Skip to main content
Loading navigation
Back to challenges

LlamaIndex Debt-for-Equity Underwriting Engine with DeepSeek R1

Build a complex CRE debt restructuring analyzer using LlamaIndex and DeepSeek R1 to evaluate debt-for-equity swap terms against property cash flows. The solution extracts financial covenants and generates risk-adjusted returns with zero parsing errors.

Business OperationsHosted by Vera
Sign in free to start

What you are building

Combines LlamaIndex's advanced RAG capabilities with DeepSeek R1 reasoning models to process distressed commercial property financial models and swap structures.

How work is evaluated

Evaluation

Evaluates financial formula accuracy and document clause extraction accuracy.

Datasets

Public datasets, plus the private datasets you have access to.

How submissions are scored

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

DSCR Math Verification

Verifies calculated revised DSCR matches expected algebraic debt reduction

Binary check

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

Dimension 2

IRR Percentage Error

Absolute error on projected equity IRR calculation • target: 0.1 • range: 0-2

Binary check

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

Learning goals

  • Construct LlamaIndex index structures optimized for complex financial PDF filings

  • Leverage DeepSeek R1 reasoning traces for debt service coverage ratio (DSCR) extraction

  • Model debt-for-equity swap valuations and post-restructuring cash flows

  • Generate verifiable equity dilution and recovery yield projections

Resources

Dataset notes

25 corporate real estate debt restructuring prospectus filings and associated debt service schedules.

Challenge setup

Evaluates financial formula accuracy and document clause extraction accuracy.

Challenge input

JSON containing loan_amount, current_dscr, proposed_equity_conversion_pct, and net_operating_income

LlamaIndex

Handles indexing and retrieval of unstructured legal/financial filings.

DeepSeek R1

Executes step-by-step financial formula reasoning.

Evaluated output

JSON containing revised_dscr, lender_equity_share_pct, projected_irr, and risk_category

Evaluation checks
  • Verifies calculated revised DSCR matches expected algebraic debt reduction
  • Absolute error on projected equity IRR calculation • target: 0.1 • range: 0-2
Configured evaluation evidence
  • Benchmark: FinRAG-ValuationBench
  • IRR Percentage Error target: 0.1
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Llama Index · Required
  • Exa · Optional
Policy Serving
  • DeepSeek R1 · Optional

Evaluation contract

  • DSCR Math Verification · Weight 1
  • IRR Percentage Error · Weight 1

Metrics

  • IRR Percentage Error · Target 0.1

Benchmarks

  • FinRAG-ValuationBench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing loan_amount, current_dscr, proposed_equity_conversion_pct, and net_operating_income

    Output: JSON containing revised_dscr, lender_equity_share_pct, projected_irr, and risk_category

    Execution harness: Python

Evaluation tools

  • LlamaIndex · Data RAG Framework
  • DeepSeek R1 · Reasoning Model

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about LlamaIndex Debt-for-Equity Underwriting Engine with DeepSeek R1