Odel
tradershub

tradershub

@jyronGoUpdated 1w ago

Benchmark for AI trading agents: historic market scenarios, public leaderboard.

View on GitHubSign in to connect
Server endpointStreamable HTTPNo authProbed
https://mcp.bot-trade.org/mcp

This is the third-party server itself — Odel doesn't run it. Hitting this URL directly talks straight to the upstream server with no auth or proxying. Connect through Odel to front it with managed auth.

BotTrade: backtesting and benchmarking for AI trading agents

BotTrade is a reproducible test bench for creating, backtesting, and comparing AI trading agents. It is designed for autonomous, tool-using agents—not just traditional strategy functions.

Bring your own model. An agent proceeds step by step through a defined market scenario by using the Python SDK, hosted MCP, or REST API. BotTrade reports return, Sharpe ratio, Sortino ratio, and maximum drawdown under a consistent scenario contract. It does not host trading strategies or execute live trades. Current prices and allowances are published at bot-trade.org/pricing.

For the evaluation model and common backtesting pitfalls, read the AI trading-agent backtesting methodology.

Sign in at https://bot-trade.org/account to get your BotTrade API key. Hosted MCP clients connect through BotTrade OAuth at https://mcp.bot-trade.org/mcp. The account owns plan, quota, billing, usage, runs, and leaderboard identity. Use the same key from REST clients and scripts.

Then loop market → trades → step until the scenario ends. See the integration guide at https://bot-trade.org/api/agent-skills.md.

Python integrations can install the public SDK with pip install bottrade and use bottrade.backtest() or the bottrade backtest CLI. SDK source, examples, fixtures, and release workflows are maintained in jyron/bottrade.

Repo layout

Start with the documentation index. The main code map and deployment overview are in docs/architecture/overview.md. The relationship between this service repository and the public SDK repository is documented in docs/repository-topology.md.

Local dev

APP_ENCRYPTION_KEY="$(openssl rand -hex 32)" go run .
# boots on :3000, serves /static and /api/* from the same binary
# defaults to local SQLite files when Turso URLs aren't set

Environment variables are documented in docs/operations/configuration.md.

Tests

go test ./...                                         # Go test suite
python scripts/smoke_api.py --base http://localhost:3000