mcpField
B
backtestport

backtestport

Local
github.com/CloudCorpRecordsFinance0Updated 1w ago

About backtestport

backtestport is an MCP server and CLI client for the hosted Backtest360 engine, aimed at strategy developers. It validates and compares strategies, runs backtests, evaluates signals, and retrieves market and macro data; it does not place trades.

Key features

  1. 01Validate strategy documents
  2. 02Run and compare backtests
  3. 03Evaluate latest signals
  4. 04Fetch market and macro data
  5. 05Browse indicators and templates

Use cases

  1. 01Check a strategy before running it
  2. 02Compare strategies on the same data
  3. 03Inspect a symbol’s price history
  4. 04Look up macroeconomic series

MCP support

Exposes Backtest360 engine, strategy, backtest, market-data, and macro-data operations as tools to MCP clients.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"cloudcorprecords-backtestport": {
"command": "npx",
"args": [
"-y",
"@reneturcios/backtestport"
]
}
}
}
Run commandnpx -y @reneturcios/backtestport

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npm@reneturcios/backtestport0.1.0
ocighcr.io/cloudcorprecords/backtestport:0.1.0

Tools

19

Saved tool list · updated 1w ago

This tool takes no parameters.

Details

From the project README

backtestport

Backtest a trading strategy without standing up your own engine. backtestport is a small client for the Backtest360 hosted backtesting engine: you hand it a strategy document and it validates the document, runs the backtest, evaluates the latest signal, compares several strategies on the same data, and fetches the market and macroeconomic data the engine can work with — plus the indicator catalog, templates and reference data you need to write the strategy in the first place.

This is a backtesting client only. It has no live-trading, order-placement or account tools, and nothing it does moves money.

Responses come back exactly as the engine returned them: nothing is trimmed, renamed or summarised on the way through, apart from the output byte cap described below.

Configuration

Four environment variables control the client:

  • BACKTEST360_API_KEY (required) — sent on every request as the X-API-Key header. When it is not set, every tool fails with BACKTEST360_API_KEY is not set before any request is made.
  • BACKTEST360_ENGINE_URL (optional) — engine base URL. Defaults to https://api.backtest360.com.
  • BACKTEST360_MCP_TIMEOUT (optional) — per-request timeout in seconds, default 300. A request that runs past it fails with backtest360 request timed out.
  • BACKTEST360_MCP_MAX_OUTPUT_BYTES (optional) — response size cap in bytes, default 100000. A longer body is truncated to the cap and a final line {"truncated_by_mcp": true} is appended, so a huge equity curve cannot flood the caller.

Reading the results

Two engine answers that look like failures are not:

  • Strategy validation returns {valid: false, errors: [...]} when the document has problems. That is a normal result — the errors are structured and locatable so you can fix the document.
  • A backtest the engine declines returns {accepted: false}. That is also a normal result.

Actual failures are errors: HTTP 429 fails with backtest360 rate limited (HTTP 429); retry after <value> when the engine sends a Retry-After header (without the suffix when it does not), any other non-2xx reply fails with backtest360 error: HTTP <status>, and an unreachable engine fails with backtest360 unreachable.

Strategy arguments (strategy, and each item of strategies) accept either a JSON object or a JSON string, which is parsed before sending; a string that does not parse fails with strategy is not valid JSON.

Three arguments shape a backtest run:

  • response_detail — summary (default), stats or full: how much of the verdict comes back.
  • max_series_points — cap on returned series length, default 500; it must be an integer of at least 2, checked locally before any request (max_series_points must be an integer >= 2).
  • include — any of trades, equity_curve, monthly_returns, yearly_returns and signal_diagnostics, each adding its series to the verdict.

Tools

Discovery and account:

  • engine_info — engine version, API contract, health.
  • get_me — what the configured key can do: permission scopes, limits, current usage, capability flags.
  • get_catalog — the reference catalogs (operators, execution modes, stop types, sizing methods, bar frequencies, metric sections); pass catalog to fetch just one.
  • list_indicators — indicator discovery with per-indicator parameter schemas; search narrows the list.
  • list_templates — predesigned strategy templates, listed compactly, or one in full when id is given.
  • get_strategy_schema — the JSON Schema for strategy documents.

Strategies and runs:

  • validate_strategy — check a strategy document without running it.
  • run_backtest — run a backtest; symbol, start, end, the shaping arguments above, and data (inline bars as an object) are all optional.
  • get_latest_signal — evaluate the most recent bar only; no P&L.
  • compare_backtests — run several strategies on the same data, side by side.
  • compute_stats — compute the metric set from an externally produced returns series.

Market and macro data:

  • search_tickers / list_tickers — the fetchable symbol universe.
  • get_data_range — available history and a bar-count estimate for a symbol.
  • get_ticker_info — symbol identity and data coverage in a single call.
  • get_quote — latest available price for a symbol.
  • get_price_history — OHLCV history over a date range (start, end, frequency; with server-side data the engine fixes bars at 1d — the argument is accepted and forwarded).
  • list_macro_series / get_macro_series — the macroeconomic series catalog, and one series' observations over an optional date range.
Endpoint provenance

The original project's README pins the 19 tools, the shaping-argument semantics, the four environment variables with their defaults, the X-API-Key header and the rule that validation results and engine rejections are data. The wire paths below are reconstructions, labelled as such: GET /engine/info, GET /me, GET /catalogs, GET /indicators, GET /templates[/{id}], GET /strategy/schema, POST /strategy/validate, POST /backtests, POST /signals/latest, POST /backtests/compare, POST /stats, GET /tickers/search, GET /tickers, GET /tickers/{symbol}/data-range, GET /tickers/{symbol}, GET /tickers/{symbol}/quote, GET /tickers/{symbol}/history, GET /macro, GET /macro/{series} — all under the configured base URL, with path segments and query values URL-encoded.

CLI

Each tool is a subcommand taking one JSON argument string:

npx @reneturcios/backtestport engine_info '{}'
npx @reneturcios/backtestport validate_strategy '{"strategy":{"name":"sma-cross","rules":[]}}'
npx @reneturcios/backtestport run_backtest '{"strategy":{"name":"sma-cross"},"symbol":"AAPL","include":["equity_curve"]}'
npx @reneturcios/backtestport get_quote '{"symbol":"AAPL"}'
npx @reneturcios/backtestport --version

Output is the tool's result text (the upstream JSON on success). Exit codes: 0 on success, 1 on a runtime error (including a missing key, a validation failure or an upstream failure), 2 on a usage error.

MCP server

npx -y @reneturcios/backtestport mcp

MCP client config:

{
  "mcpServers": {
    "backtestport": {
      "command": "npx",
      "args": ["-y", "@reneturcios/backtestport", "mcp"],
      "env": {
        "BACKTEST360_API_KEY": "<your key>"
      }
    }
  }
}

Docker

docker run -i ghcr.io/cloudcorprecords/backtestport
docker run --rm ghcr.io/cloudcorprecords/backtestport get_quote '{"symbol":"AAPL"}'

Development

npm install
npm run build   # tsc -> dist/
npm test        # tsx --test test/*.test.ts (fake fetch only, no network)
npm run dev     # run the CLI from source

License

MIT — Copyright (c) 2026 Rene Turcios

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.