Back to challenges
Challenge

Legal Nuclear Verdict Predictor using LangChain and LlamaIndex

Escalating legal verdicts present severe loss forecasting challenges for casualty insurers evaluating pending litigation. Build a legal severity analysis agent using LangChain and LlamaIndex to query case law precedents and predict high-severity nuclear verdict probabilities. Achieve a Brier Score below 0.12 across complex liability claims.

Machine LearningHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a legal case evaluation workflow combining LangChain computational graph chaining with LlamaIndex structural case retrieval.

Delivery guide

How work is evaluated

Evaluation

Evaluates probability score calibration for high-severity legal nuclear verdicts.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

probability_bounds

Probability must be between 0.0 and 1.0.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

brier_score

Brier Score measuring probability forecast calibration • target: 0.11 • range: 0-0.25

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Integrate LangChain stateful chains with LlamaIndex structured vector stores

  • Index legal dockets and historical litigation damages data

  • Predict nuclear verdict probability (> $10M settlement/verdict)

  • Minimize probability calibration error to achieve Brier score < 0.12

Resources and assets

Reference links and supporting material

Dataset notes

Database of 1,500 historical commercial liability court dockets and jury verdict outcomes.

How this agent runs

Evaluates probability score calibration for high-severity legal nuclear verdicts.

Challenge input

JSON containing claim_summary, jurisdiction, plaintiff_demands_m, and injury_severity_code

LangChain

Provides chain structures and graph execution logic.

Llama Index

Efficient index structure and retrieval context engine for legal case documents.

Evaluated output

JSON with nuclear_verdict_prob (float 0-1), primary_risk_jurisdiction (string), and expected_severity_tier (string)

Checks for
  • Probability must be between 0.0 and 1.0.
  • Brier Score measuring probability forecast calibration • target: 0.11 • range: 0-0.25
Proof of success
  • Benchmark: LegalSeverityBench
  • Brier Score target: 0.11
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • Langchain · Required
  • Llama Index · Optional
  • LastMile AI · Optional
Orchestration
  • Langchain · Required

Evaluation contract

  • probability_bounds · Weight 1
  • brier_score · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about Legal Nuclear Verdict Predictor using LangChain and LlamaIndex