What evidence is that?
Here are some contrarian indicators I have on my radar:
1. RCP averages - She leads in MI and WI, while he leads in PA, just narrowly. PA is now the must-win for Trump. VA and NH and NM are pipe dreams for him. Whereas NC is not a pipe dream for her.
2. 55-45 female-male voting ratio in early voting. This may be a misleading indicator and may correct itself on election day, but all polls had a 52-48 or 51-49 turnout model.
3. Multiple recent gaffes from the Trump campaign (comedian, comments against Liz Cheney, etc.) whose fallout is reverberating on social media. The Biden/Harris campaign gaffes and fallout are well covered here. But the Trump gaffes are ignored or downplayed. They are having an effect, like the Biden gaffe, and perhaps may cancel each other out.
4. The hidden Harris voter. While pollsters have had 3 election cycles and 8 years to improve their Trump forecast models, they only have had 3 months for Harris voter polling. It is possible they are missing some Harris voters on the fringe. On the other hand, the likelihood of polls again underpredicting Trump support like they did in 2016 and 2020 is nearly zero. And it is possible some polls are actually "overpredicting" Trump's strength by overcompensating for past poll misses.
5. Trump's own schedule - he is doubling down on rallies in NC. This state looks very tentative for him, other signs notwithstanding.
6. Dem ballot harvesting machine is second to none. We have no comparable ground game, except to complain about fraud afterwards.
Having said all that, I hope he wins, but people here chest thumping and being certain about him getting 312 EVs or 322 EVs etc. are repeating the same mistake people here have made in 2006, 2008, 2012, 2018, 2020, and 2022.