Skip to main content
Back to challenges
Challenge

Agent for Autonomous Market Intelligence

This challenge tasks developers with building a sophisticated agentic system for real-time financial market analysis. The system will leverage the advanced reasoning capabilities of Gemini 2.5 Pro, particularly its 'Deep Think' mode, to process diverse financial data streams. Using LangGraph, you will orchestrate a dynamic, graph-based workflow where agents autonomously analyze news, company reports, and prediction market data (like from Kalshi or Polymarket, simulated) to generate actionable investment insights and risk assessments. The core of this challenge lies in implementing a hybrid reasoning system: instant processing for rapid market reactions and deep, extended thinking for complex economic forecasting. Agents will utilize the Model Context Protocol (MCP) for seamless tool integration, allowing them to fetch real-time stock prices, analyze financial statements via RAG, and query prediction markets. The goal is to create a robust, self-improving financial analyst capable of identifying opportunities and threats with high precision.

Special Purpose AgentsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a graph-based multi-agent system using LangGraph with Gemini 2.5 Pro's Deep Think mode for real-time financial market insights and risk assessment.

Delivery guide

How work is evaluated

Evaluation

The system will be evaluated on its ability to accurately analyze real-world financial scenarios, provide well-reasoned justifications, and effectively utilize its toolset. The quality of the LangGraph workflow and the effective use of Gemini's Deep Think mode will be key.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Learning goals

What you should walk away with

  • Design hybrid instant/deep reasoning systems for rapid and in-depth financial insights.

  • Leverage Gemini 2.5 Pro's Deep Think mode for advanced market reasoning and forecasting.

  • Implement MCP-enabled tool integration for diverse financial data sources.

  • Create agents capable of autonomous hypothesis generation and testing in financial markets.

  • Build dynamic, graph-based agent workflows using LangGraph for financial analysis.

Resources and assets

Reference links and supporting material

Dataset notes

Sample data for 1 tasks

Challenge setup

The system will be evaluated on its ability to accurately analyze real-world financial scenarios, provide well-reasoned justifications, and effectively utilize its toolset. The quality of the LangGraph workflow and th...

Challenge input

{ "event_description": "<news headline or brief summary>", "company_ticker": "<STOCK_TICKER>", "event_date": "YYYY-MM-DD" }

Agent execution

No agent tool is listed for this challenge.

Evaluated output

{ "predicted_movement": "<UP|DOWN|FLAT>", "magnitude_percent": "<float>", "confidence_score": "<0-1.0>", "rationale": "<detailed justification>", "...

Evaluation checks
  • Output is a valid JSON object matching the schema.
  • Agent successfully called at least 2 external tools (e.g., news API, stock API, prediction market).
  • Evidence of a multi-step LangGraph execution (e.g., trace logs showing multiple nodes/agents activated).
Configured evaluation evidence
  • PredictionAccuracy target: 0.75
Configured runtime
  • Python execution harness

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

No tool records are attached.

Evaluation contract

  • Output is a valid JSON object matching the schema.
  • Agent successfully called at least 2 external tools (e.g., news API, stock API, prediction market).
  • Evidence of a multi-step LangGraph execution (e.g., trace logs showing multiple nodes/agents activated).

Metrics

  • PredictionAccuracy · Target 0.75
  • RationaleQuality · Target 4
  • ConfidenceScoreAccuracy · Target 0.7

Workflow tasks

  1. Task 1

    Input: { "event_description": "<news headline or brief summary>", "company_ticker": "<STOCK_TICKER>", "event_date": "YYYY-MM-DD" }

    Output: { "predicted_movement": "<UP|DOWN|FLAT>", "magnitude_percent": "<float>", "confidence_score": "<0-1.0>", "rationale": "<detailed justification>", "data_sources_used": [{"name": "<source_name>", "url": "<url>"}] }

    Execution harness: Python

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Agent for Autonomous Market Intelligence