Rarely will you hear a poll called “good news” for a candidate when it has him down eight in his home state, but this really is the most encouraging news Rubio’s had in ages. And the best news anti-Trumpers have had since, well, Saturday afternoon.

Rubio 48, Trump 23 in the early vote? Gadzooks.

Trump has support from 38% of likely primary voters compared to 30% who back Marco Rubio. Ted Cruz earns 17% support and John Kasich has 10%. Rubio actually leads Trump by 48% to 23% among the nearly 1-in-5 voters who have already cast their ballots in this “early vote” state. Trump has a 42% to 26% lead among those who have yet to vote.

Rubio leads Trump 41% to 30% in the southern part of the state. Trump has a significant 44% to 22% advantage in Florida’s central region. The race is closer in the northern tier of the state, with Trump at 36% and Rubio at 32%. Nearly one-fifth of the likely electorate is non-white, mostly Cuban or other Hispanic. This group appears to prefer Rubio over Trump, but the sample size is too small to report exact percentages…

With all the discussion about hands in this election, Monmouth decided to test how Rubio and Cruz would do mano a mano against Trump. Trump (47%) and Rubio (45%) are basically tied in this hypothetical two person race. Trump (48%) would potentially have a decided edge over Cruz (40%) in a head to head vote in Florida.

Even for an election where the conventional wisdom is getting smashed daily, those numbers are way, way off from what most people expect. It’s Trump who’s supposed to be cleaning up in the early vote, not Rubio. Trump’s the one who’s been leading in the polls there for months and Trump’s the one with the most enthusiastic voters. If anyone’s banking votes early, you’d expect it to be him. In Louisiana on Friday night, he jumped out to a huge lead in the first returns thanks to his strength in the early vote — only to barely hold on as Ted Cruz performed very strongly among people who voted on election day. If Rubio’s crushing Trump in the early vote, it means, I guess, that his turnout operation there is much better than everyone expects.

And if he’s winning big early, thanks in part to a solid organization, it becomes very hard for me to believe that he’s about to get pasted by Trump among election-day voters. After all, following a poor debate last week, withering attacks from Rubio, and some Cruzmentum in Saturday night’s races, Trump’s trend line has been downward lately (the latest evidence comes from Michigan). Trump hasn’t performed well nationally among late deciders either, which makes sense to my anti-Trumper brain: How many undecideds go into the booth and end up thinking, “C’mon, I’m not really going to vote for this guy, am I?” Ahem:

Trump % of early voters / % of election day voters
AR 33 / 28
GA 44 / 38
OK 33 / 28
TX 28 / 17

— Adrian Gray (@adrian_gray) March 7, 2016

On top of all that, outside groups are now (finally!) spending millions bombing Trump with attack ads in Florida. So no, I don’t buy that Trump is crushing Rubio among election-day voters. But I’m not sure why anyone would buy that Rubio’s up big among early voters either. It’s not just a matter of the CW being wrong; it’s a matter of the total sample in Monmouth’s poll consisting of just 400 Florida voters, which means their subsample of earlies is just … 80 people or so. How much can you trust a group that small to accurately reflect statewide opinion?

On the other hand, Rubio’s early-voter lead is big enough that it may be safe even among a small subsample with a large margin of error. And if you’re looking to give Monmouth’s data more credence than other recent polls of Florida showing Trump up big, note they drew their sample from “a list of registered Republican voters, who participated in a primary election in 2012 or 2014 or voted in both of the last two general elections, and indicate they will vote in the presidential primary on March 15, 2016.” Normally that screen might be considered too tight; after all, isn’t Trumpmania all about bringing in people who haven’t been voting recently? Don’t forget, though — Florida is a closed primary. If you’re an independent, a “Trump Democrat,” or a Republican who just never got around to registering with the party, you can’t walk into your polling place on Tuesday in Florida and vote. You’re out of luck. Using a sample composed of people who’ve voted in the past thus might give you a better sense of what’s really likely to happen in Florida than you’d get from simply asking respondents, “Do you intend to vote last week?” And as it turns out, polls that have used that more rigorous method of predicting actual voters next week tend to show a smaller lead for Trump than other polls do. Three separate surveys, one from Tarrance, one from AIF, and now this one from Monmouth, that have used the voter file for their sample show the race within single digits. That’s reason to keep hope alive if you’re a Rubio fan.

I still don’t see any way out of the dilemma I described last night, though, especially if it really is true that Rubio’s banked a big lead among early voters. If Cruzmentum is a real thing on the right after Saturday, with Rubio’s support collapsing everywhere, then Cruz is arguably your better bet to win in Florida. But there are still bound to be many thousands of Floridians who will stick with Rubio out of loyalty, and many thousands more who may have already cast a ballot for Rubio as early voters. We may be headed for an outcome in Florida where Rubio and Cruz split the right-wing vote fairly evenly, which means a crushing win for Trump and much better odds that he’ll go on to win the nomination. Cruz isn’t letting up, either: One of his Super PACs just unleashed a new barrage of ads aimed at … Rubio, not Trump. (You can watch them on their YouTube page.) They’re going to try to take him out on Tuesday night. I think Cruzmentum is probably strong enough to do that, although not strong enough to wrest Florida from Trump.

Update: An alternate theory from Benjy Sarlin: What if Rubio was on his way to a shockingly strong victory in Florida, as reflected in the early-vote numbers here, but is now in free fall there and elsewhere as his supporters defect to Cruz (and Kasich)? That was the lesson from this weekend — Cruz finished strong in all four states because a chunk of Rubio fans switched to support him. If Republicans are giving up on Rubio, that might foretell a disappointing finish in Florida next week and explain why Monmouth is seeing his support among election-day voters as weaker than you might expect.