Skip to main content
Glama

start_test

Runs one of QuanterLab's ready tests on a stock or fund as a circuit of the lab's cards: the rule is tuned on past years only, then tested on the last three years, which the tuning never saw, against simply holding the same stock. It takes no settings to start: each test has its own default stock, below, and a test picked from get_started's card or named in words runs as it is.

    Tests: mean_reversion (a fitted model that buys when the price has fallen unusually far below its recent
    level; default KO), trend_following (a model that rides a trend while it lasts; SPY), momentum (holds
    while the last few months were up; QQQ), buy_the_dip (buys after a sharp fall, the 7-day RSI under 30; SPY), famous_rule (the 200-day line: hold above the 200-day average, cash below; SPY), and mvo
    (a book of up to 20 stocks of an index picked each quarter by mean-variance optimisation, from the companies
    in the index on that date; it takes `index`, not a ticker, and is set against holding the index's fund).

    Its card draws the circuit as the lab does and lights each card as it runs, with the result under it.
    get_primitives_run_progress follows it by its task_id; when it has finished it answers with the result,
    report_url and offers, each with its exact call, the first one tracking the test live
    (start_paper_trade). The circuit card shows the result, so a ready test needs no show_report_card.

    Every account runs it within a daily allowance; a refused call does not count, and get_started and
    every run's answer give today's count.

    A run that takes more than two minutes does not start at once: the answer gives its time and why, the
    person is asked ("This one takes about 25 minutes (20 years of the S&P 500). Run it?"), and the same call
    with confirm: true starts it after their yes.

    Args:
        test: mean_reversion, trend_following, momentum, buy_the_dip, famous_rule or mvo.
        ticker: Optional. The stock or fund to test it on, as its symbol ('AAPL', 'SPY'); else the test's own.
        index: Optional, for mvo: nasdaq100 (its own) or dow30. The book on the S&P 500 does not run here
            yet (more than a run here does in full in under a minute), and another index (the DAX, the
            FTSE 100) has no dated members: pass what the person asked all the same, and QuanterLab answers
            with its note and the same book on the Nasdaq 100, one tap away, and nothing runs.
        settings: Optional. The one setting a test lets a person change, by name: mean_reversion
            {"z_entry": 1 to 3.5}, trend_following {"trend_threshold": 0.25 to 3}, momentum {"months": 1 to
            6}, buy_the_dip {"rsi_below": 10 to 40}, famous_rule {"days": 20 to 300}, mvo {"holdings": 5 to
            30}, e.g. {"days": 100} for
            the 100-day line; and on every test {"stop_loss": 0.5 to 30}, the lab's Stop Loss card added
            under the forward test, which sells everything once the run has lost that many percent since the
            start of the month. A test changed by talking ("use the 100-day line instead", "add a stop
            loss") is the same call: the same test and stock and the last run's settings with that one change;
            the card draws the changed circuit and runs it. Without settings the test runs as it stands.
        question: Optional. One sentence restating what the run should answer, not the person's message: the
            title of the report, and kept only as that.
        study: Optional, instead of test. A published paper's id (from search or fetch). Unchanged, it answers
            at once with the paper's registered result: the run its last registered window recorded, from that
            window's date to the day it was recorded, and it uses none of today's runs. Changed (its settings)
            or with to_today, it runs the paper's circuit from that window's date to today. It takes no ticker;
            its `settings` are the settings the paper names (fetch lists them, e.g. {"symbol": "QQQ"} for which
            fund, {"spread_bps": 20} for the trading cost), and a refusal names them.
        to_today: Optional, with study. True runs the paper's circuit unchanged from its last registered window
            to today, a run like any other: what the person asks for with "run it up to today", or a tap on
            that offer after a paper's registered result.
        confirm: Optional. True only after the person said yes to the time a first call of this run gave.
    

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "start_test"
  }
}

Input Schema

NameRequiredDescriptionDefault
testNo
indexNo
studyNo
tickerNo
confirmNo
questionNo
settingsNo
to_todayNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • addedInput schema / properties / confirm
      Added value: +{
      +  "anyOf": [
      +    {
      +      "type": "boolean"
      +    },
      +    {
      +      "type": "null"
      +    }
      +  ],
      +  "default": null,
      +  "title": "Confirm"
      +}
  2. First observed

TDQS

A4.7/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description adds substantial behavior beyond the annotations: daily allowance and refusal handling, two-minute confirmation flow, task_id progress tracking, returned result/report_url/offers, and study-mode registered-result behavior. These details are consistent with readOnlyHint=false, destructiveHint=false, idempotentHint=false, and openWorldHint=true.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is long but mostly front-loaded and earns its length by covering multiple modes and eight parameters. There is some repetition around defaults and dense parenthetical phrasing, which keeps it from being maximally concise.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For an eight-parameter, multi-mode tool with no output schema, the definition is complete: it explains run behavior, confirmation, progress tracking, returned fields, daily limits, and study-mode variations. Nothing essential for correct invocation appears to be missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 0%, so the description must carry all parameter meaning, and it does: it documents all eight parameters, test names and defaults, index restrictions, settings ranges, study semantics, question handling, and confirm usage. It adds far more meaning than the bare schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and resource ('Runs one of QuanterLab's ready tests') and immediately defines the stock/fund scope and circuit mechanics. The six named tests, their logic, and their defaults make the purpose unambiguous and distinguish it from a generic backtest.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives clear context for starting a ready test, using study instead of test, and applying confirm or to_today, and it notes that progress is tracked via get_primitives_run_progress and that no show_report_card is needed. It does not explicitly state when not to use this tool versus a custom-backtest sibling, so it falls short of full when/when-not guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources