[Phase 1] Bookmaker Disagreement Detection

[WFord26]

---

name: "Phase 1: Bookmaker Disagreement Detection"
about: Find value by identifying games where bookmakers strongly disagree
title: "[Phase 1] Bookmaker Disagreement Detection"
labels: enhancement, phase-1, analytics
assignees: ''

Overview

Detect and highlight games where bookmakers have significantly different odds, indicating market uncertainty and potential value opportunities.

Business Value

• Value Detection: High disagreement = uncertainty = potential value
• Arbitrage Opportunities: Extreme disagreement can create arbs
• Market Inefficiency: Shows where the market hasn't reached consensus
• User Education: Teaches users to spot value

Technical Requirements

Database Changes

model MarketConsensus {
  id                  String   @id @default(dbgenerated("gen_random_uuid()")) @db.Uuid
  gameId              String   @map("game_id") @db.Uuid
  marketType          String   @map("market_type") @db.VarChar(20)
  calculatedAt        DateTime @default(now()) @map("calculated_at") @db.Timestamptz(6)
  
  // Consensus metrics
  consensusLine       Decimal  @map("consensus_line") @db.Decimal(5,1)
  standardDeviation   Decimal  @map("standard_deviation") @db.Decimal(5,2)
  outlierBookmakers   Json     @map("outlier_bookmakers")
  
  // Value indicators
  disagreementScore   Int      @map("disagreement_score")  // 1-100
  bestValue           Json     @map("best_value")  // Which side has best odds
  
  game Game @relation(fields: [gameId], references: [id])
  
  @@index([gameId])
  @@index([disagreementScore])
  @@index([calculatedAt])
  @@map("market_consensus")
}

Backend Services

File: src/services/market-consensus.service.ts

class MarketConsensusService {
  async calculateConsensus(gameId: string, marketType: string): Promise<Consensus>
  async findHighDisagreement(threshold: number): Promise<Game[]>
  async identifyOutliers(gameId: string): Promise<Bookmaker[]>
  async calculateDisagreementScore(gameId: string): Promise<number>
}

Scheduled Job: Calculate consensus every 15 minutes for upcoming games

• Cron: */15 * * * *
• Query games in next 48 hours with odds
• Calculate consensus for each market type
• Store results and disagreement scores

Algorithm

Consensus Line Calculation:

// For spreads/totals
consensus = median(allBookmakerLines)

// For moneylines
impliedProbs = lines.map(line => toImpliedProbability(line))
consensusProb = median(impliedProbs)
consensusLine = toAmericanOdds(consensusProb)

// Standard Deviation
stdDev = calculateStdDev(allLines)

// Disagreement Score (1-100)
score = min(100, (stdDev / consensus) * 100 * 10)

Outlier Detection:

• Lines > 2 standard deviations from consensus
• Store bookmaker name and how far off

API Endpoints

• GET /api/analytics/disagreement/live - Current high disagreement games
• GET /api/analytics/disagreement/game/:gameId - Disagreement for specific game
• GET /api/analytics/disagreement/trends - Historical disagreement patterns
• GET /api/analytics/disagreement/bookmaker/:bookmaker - How often a bookmaker is an outlier

Frontend Components

Dashboard Widget: HighDisagreementGames.tsx

• List of top 5 games with highest disagreement
• Show disagreement score (1-100)
• Click to see full consensus breakdown
• Real-time updates every 60 seconds

Detailed Modal: DisagreementBreakdown.tsx

• All bookmaker lines for the game
• Visual representation of spread/distribution
• Highlight outliers in red
• Show which side has best value
• Historical disagreement for this matchup

Filters Page: ValueOpportunities.tsx

• Filter by disagreement score
• Filter by sport
• Filter by time until game starts
• Sort by score, time, sport

Calculations & Thresholds

Disagreement Categories:

• Low (0-30): Normal market efficiency
• Medium (31-60): Some uncertainty
• High (61-80): Significant disagreement
• Extreme (81-100): Major market inefficiency

Alert Thresholds:

• Score > 70: Push notification to users
• Score > 85: Email alert for major opportunities

Acceptance Criteria

• Database migration completed
• Consensus calculation service implemented
• Scheduled job running reliably (15-min intervals)
• Dashboard widget showing high disagreement games
• Detailed breakdown modal with visual distribution
• API endpoints documented
• Unit tests for consensus calculations
• Historical data showing disagreement accuracy

Dependencies

• Odds sync service must have multiple bookmakers
• Minimum 5 bookmakers per game for reliable consensus

Estimated Effort

• Backend: 4 days
• Frontend: 3 days
• Testing & Tuning: 2 days
• Total: 9 days

Success Metrics

• Detect 10+ high disagreement games per day
• Alert notifications sent within 1 minute of detection
• Users click through on 20%+ of alerts
• 5+ bookmakers per consensus calculation (average)

Future Enhancements

• Machine learning to predict which side wins when disagreement is high
• Historical win rates by disagreement level
• Bookmaker accuracy tracking (who's right when they're outliers)