AI trading bot: TradingAgents setup (source 9982)

Summary: TradingAgents is an open-source framework that mirrors a real trading firm (Tauric Research, Apache-2.0, 120K GitHub stars). Four analysts (fundamentals, news, sentiment, technical) write reports, a bull and a bear researcher debate, a trader decides, a risk team reviews, and a portfolio manager approves or rejects. Approved orders go to a simulated exchange: no real money. The authors say it is "designed for research purposes" and "not intended as financial, investment, or trading advice". Neither is this guide.

1. Install (Python 3.11 or later)

git clone https://github.com/TauricResearch/TradingAgents.git
cd TradingAgents
conda create -n tradingagents python=3.13
conda activate tradingagents
pip install .

Without conda, with uv:

uv venv --python 3.13
source .venv/bin/activate
uv pip install .

2. Keys

One model provider is enough; the Alpha Vantage data key is free (25 requests a day, alphavantage.co/support).

export XAI_API_KEY="..."            # Grok (or ANTHROPIC_API_KEY, OPENAI_API_KEY, DEEPSEEK_API_KEY)
export ALPHA_VANTAGE_API_KEY="..."  # free key
export FRED_API_KEY="..."           # optional, free (macro data)

SEC EDGAR needs no key. The software is free; Grok / Claude / GPT API usage is billed by the provider.

3. Run

tradingagents

You pick the ticker, the analysis date, the analysts, the model provider and the research depth; the result is saved as complete_report.html. Without prompts:

tradingagents --ticker NVDA --date 2026-11-05 --analysts market,news,fundamentals

4. Fully free: a local model (Ollama)

ollama pull <model-name>

Set llm_provider: "ollama"; the default endpoint is http://localhost:11434/v1 (OLLAMA_BASE_URL for a remote server). For a model not in the list, choose "Custom model ID" in the CLI. OpenAI-compatible endpoints such as LM Studio work too.

5. With Docker

cp .env.example .env        # put your keys in .env
docker compose run --rm tradingagents
# with Ollama: docker compose --profile ollama run --rm tradingagents-ollama

6. Second tool: AI Hedge Fund (virattt, MIT, 64.5K stars)

pipx install aihf
aihf

On first run it asks for a Financial Datasets API key and one model key. It has a paper-trading mode and a backtest mode; "the system does not actually make any trades", educational only.

7. The paid option: Grok Bot (xAI)

x.ai/bot: Premium $30 a month (the bot's own cloud computer, weekly usage included), Super $100, Ultra $200. You can put several bots in one chat so they pass work to each other. Trading is not an official feature; people on X build those setups themselves. On November 1, 2026 Musk said "the cost of serving Grok Bot was unsustainably high relative to revenue" and weekly limits were cut.

Warning

Profit claims such as "$90 to $14,220 in a week" come from a single X post; nobody has verified them. Test in simulation first, never with real money.

AI Trading Desk