I know that Trafalgar got it right in 2016, but now that pollsters are compensating for the errors they made in 2016 (esp. weights by education), Trafalgar's polls having a consistent slant for GOP candidates should make us treat them like R internals at this point.
They did well in some places, and poorly in others.
MI/PA/FL and I believe Wisconsin (I can't recall if they polled there in 2016) were very accurate but they had an awful R slant in AZ/NV/TX/GA and D slant in OH.
They're still good to watch in those states they did well in.