Consensus Capital bridges traditional financial analysis with real-time prediction market probabilities to reveal hidden macro risks and non-obvious correlations in investment portfolios.

By combining Polymarket prediction data, semantic vector search, and AI-driven workflows, Consensus Capital gives traders a probabilistic edge that charts and earnings reports alone can’t provide.

• 🥇 McHacks 13 — Auth0 Prize Winner
• 🥇 McHacks 13 — Best Project Using Gumloop

During our brainstorming sessions, we kept coming back to one phenomenon: Polymarket.

In recent years, prediction markets have consistently forecasted major global events with higher accuracy than traditional news sources. Yet, despite this predictive power, stock traders largely ignore these signals.

We realized that traders are operating with incomplete data. While markets obsess over technical indicators and earnings, they miss the probabilistic expectations that drive sentiment and volatility.

In an era of peak uncertainty, the real edge isn’t just getting the news first — it’s connecting macro probabilities directly to your portfolio. Consensus Capital was built to unlock that missing dimension.

Consensus Capital connects real-time prediction markets with traditional asset data to stress-test portfolios against macro-level risks.

• Ingests the top 50 most active markets from Polymarket
• Runs a multi-stage AI workflow using Gumloop
• Analyzes portfolios against geopolitical, regulatory, and economic events
• Uses cosine similarity to uncover non-obvious correlations

Instead of reacting to headlines, users gain a 360-degree view of portfolio exposure that traditional tools miss.

• Modular, responsive dashboard
• Strong type safety for AI-generated nested JSON
• Clean visualizations for similarity scores and risk alerts

• Scalable REST API
• Enterprise-grade session management
• Auth0 integration with:
  • Passwordless login
  • MFA
  • Biometric Passkeys (FaceID / TouchID)
• MongoDB for flexible portfolio storage

• Dedicated microservice for:
  • Polymarket & Yahoo Finance ingestion
  • Embedding generation
  • Cosine similarity computation

• Automated workflows acting as the system’s brain
• Stock classification
• Noise filtering (sports betting, irrelevant markets)
• Natural-language insight generation

• Bridging deterministic prices with probabilistic events
• Keyword matching failed to capture real relevance
• Pivoted to semantic analysis using vector embeddings
• Required careful API design between Java and Python services

• Enterprise Auth in 24 Hours
  Full Auth0 integration with MFA and biometrics without hurting UX.

• Smart Caching Layer
  Prevents API rate limits and delivers instant insights.

• Semantic Relevance Engine
  Understands that a Taiwan blockade is relevant to Nvidia — even without keyword overlap.

• Seamless Multi-Stack Integration
  React, Spring Boot, Python, and Gumloop working together cleanly.

• LLMs Need Guardrails
  Clean, parsable JSON requires strict prompting and validation.

• Meaning > Text
  Semantic similarity dramatically outperforms keyword matching.

• Hybrid Stacks Are Hard (But Worth It)
  Java + Python forced disciplined API design and data contracts.

• Rapidly Learning New AI Platforms
  Gumloop’s learning curve paid off with powerful orchestration.

• Deeper, more actionable insights
• Expanded what-if scenario modeling
• Custom Auth0 refresh tokens
• Transparent explanations behind every correlation

MIT