They're a professional polling organization. They wouldn't get paid if they guessed. If they were right in 2016 and tank now, then there most likely some amount of "luck" involved. If they get it right again, then I don't know what that statistical odds would be at them guessing correctly two elections in a row. They didn't just get the overall result in 2016, they got something like 48 states right, and the two they didn't get right they didn't actively poll in.
I don't know exactly how polling works, but it's done by professional statisticians and the like and is a big-money field.
What Trafalgar got right in 2016 and everyone else got wrong was determining with accuracy who was lying to them.
Like I said, if they get it wrong this year then it would appear that 2016 involved some amount of luck. If they get 48 of 50 states right again this year, then I have a hard time believing it's guessing. The odds against that would be mind boggling and bordering on impossible.