The work

Each task hands you an evidence cutoff — a date — and asks what a strong analyst would have concluded at that moment. You write the report: where the race stands, what the polling supports and what it does not, a directional judgment on win probability or market-implied odds, and the causal mechanism connecting an event to its expected political or market impact. Some tasks flip the direction: you read a model's analysis of a race you know intimately and grade it, flagging where it leaned on hindsight, misread a poll's likely-voter screen, or asserted a mechanism that does not operate in that state.

Hindsight is the central discipline. If the cutoff is three weeks before a runoff, you cannot cite the runoff result, a later poll, or anything you learned afterward — and the graders check. The same rule bars confidential campaign data, client work, and material non-public information. What makes the data valuable is that your reasoning is reconstructible from public evidence available at the cutoff.

What the screen looks for

  • State specificity. Which races, which cycles, which counties or media markets. Generic national forecasting experience does not clear this on its own; it is welcome when paired with deep knowledge of one state.
  • Live judgment, not retrospective commentary. Evidence you made a call before the result was known — published ratings, internal forecasts, polling memos, odds views — and how it turned out.
  • Polling literacy under follow-up. Weighting, likely-voter models, mode effects, house effects, differential nonresponse, and how you handle a sparse or contradictory poll set in a low-information statewide race.
  • Causal reasoning. Whether you can name a transmission channel from event to outcome rather than asserting correlation.

Expect roughly 8+ years of direct experience, or unusually strong evidence of equivalent performance. Brand-name affiliation helps on first pass but the work sample decides.

Logistics

Remote and asynchronous, contractor engagement, work claimed from a task queue rather than scheduled shifts. Most contributors treat it as part-time alongside existing roles; throughput matters more than specific hours. Pay in the $150–250/hr band reflects rates observed on this platform for state-expert political work and is not a guarantee — it varies with task type, review outcomes, and how narrow your state coverage is.