Finsider Labs

[PROPOSED EVALUATIONS]

Earnings quality evaluation protocol

Adjustment evidence, arithmetic, period consistency, and treatment of unresolved candidates.

Proposed evaluation only. No public dataset, completed benchmark run, ranked models, or measured results are available for these tracks.

Proposed controls and evaluation

Use synthetic examples with known arithmetic and practitioner-reviewed cases with an explicit review rubric. Test signs, period boundaries, double counting, and exclusion of pending items. Evaluate arithmetic agreement separately from reviewer disagreement about economic treatment. Do not score every judgment difference as a software error. No benchmark or performance uplift is reported.

Before reporting results

Publish the task definition, data permissions, source coverage, reference-label process, metrics, denominators, and limitations. Separate calculation correctness from professional judgment and disclose unresolved disagreement.