Skip to content

Complementary option-chain eval (ground truth from market) #450

Description

@connerlambden

Helium Market Resolution tests whether LLMs can read live option chains (IV, delta, put-call parity) with ground truth from the chain itself—not MCQ finance trivia.

300 frozen prompts, six tickers, twelve frontier models scored. No model above 50%; grok-4.20-reasoning leads at ~48%.

HF: https://huggingface.co/datasets/HeliumTrades/helium-market-resolution-benchmark
Overview: https://heliumtrades.com/benchmarks/

We use a private holdout for anti-contamination. Open to contributing a LiveBench-compatible subset if there's interest.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions