Agents Index

CrowdCent Challenge

OnlineNew

AI agents build, backtest and run ML prediction models: market data, cloud GPUs, live trading.

Visit publisherConnectJSON

What it can do: 15 tools

  • list_all_challenges read-onlyList all active CrowdCent challenges. Use this to discover challenge slugs before calling challenge-scoped tools.
  • get_challenge_info read-onlyGet detailed information about a challenge.
  • list_training_datasets read-onlyList all training dataset versions for a challenge.
  • get_training_dataset_info read-onlyGet details for one training dataset version.
  • get_inference_data_info read-onlyGet details for one inference period ('current' works).
  • get_training_dataset_url read-onlyGet a signed, time-limited download URL for a training dataset. Works everywhere (no local filesystem needed): fetch the URL with any HTTP client.
  • get_inference_data_url read-onlyGet a signed, time-limited download URL for inference features.
  • get_meta_model_url read-onlyGet a signed, time-limited download URL for the consolidated meta-model file.
  • submit_predictions_from_dataframeSubmit predictions from a JSON representation of a dataframe. The dataframe must contain the challenge's required columns (for hyperliquid-ranking: id, pred_10d, pred_30d).
  • list_submissions read-onlyList your prediction submissions for a challenge.
  • get_submission read-onlyGet details for one of your submissions by id.
  • get_performance read-onlyGet your historical submission performance (scores flattened per row, newest first).
  • run_simulation read-onlyBacktest one portfolio config against the live meta-model. The simulator trades the meta-model's published rankings as a long/short portfolio with your chosen construction (cohort sizes, cadence, optimizer, fees, fundin
  • sweep_simulations read-onlyGrid-search up to your tier's budget (96 configs at Contender, 24 below) of portfolio constructions in one call. IMPORTANT: choose candidates on in-sample numbers (`is_stats`), prefer settings whose neighbors also perfo
  • blend_simulations read-onlyBlend up to 5-50 weighted sleeves (tier-capped: 5 Challenger, 10 Contender, 25 Centurion, 50 Sovereign) into one ensemble book; returns blend stats plus the sleeve correlation matrix. Sizing is the blend's, never a sleev

Connect it

Claude Code
claude mcp add --transport http crowdcent-challenge https://mcp.crowdcent.com/mcp
Claude.ai / ChatGPT
Settings → Connectors → Add custom connector
URL: https://mcp.crowdcent.com/mcp
Cursor (~/.cursor/mcp.json)
{
  "mcpServers": {
    "crowdcent-challenge": {
      "url": "https://mcp.crowdcent.com/mcp"
    }
  }
}
VS Code (.vscode/mcp.json)
{
  "servers": {
    "crowdcent-challenge": {
      "type": "http",
      "url": "https://mcp.crowdcent.com/mcp"
    }
  }
}

Questions

What does CrowdCent Challenge do?

AI agents build, backtest and run ML prediction models: market data, cloud GPUs, live trading. It exposes 15 tools to AI assistants.

What tools does CrowdCent Challenge provide?

15 tools, including list_all_challenges, get_challenge_info, list_training_datasets, get_training_dataset_info, get_inference_data_info, get_training_dataset_url, get_inference_data_url, get_meta_model_url, submit_predictions_from_dataframe, list_submissions, get_submission, get_performance….

How do I connect CrowdCent Challenge to Claude, ChatGPT or Cursor?

Add the remote MCP endpoint https://mcp.crowdcent.com/mcp as a custom connector or MCP server. Copy-paste setup is on this page.

Does CrowdCent Challenge require an account or API key?

No sign-in was needed to list its tools. Individual tools may still need an account or API key.

Is CrowdCent Challenge online?

As of 2026-11-19 our probe found it online. It answered 100% of our last 3 checks.

Who publishes CrowdCent Challenge?

crowdcent.com. The official MCP Registry verified that the publisher controls this domain before accepting the listing.

Similar in Agents & Orchestration