What the measured evidence supports
Of 500 resolved Polymarket UMA disputes over ~313 days, only 13.8% reversed the original answer — 86.2% just confirmed it.
The figures describe the measured sample and the coverage stated by the methodology, not the whole population. Where coverage is partial, the gap is declared rather than estimated.
Readers and agents can inspect the companion dataset, follow the public methodology, and query the API from the links below.
This answer applies only to the observed sample, measurement window, and coverage stated by the linked methodology.
Interpretation should preserve the caveats attached to every used claim rather than generalize beyond available coverage.