It’s a two-point lead, 43/41, which is within the margin of error, but man oh man did he need some good news out of Florida after the bloodbath in the last two polls.

This FAU survey is the first time he’s led in the state since before the conventions. How we square this result with the recent Monmouth and Saint Leo’s polls showing Hillary up by nine and 14 points, respectively, lord only knows.

For the presidential race, both candidates are underwater in terms of their favorability ratings in Florida. Trump scores a 41 percent favorable and 56 percent unfavorable rating, while Clinton has 40 percent favorable and 58 percent unfavorable rating. Clinton, meanwhile, has a loyalty score of 90 percent (those who have a favorable opinion and are voting for her), while Trump has a loyalty score of 94 percent. Trump leads among males 46 percent to 36 percent, but trails Clinton among women 45 percent to 41 percent. Independents are voting for Trump by a wide margin of 47 percent to 26 percent.

Trump leads among white voters 49 to 33 percent, but trails with African Americans 68 to 20 percent, as well as Hispanics 50 to 40 percent.

“The race between Clinton and Trump among Hispanics in Florida is closer than it is nationally,” said Monica Escaleras, Ph.D., director of the BEPI. “Some of that is probably the Cuban vote. Trump’s support among Latinos in Florida is helping him stay competitive.”

Credible or not? Well, FAU has Marco Rubio up five points over Patrick Murphy, 44/39, in Florida’s Senate race, which is right in line with other polls of that election. If there were some heavy Republican “house effect” happening here inflating Trump’s numbers, you’d expect a super-sized lead for Rubio too. Instead you’ve got a plausible result for the Senate race, lending credence to the presidential numbers. (Then again, yesterday’s Saint Leo poll showing Hillary blowing out Trump also had a plausible Senate result.) FAU also did reasonably well in predicting the Florida primary this year, although not as well as Saint Leo did. FAU’s final primary poll had Trump winning by 23 points; he went on to win by 18.7. Overall, FAU scores a weak grade of C+ on FiveThirtyEight’s pollster scorecard (with a four-point Republican-leaning “house effect,” meaning that you’d expect an actual result of Clinton by two in Florida based on these numbers), but Saint Leo also scores a C+. There’s no reason, in other words, to give yesterday’s “Hillary blowout” poll any more inherent credibility than today’s “Trump leads narrowly” result. People might pay added attention to Saint Leo only because the result is more in line with the results of other polls of the state. Which is a different thing than saying all of the other polls must be right.

On the other hand, there are some numbers in the FAU data that jump out as … highly surprising. The Hispanic result, with Clinton up 50/40, isn’t one of them. Obama won Hispanics 60/39 in Florida four years ago, and Trump has performed as well with Latinos as Romney did in various polls this year. It’s a little surprising, I guess, that he’d actually be ahead of Romney’s pace, but maybe Rubio’s backing has helped Trump a bit. The split among black voters, though, is very surprising: 68/20 is an awfully good showing for Trump among African-Americans when Obama won that group … 95/4. Maybe his outreach to black voters over the past 10 days is helping? We’ll need more data, but that’s a good start. Another eye-popping pro-Trump result: Clinton leads among voters aged 18-34 by just three points, 38/35. Young voters have been one of the most consistently poisonous groups for Trump, with him finishing behind Gary Johnson among the 18-29 crowd in more than one other poll this year. To be within three points of Clinton is mind-boggling considering that Obama won the 18-24, 25-29, and 30-39 groups by double digits in Florida in 2012. (His margin among voters 18-34 was 34 points.) If you’re looking to be skeptical of these numbers, that’s the best reason.

Two other numbers worth noting. FAU sees the same familiar gender gap that other pollsters see, with Trump winning men while Clinton wins women, but their result is unusual in how narrowly Clinton wins women, by a mere four points. Quinnipiac’s recent poll of Florida, which had the race dead even at 43, found her up 13 points among women. In 2012, Obama won women 53/46. The first major-party woman nominee is underperforming him? Hmmm. Finally, Trump is killing it here among independents with a 47/26 lead, which also contradicts the 2012 numbers — although in this case maybe there’s good reason for that. Obama won Florida indies, 50/47, but it may be that between Trump’s “outsider” quasi-independent brand and the presence of Johnson and Jill Stein in the race to poach left-leaning independents from Hillary that he’ll strongly overperform Romney this year (or she’ll weakly underperform Obama’s totals). Worth noting, though: The same Quinnipiac poll that had Florida tied recently also had independents splitting 37/36 for Clinton. An 11-point discrepancy on indies in the polls matters a lot, needless to say.

By the way, a separate poll out of North Carolina today has Clinton winning there by two points, which is right in line with other surveys. Romney won that state by two percent four years ago.