Team predictions, not blind automation.
FEA Team Predictions are editorial/team race ratings informed by public ratings, polling, candidate quality, fundraising, state and district partisanship, national environment, election history, and internal model diagnostics.
How this works.
The automated model is maintained as an internal diagnostic tool in Model Lab. It can surface polling gaps, baseline disagreements, release-gate problems, and margin/probability signals, but it does not automatically decide the public prediction.
Public predictions are manually editable. The team can accept, adjust, or ignore model suggestions after reviewing the broader political context.
County-level predictions, when present, are also manual team predictions. Missing county predictions render as unrated rather than inheriting model output.