Odisha

Every seat above is assigned by a deterministic vote-share rule, never a probability model — this is a sandbox, not a forecast. A projection starts from each seat's own recorded baseline vote shares, then applies whatever transfers, new parties, turnout changes, pre-poll pacts and national/state swing you have set, in that order, and finally reads off the plurality winner (NOTA is never counted as a winner).

The national swing slider debits points from the trailing bloc's member parties pro-rata to their own share in each seat, and credits the leading bloc's strongest local member there — so a seat where a bloc member polled zero loses nothing from that member, and a swing never moves more points than a bloc actually holds. A pre-poll pact's "chemistry" setting controls how much of a withdrawing partner's vote actually transfers to the pact's nominee; a post-poll coalition is pure arithmetic over the resulting seat tally.

No confidence intervals, no Monte Carlo simulation, and no polling input of any kind — there is nothing here to calibrate such a distribution against, so none is invented. Read every number as "what would it take", not a prediction.