1992-2012
Deep red -- Democrats win every Presidential race.
Medium red -- Democrats win all but one Presidential race.
White -- always went with the winner
Pale blue -- went for the winner in all election, but in that exception went for the Republican
Medium blue -- Republicans win all but one Presidential race.
Yellow -- twice Democratic, but seeming to now drift Democratic
Green -- twice Democratic but seeming to drift Republican (Missouri in a light shade because Obama was close in 2008, others deep green)
Deep blue --Republicans win every Presidential race.
... This is where we reasonably start. Any information from before 1992 is irrelevant to current reality. Anyone who predicts that a state in deep (or even medium) red or deep (or even medium) blue is going to vote differently from how it has voted in the last six elections has some explaining to do. How California or Texas voted in 1976 or earlier no longer matters. States in pale shades or white can be understood as swing states in anything near a 50-50 election.
No state is in pink, so 'reasonable swing states' according to state voting patterns of the last twenty years suggest that
CO FL MO NV OH VA
are the real swing states.
If you see something out of recent norms happening in Arizona or Iowa, then you can add those. I would be tempted to replace MO with NC based on 2012. You might make an argument that some Democrat is a better match for states in green or that Virginia has gone Democratic twice only because of Obama and will not do so again. You can argue such and still be wrong, but you might have a reason.
But if you say something like "Kansas must be getting sick of Republican pols", "the fast-growing Hispanic population dooms the Republicans in Arizona", "economic distress in the Rust Belt will cause blue-collar workers to abandon the Union Bosses for free-market solutions", or "Scott Walker is sure to win Wisconsin as a favorite son" you have some explaining to do and a need for evidence to support your position. Polls will be adequate.