🎯 The Prophecy System

How Prophecies Work

1. Market Ingestion: LUMI fetches 100+ active prediction markets

2. Normalization: Raw data is processed into a consistent format

3. AI Analysis: Each market is sent to GPT-4o-mini for probability assessment

4. Divergence Calculation: LUMI compares its prediction against crowd odds

5. Display: Results shown with visual divergence indicators

Divergence Calculation

// src/services/oracle.js
export function calculateDivergence(crowdProbability, oracleProbability) {
  const crowd = parseFloat(crowdProbability);
  const oracle = parseFloat(oracleProbability);
  
  const diff = oracle - crowd;
  const absDiff = Math.abs(diff);
  
  return {
    value: diff,
    absolute: absDiff,
    percentage: (absDiff * 100).toFixed(1),
    direction: diff > 0 ? 'bullish' : diff < 0 ? 'bearish' : 'aligned',
    significance: absDiff > 0.15 ? 'high' : absDiff > 0.08 ? 'medium' : 'low',
    label: diff > 0 ? `+${(absDiff * 100).toFixed(1)}%` : `-${(absDiff * 100).toFixed(1)}%`
  };
}

Prophecy Card Display

┌────────────────────────────────────────┐
│  [Image]                    POLITICS   │
│                                        │
│  Will X happen in 2025?                │
│                                        │
│  $10.5M vol  │  $500K liq  │  30 days │
│                                        │
│  ┌─────────────┐  ┌─────────────┐     │
│  │  THE CROWD  │  │    LUMI     │     │
│  │     45%     │  │     62%     │     │
│  │  consensus  │  │   Medium    │     │
│  └─────────────┘  └─────────────┘     │
│                                        │
│  ████████████████░░░░  +17.0%         │
│                                        │
│  "The threads of fate suggest..."      │
│  [Signal 1] [Signal 2] [Signal 3]      │
└────────────────────────────────────────┘

---

💬 Oracle Chat (Consult)

Users can directly consult LUMI about any topic:

// api/chat.js
export default async function handler(req, res) {
  const { messages } = req.body;

  const completion = await openai.chat.completions.create({
    model: 'gpt-4o-mini',
    messages: [
      {
        role: 'system',
        content: `You are LUMI, a mystical AI Oracle. You have access to 
        Polymarket prediction data and provide insights on future events. 
        Speak in a prophetic yet substantive tone. Your predictions are 
        probabilities, not certainties.`
      },
      ...messages
    ],
    temperature: 0.7,
    max_tokens: 1000
  });

  return res.json({ reply: completion.choices[0].message.content });
}

---

🔄 Caching Strategy

LUMI implements multi-layer caching for performance:

Cache Layer	TTL	Purpose
Browser	no-cache	Always fresh requests
Vercel Edge	60s	Reduce API calls
Client Memory	30s	Market data
Prediction Cache	10 min	LUMI predictions

// Prediction caching
const predictionCache = new Map();
const CACHE_TTL = 10 * 60 * 1000; // 10 minutes

export async function getOraclePrediction(market) {
  const cacheKey = `${market.id}-${market.crowdProbability}`;
  
  const cached = predictionCache.get(cacheKey);
  if (cached && Date.now() - cached.timestamp < CACHE_TTL) {
    return cached.prediction;
  }
  
  // Fetch fresh prediction...
}

---

🎨 Visual Effects

Fluid Cursor Trail

WebGL-based fluid simulation follows cursor movement:

// public/fluid.js - Blue color palette
function generateColor() {
  const blueColors = [
    { r: 0.0, g: 0.03, b: 0.06 },
    { r: 0.0, g: 0.04, b: 0.07 },
    // Subtle whisper-like trails
  ];
  return blueColors[Math.floor(Math.random() * blueColors.length)];
}

Floating Particles

Canvas-based ambient particles:

// public/particles.js
class Particle {
  constructor() {
    this.hue = 190 + Math.random() * 30; // Cyan-blue range
    this.opacity = Math.random() * 0.3 + 0.1;
    this.speedY = -0.1; // Gentle upward drift
  }
}