PredictCollusion

The Problem

Procurement professionals struggle to identify collusion risks in upcoming tenders due to fragmented historical data and subtle behavioral patterns. Existing approaches rely on manual spreadsheet analysis that is time-consuming, inconsistent, and often misses emerging collusion signals until after contract award.

The Solution

This tool provides instant collusion risk scoring by analyzing bidder behavior patterns, market structure, and historical anomalies. It combines multiple detection heuristics into a single deterministic calculation that procurement teams can use during tender preparation to adjust evaluation criteria and monitoring intensity.

Input Parameters

Functional Requirements (Structured)

FR-001: Market Dominance Score

• Inputs: market_concentration, bidder_count
• Output: dominance_score (number, %)
• Constraint: 0-100%, higher indicates concentrated market
• Formula hint: dominance = f(market_concentration, 1/bidder_count) weighted combination

FR-002: Bid Pattern Anomaly Score

• Inputs: bid_spread_previous, identical_bid_items
• Output: pattern_score (number, %)
• Constraint: 0-100%, higher indicates suspicious bid patterns
• Formula hint: pattern = w1(small_bid_spread_penalty) + w2(identical_pricing_penalty)

FR-003: Behavioral Collusion Index

• Inputs: winner_rotation, geographic_clustering
• Output: behavioral_score (number, %)
• Constraint: 0-100%, higher indicates collusive behavior patterns
• Formula hint: behavioral = f(1/winner_rotation, geographic_clustering) with threshold effects

FR-004: Composite Collusion Risk

• Inputs: dominance_score, pattern_score, behavioral_score
• Output: collusion_risk (number, %)
• Constraint: 0-100%, weighted composite
• Formula hint: risk = 0.4*dominance + 0.4*pattern + 0.2*behavioral

FR-005: Risk Classification

• Inputs: collusion_risk
• Output: risk_level (text, category)
• Constraint: Must be one of: Low, Moderate, High, Critical
• Formula hint: classification based on thresholds: <30% Low, 30-60% Moderate, 60-85% High, >85% Critical

Calculation Dependencies

dominance_score <- (market_concentration, bidder_count)
pattern_score <- (bid_spread_previous, identical_bid_items)
behavioral_score <- (winner_rotation, geographic_clustering)
collusion_risk <- (dominance_score, pattern_score, behavioral_score)
risk_level <- (collusion_risk)

Output Results

Validation Test Cases

Test Case 1: Competitive Market

Inputs:
  bidder_count: 8
  market_concentration: 1200
  bid_spread_previous: 22
  winner_rotation: 4
  geographic_clustering: 25
  identical_bid_items: 2

Expected Outputs:
  dominance_score: 18.2
  pattern_score: 15.8
  behavioral_score: 24.5
  collusion_risk: 18.9
  risk_level: Low

Test Case 2: High Risk Scenario

Inputs:
  bidder_count: 3
  market_concentration: 4800
  bid_spread_previous: 3.5
  winner_rotation: 2
  geographic_clustering: 85
  identical_bid_items: 28

Expected Outputs:
  dominance_score: 82.7
  pattern_score: 76.4
  behavioral_score: 88.3
  collusion_risk: 80.8
  risk_level: High

Test Case 3: Borderline Case

Inputs:
  bidder_count: 5
  market_concentration: 2800
  bid_spread_previous: 8
  winner_rotation: 3
  geographic_clustering: 60
  identical_bid_items: 12

Expected Outputs:
  dominance_score: 54.3
  pattern_score: 48.2
  behavioral_score: 52.7
  collusion_risk: 51.9
  risk_level: Moderate

Domain Expertise

• HHI above 2500 indicates highly concentrated markets with elevated collusion risk
• Bid spreads below 5% in competitive markets often signal price signaling
• Winner rotation below 2 in 5 tenders suggests bid suppression or market allocation
• Geographic clustering above 70% significantly increases local cartel formation probability
• Identical pricing on >15% of line items strongly suggests communication between bidders
• Construction, transportation, and waste management sectors show highest historical collusion rates

Who Uses This Tool

Procurement officers, contract managers, and anti-corruption analysts in public and private sector organizations. These professionals need rapid, consistent risk assessments during tender preparation and evaluation phases, replacing subjective judgments with data-driven insights.