What problem it solves

Complex trading decisions require evaluation across financial fundamentals, market sentiment, news, technical indicators, portfolio holdings, and risk. TradingAgents addresses this by decomposing trading tasks into specialized LLM-powered roles that discuss findings and assess proposed transactions.

Key features

• Parallel analyst research — Fundamental, sentiment, news, and technical analysts work concurrently with their own tools. Their reports are collected before the research debate begins.

  Source: README

• Structured debate and decision-making — Bullish and bearish researchers assess analyst findings, while the trader composes the reports into a proposed trading decision. The risk team evaluates the strategy, and the Portfolio Manager approves or rejects it.

  Source: README

• Multiple model providers — The framework supports several hosted and local model providers, including OpenAI, Google, Anthropic, DeepSeek, Ollama, and Azure OpenAI. Quick-thinking and deep-thinking model tiers can also use different providers.

  Source: README

• Point-in-time financial data — Historical runs can read US company statements as they stood on the analysis date, including first-reported figures and filing availability. Other market data is also handled according to what was published at the time.

  Source: README

• Persistence, recovery, and backtesting — Completed decisions are retained in a memory log, and optional LangGraph checkpoints allow interrupted runs to resume. The package also supports evaluating decisions across ticker and date grids.

  Source: README

Quick start

Copied verbatim from the project's README.

conda create -n tradingagents python=3.13
conda activate tradingagents

pip install .

from tradingagents.graph.trading_graph import TradingAgentsGraph
from tradingagents.default_config import DEFAULT_CONFIG

ta = TradingAgentsGraph(debug=True, config=DEFAULT_CONFIG.copy())

# forward propagate
state, decision = ta.propagate("NVDA", "2026-09-01")
print(decision)

# the same report tree the CLI saves, under results_dir/reports
ta.save_reports(state, "NVDA")

Who it may suit

This project may suit researchers, developers, and financial software practitioners who want to study or extend multi-agent approaches to market analysis and simulated trading workflows.

Topics to discuss

• Which market-data sources and point-in-time rules should be added or audited next?

• How should agent debate rounds and model assignments be evaluated for research quality?

• What portfolio and reporting workflows would be most useful for framework adopters?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/TauricResearch/TradingAgents#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.