QM Screener documentation

7 momentum screens run over the full US equity market every trading day after the close. This page is how to reach that data from outside the browser. The short answer is MCP: add one URL to an AI assistant and it can read the screens, the sector ranking and — on a subscriber account — every passing name and its chart data. Everything is also available over plain HTTP for scripts.

Connect an AI assistant (MCP)

The MCP server is at:

https://qmscreener.com/mcp

It speaks MCP over Streamable HTTP, which is what Claude, ChatGPT and other assistants mean by a remote MCP server. Three steps:

• Paste that address into your assistant's MCP or connectors settings.
• It will send you back here to sign in.
• Approve the access it asks for, and the tools below appear in the assistant.

There is nothing to install and no API key to paste: the assistant registers itself and gets a token through OAuth 2.1 when you approve it.

Tools

• list_screens — List the momentum screens this site runs and what each one tests for.
• todays_top_setups — The highest-ranked passing tickers per screen from the most recent run, with the total that passed.
• sector_strength — Trailing returns for 30 US sector, industry and theme ETFs measured against SPY.
• market_breadth — How many names pass each screen per session, as a daily series beside SPY, QQQ and IWM closes. History before the feature shipped was replayed against today's universe, so names that have since delisted are missing from those sessions.
• screen_results — One screen's full result set — every passing ticker with the metric values that decided it, plus the near-miss and value lists and the run's SPY benchmark return. Subscriber account.
• price_history — Daily OHLCV bars for one US ticker. Subscriber account.

On a free account an assistant can read the daily top setups, the sector strength ranking and market breadth. Full screen results and price history need a subscription — see https://qmscreener.com/pricing.md.

Market breadth history before the feature shipped is a replay: History before the feature shipped was replayed against today's universe, so names that have since delisted are missing from those sessions.

Server card

https://qmscreener.com/.well-known/mcp/server-card.json describes the endpoint, the transport and this same tool list, for a client that reads a card before it connects.

For scripts

The same data is available over HTTP for code that is not an assistant: the machine-readable spec is at https://qmscreener.com/openapi.json (OpenAPI 3.1), how a caller obtains a token is at https://qmscreener.com/auth.md, and https://qmscreener.com/llms.txt is a model-readable summary of the whole site. The API is the same product behind the same paywall — a caller is served exactly what the account behind its token is served in the browser, and no more.

Errors

Every failure is JSON, never an HTML page:

{
  "error": "subscription_required",
  "code": "subscription_required",
  "message": "This endpoint needs an active subscription.",
  "hint": "Start a 7-day free trial at https://qmscreener.com/subscribe.",
  "docs": "https://qmscreener.com/docs#errors"
}

code is the stable identifier to branch on; message and hint are for humans and for an agent explaining itself.

Rate limits

Responses carry the RFC 9331 RateLimit header, and a 429 carries Retry-After in seconds. Self-throttle on those rather than on a fixed sleep.

Agent skills

https://qmscreener.com/.well-known/agent-skills/index.json indexes task recipes: todays-momentum-setups, screen-methodologies, ticker-price-history, market-breadth, momentum-watchlist.

Data and limits

End-of-day US equities, refreshed once per trading day after the close. No intraday, no fundamentals, no news. Every filter comparison is strictly greater-than, and every row carries the values that decided it, so any result can be audited without re-running the screen. Not financial advice.

The criteria behind the screens, and the thresholds each one uses, are explained in how to screen for momentum stocks.

Contact

support@qmscreener.com · https://qmscreener.com/contact