Simpler agent-based framework for market research; less multi-agent orchestration but smaller operational surface. Focuses on data aggregation and reporting rather than autonomous trading.

NVIDIA Nemo Guardrails + LLM Backbone

General-purpose multi-agent framework with more robust safety and guardrail controls. Requires manual composition of trading agents but offers stronger compliance and audit trails.

Backtrader + Manual LLM Integration

Battle-tested backtesting library with human-written decision logic. More overhead than TradingAgents but avoids LLM-driven non-determinism and is widely used in prop trading.

Software development agency

Build on TradingAgents with DEV.co software developers

TradingAgents is ideal for quant researchers and trading teams exploring LLM-driven strategies. Start with backtesting on historical data, audit agent decision logs, and gradually integrate with your risk infrastructure. Consult legal and compliance before live trading.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

TradingAgents FAQ

Can I use TradingAgents to trade real money?
The project disclaims use as financial, investment, or trading advice and is designed for research. Live trading would require independent legal/compliance review, robust risk controls, and acceptance of full liability. Not recommended without institutional backing and regulatory approval.
Which LLM should I use for best trading performance?
Unknown from the data. Framework supports GPT-5.x, Gemini 3.x, Claude 4.x, Grok 4.x, DeepSeek, and open models. Performance depends on model capability, temperature, training data freshness, and market conditions. Backtesting with your chosen provider is required.
How do I integrate TradingAgents with my existing trading platform?
Framework outputs structured decisions (analyst reports, trader proposals, risk assessments) as JSON. You must build integration logic to parse these, validate them, and route to your execution engine. No off-the-shelf connectors to major brokers are mentioned in the data.
What data sources does TradingAgents support?
Confirmed: Alpha Vantage, FRED, Polymarket, StockTwits, Reddit, news APIs. Framework also mentions sentiment and macro indicators. Full vendor list and API status are not provided; refer to the documentation and GitHub issues for current support.

Custom software development services

From first prototype to production, DEV.co delivers software development services around tools like TradingAgents. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across ai frameworks and beyond.

Ready to experiment with multi-agent trading?

TradingAgents is ideal for quant researchers and trading teams exploring LLM-driven strategies. Start with backtesting on historical data, audit agent decision logs, and gradually integrate with your risk infrastructure. Consult legal and compliance before live trading.