Claude
- Open Manifold-BT Backtest in the Claude directory ↗
- Click Connect
- No sign-in needed: ask away

Manifold-BT Backtest turns a question about a trading rule into a research workflow that your AI runs in its own code environment.
Step 1 installs the Manifold-BT backtesting engine from PyPI. Each following step runs there, on the bars you upload (CSV or Parquet) or on a synthetic series, and prints a few numbers. The connector checks them against a written gate (enough trades, a Sharpe ratio distinguishable from zero, a result that survives doubled costs and a one-bar delay, a parameter choice that beats what the same search finds in noise) and hands over the next step, a retry with the reason, or the verdict.
Five workflows: test an idea; tune parameters by keeping the stable region instead of the single best cell; validate out of sample; size exposure when the edge is an estimate; diagnose a result that looks too good.
The connector computes nothing and stores nothing: your data and strategy stay in your AI's environment. No account is needed. Code execution with network access to package managers must be enabled in your AI. A backtest measures a rule on past data; it is not a forecast and not investment advice.
A backtest that looks too good, or a live result that diverges from its backtest.
https://mcp.manifoldbt.com/backtest/mcphttps://mcp.manifoldbt.com/backtest/mcpmcp.manifoldbt.comdiagnose_resultA backtest that looks too good, or a live result that diverges from its backtest.size_exposureA measured edge to turn into an exposure, given the edge is estimated, not known.test_ideaA rule or idea to check once, honestly, before any tuning.tune_parametersParameters to choose: map the surface, bound the stable region, pool it, and correct for the search.validate_out_of_sampleA rule or a tuned region to confirm on data it was not fitted on.The AI reads these server instructions before using this connector, every time.
Step-by-step backtesting research with the manifoldbt engine (0.35.2), run in the client's code environment. returns a workflow whose first step installs manifoldbt; each step's code prints a RESULT line, and applies that step's gate to it and returns the next step. returns passages of the engine reference and of the research method; maps an engine error message to its cause and fix; lists the licence tiers and Community limits. The server holds no engine and no market data: backtests run in the client: pip install manifoldbt in Claude's code sandbox or on the user's machine, on the user's own data (an uploaded CSV, a DataFrame, or a store ingested locally). Nothing here is investment advice: a backtest measures a rule on past data.
check_stepget_capabilitiessearch_docsexplain_errorstart_researchNone in the store listings or on the server
Manifold-BT Backtest turns a question about a trading rule into a research workflow that your AI runs in its own code environment. Step 1 installs the Manifold-BT backtesting engine from PyPI. Each following step runs there, on the bars you upload (CSV or Parquet) or on a synthetic series, and prints a few numbers.
Open its listing in the app's connector directory and connect. No sign-in is needed. Step-by-step setup for each app is on this page.
The Manifold-BT Backtest MCP server URL is https://mcp.manifoldbt.com/backtest/mcp. Paste it as a custom connector in Claude, ChatGPT, Gemini or Grok, or add it to Claude Code, Cursor or any MCP client. Setup steps.
Yes. Manifold-BT Backtest is published or run by Manifold-BT, so it is the official MCP server. See other official connectors.
It is listed in 1 store: Claude. Apps where it is not listed can usually still add it as a custom connector.
It has 5 tools: 3 read data and 2 change data. See every tool and its inputs.
It can change data but not delete it. 2 tools can create or update things, for example:
None that we found. None in the store listings or on the server.
Here are examples from its store listing. Type one into your AI after you connect.
No. Its answers come back as text in the chat. Connectors with interactive UI.
This page updates every day. To ask about Manifold-BT Backtest from Claude, ChatGPT or a coding agent, connect the Agent Ready MCP server: it tracks ranks, tools and new listings across the AI plugin stores.