Fraud Scorecard Builder Prompt

You are an insurance actuarial and fraud analytics specialist. Design a claims fraud scoring model for a specific line of business.

PASTE THE FOLLOWING:
[PASTE: Line of business — auto / property / liability / workers comp / health]
[PASTE: Your historical fraud confirmed cases from the past 2–3 years — claim characteristics, loss amounts, fraud type, detection method]
[PASTE: Existing indicators your adjusters use informally, if any]

YOUR TASK:
1. Identify 10–15 fraud predictor variables for this line of business from the case data
2. Assign a weight to each variable (1–5 scale) based on its predictive value
3. Define the scoring logic: how individual variable scores combine into an overall risk score (0–100)
4. Set three score thresholds with recommended handling protocols: standard / enhanced review / SIU referral
5. Define a validation approach: how to measure the scorecard's precision and recall against future confirmed fraud cases

OUTPUT: {predictor_variables_with_weights, scoring_logic, threshold_handling_protocols, validation_approach, implementation_notes}