Will eventually resolve to the party which won the popular vote for the race, as determined by the local, state, and ultimately federal governments.
In the case the race is called by the major news organizations, the market may optionally resolve early.
This market covers events up to and including Election Day, November 5: if Party A's candidate X wins the popular vote that day, but intervening events cause the next inaugurated Senator to be from Party B, the market will resolve to Party A.
If candidate X switches party affiliations, the switch will not affect the outcome if it is after the beginning of Election Day, local time.
Analysis from the Manifold newsletter:
New Mexico: Sen. Martin Heinrich (D) vs. Nella Domenici (R)
Democrat win chances (Manifold, 16 traders): 95% (I bet up from 90%)
Democrat win chances per models: 95% (The Hill), 98% (538), 95% (Split Ticket)
Expert forecasts: Solid D (Cook), Safe D (Sabato), Solid D (Inside Elections)
Polling average (538): Heinrich+10
Election result in 2018, a Democratic wave year: D+24 (Heinrich vs. Rich)
Election result in 2012, amid Obama's reelection: D+6 (Heinrich vs. Wilson)
Recent New Mexico presidential election results: Biden+11 (2020), Clinton+8 (2016), Obama+10 (2012)
Commentary:
At 90%, I felt that this Manifold probability (in a state which went red in 2004 but has since trended blue) was just low due to laziness; a lot of these are, frankly. So I bet it up to match the models! I did leave some free alpha on the table in these other markets, though, as a present for reading this far.