Build on Tradingale data
Tradingale computes descriptive metrics across 250+ crypto pairs and 100+ US stocks, and serves them two ways: a REST API and an MCP server. Your AI builds the software. What you build is yours.
Recipes
Not a blog. Each recipe is a complete, reproducible path from a blank folder to running software.
The Tradingale Runner
The bot, back as open-source software you run yourself: one-click Railway deploy or one prompt to your AI, the Tradingale dashboard, paper by default, free on BTC. Live Kraken stays behind an explicit flag with your keys on your side.
Read the recipe →REST + MCPYour own companion, in one evening
Want something custom instead of the Runner? A personal dashboard built by your AI from a single prompt, with the full field reference. The DIY door to the same data.
Read the recipe →EngineeringThe Sequence Automation Handbook
The production engineering playbook from our execution era, paired with martingale-kit, our open-source engine on GitHub: exact ladder formula, one-active-sell state machine, price grids, paper simulator. Your agent only writes the venue adapter.
Read the recipe →MCPConnect your AI in 60 seconds
The connection reference for every MCP client: claude.ai custom connectors (OAuth, no token), Claude Code and Desktop, Cursor, ChatGPT, and the secret-URL option for clients without OAuth. Seven read-only tools.
Read the recipe →How it works
Get a token
Tokens live in Settings > API. One bearer token covers both the REST API and the MCP server, on the same monthly quota.
Point your AI at the data
Agents connect through MCP; code talks plain JSON over REST. Every instrument carries the same descriptive metrics: Martingale Score, Startingale, component scores, and the model's structural parameters.
Own what you build
The software your AI writes runs on your machine, under your control. Tradingale serves the data and stops exactly there.
Why we stick to the data
Running order execution well is its own full-time craft: every venue has its own order types, tick sizes, rounding grids, rate limits and API quirks, and they all change without notice. That plumbing is not our trade. Our trade is martingale intelligence: two descriptive metrics and the model structures behind them, computed the same way for everyone, every day. That is what we sell.
How you integrate this data into your own strategy and tooling is entirely up to you. Some read it on the scoreboard, some journal snapshots, some wire their own agents to the API or the MCP server with Claude, ChatGPT, Gemini, Cursor or anything else that can read JSON. Your stack, your rules, your responsibility.
Tradingale never places orders, holds funds, or gives advice. What you build and run is yours.
Data for informational purposes only. Not investment advice.