Establish generalization to broader forecasting and decision-making settings

Establish whether the results obtained for structured horse-racing forecasting generalize to other sports, unstructured information environments, and broader decision-making tasks.

Background

The evaluation focuses on structured forecasting tasks using time-gated historical horse-racing records, priced data tools, persistent programs, and fixed retrospective decision rules. This design does not test settings involving other sports, less structured evidence, or more general forms of decision-making.

The limitations section explicitly identifies transfer to these broader domains as unresolved rather than demonstrated by the benchmark.

References

Generalization to other sports, unstructured information environments, and broader decision-making tasks remains to be established (Srinivasan, 2026).

— JockeyBench: Evaluating LLM Agents on Horse-Racing Decision-Making  (2610.19589 - Xie et al., 14 Oct 2026) in Limitations