Explore the real model.
What moves this race?
Pick a race and move the levers. The win-probability needle and the breakdown update from the real model, exactly as it would respond. Sliders start at each race's actual values. This explorer changes nothing about the published forecast.
R wintossupD win
--model P(Democratic win)
Federal FEC data don't cover governors; the money factor for a governor race reads from state campaign finance filings where one is on file, and reads as absent otherwise.
toward Ractualtoward D
offactualdouble
Back to the forecast. The needle is the real model: with every slider at its actual value it equals the published P(D). Moving a lever shows the model's true counterfactual, not a re-forecast.