SCORING.md

📊 TREMOR Intensity Scoring

How we measure market earthquakes

The Concept

Just like seismologists measure earthquakes on the Richter scale, TREMOR measures prediction market movements on an intensity scale. When probabilities shift dramatically, we detect and quantify the "tremor" in the market.

🎯 Scoring Scale (0-10)

Our intensity scale maps market movements to easily understood severity levels:

┌──────────────────────────────────────────────────────┐
│  0    1    2.5      5.0        7.5           10      │
│  │────│─────│────────│──────────│─────────────│      │
│  ⚪ LOW   🟡 MODERATE  🟠 HIGH    🔴 EXTREME          │
└──────────────────────────────────────────────────────┘

Real-World Examples

Score	Intensity	Market Example	Earthquake Equivalent
0.5	⚪ Low	0.5% price move, $500 volume	Barely perceptible
2.0	⚪ Low	2% price move, $5k volume	Minor tremor
4.0	🟡 Moderate	4% price move, $20k volume	Noticeable shake
6.0	🟠 High	8% price move, $50k volume	Strong movement
8.0	🔴 Extreme	15% price move, $100k volume	Major earthquake
10.0	🔴 Extreme	20%+ price move, $200k+ volume	Catastrophic event

🧮 The Algorithm (v2)

Core Formula

intensity = priceChangeScore × volumeMultiplier

Price Change Mapping

The relationship between price change and base score:

Price Change (%)    Base Score    Visual
─────────────────────────────────────────
0-1%               0-1          ▁▁
1-5%               1-5          ▁▁▃▃▅
5-10%              5-7.5        ▅▅▆▇
10-20%             7.5-10       ▇▇█
20%+               10           █████

Volume Multiplier

Volume adds credibility to price movements:

Volume ($)         Multiplier    Effect
─────────────────────────────────────────
< $1,000          0.0          No score
$1,000-$5,000     0.1-0.5      Reduced
$5,000-$10,000    0.5-1.0      Gradual
> $10,000         1.0          Full score

📈 Visual Examples

Example 1: Moderate Tremor (Score: 4.2)

Event: "Will Bitcoin reach $100k by March?"
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Price Movement:
Before: 35% → After: 39% (4% change)
        ████████░░░░░░░░░░

Volume: $25,000
        ████████████░░░░░░

Intensity: 4.2 🟡 MODERATE
        ████████▌░░░░░░░░░

Example 2: Extreme Tremor (Score: 8.5)

Event: "Will the Fed cut rates this month?"
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Price Movement:
Before: 20% → After: 35% (15% change)
        ████████████████░░

Volume: $150,000
        ██████████████████

Intensity: 8.5 🔴 EXTREME
        █████████████████░

Example 3: Low Activity (Score: 0.0)

Event: "Will it rain in Seattle tomorrow?"
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Price Movement:
Before: 70% → After: 72% (2% change)
        ██░░░░░░░░░░░░░░░░

Volume: $750
        █░░░░░░░░░░░░░░░░░

Intensity: 0.0 (Below threshold)
        ░░░░░░░░░░░░░░░░░░

⏱️ Time Windows

TREMOR analyzes movements across multiple timeframes:

5-Minute Window

• Purpose: Catch breaking news
• Use Case: "BREAKING: Major announcement"
• Sensitivity: Highest
• Example: Election results coming in

1-Hour Window

• Purpose: Identify trending movements
• Use Case: Sustained market sentiment shift
• Sensitivity: Moderate
• Example: Policy announcement digestion

24-Hour Window

• Purpose: Major narrative changes
• Use Case: Complete probability reassessment
• Sensitivity: Lower (needs bigger moves)
• Example: Black swan events

📊 Multi-Market Aggregation

For events with multiple markets, we aggregate intelligently:

def aggregate_event_score(markets):
    # Take the maximum individual market score
    max_score = max(market.score for market in markets)

    # Boost slightly for correlated movements
    if multiple_markets_moving():
        max_score *= 1.1  # 10% boost

    return min(max_score, 10)  # Cap at 10

Example: Presidential Election

Event: "2024 Presidential Election"
├── Market A: "Will Trump win?"
│   └── Score: 6.5 🟠
├── Market B: "Will Biden win?"
│   └── Score: 6.3 🟠
└── Market C: "Will there be a third party winner?"
    └── Score: 2.1 ⚪

Event Score: 6.5 🟠 (Maximum of all markets)

🔄 Evolution History

v1 (Original)

• Simple linear scaling
• Single time window
• Basic volume threshold

v2 (Current)

• Non-linear scaling for realistic intensity
• Multi-window analysis
• Graduated volume multiplier
• Better handling of edge cases

v3 (Future)

• Refined thresholds based on historical data
• Market-specific adjustments
• Improved baseline calculations

🎯 Calibration Examples

To ensure our scoring is meaningful, here's how real events would score:

Real Event	Market Move	TREMOR Score	Accuracy
Trump indictment news	15% in 1hr	8.2 🔴	✅ Major event
Fed rate decision	8% in 5min	6.5 🟠	✅ Significant
Celebrity endorsement	3% in 24hr	2.8 🟡	✅ Minor news
Random fluctuation	1% in 1hr	0.8 ⚪	✅ Filtered out

🔧 Technical Implementation

Database Query

-- Get highest scoring events in last hour
SELECT
  e.title,
  s.maxScore,
  s.topMarket,
  s.metadata
FROM scores s
JOIN events e ON s.eventId = e._id
WHERE s.timestamp > NOW() - INTERVAL '1 hour'
ORDER BY s.maxScore DESC
LIMIT 20

Score Computation

// Runs every minute via cron
async function computeScores() {
  const events = await getActiveEvents();

  for (const event of events) {
    const markets = await getEventMarkets(event.id);
    const snapshots = await getRecentSnapshots(markets);

    const scores = {
      score5m: calculateScore(snapshots, '5m'),
      score1h: calculateScore(snapshots, '1h'),
      score24h: calculateScore(snapshots, '24h'),
    };

    await saveScore(event.id, scores);
  }
}

📈 Future Improvements

• More accurate volume weighting
• Better handling of market-specific patterns
• Improved normalization for different market types
• Refined time window calculations

🎓 Understanding the Numbers

For Traders

• 2.5+: Worth investigating
• 5.0+: Significant opportunity/risk
• 7.5+: Major market event

For Observers

• Low (0-2.5): Normal market noise
• Moderate (2.5-5): Something interesting
• High (5-7.5): Important development
• Extreme (7.5+): Breaking news territory

---

The scoring algorithm is continuously refined based on real-world data and user feedback.