It's definitely believable at this stage when the NRSC smear campaign hasn't started yet. It's worth noting that this firm was pretty accurate in 2012, even if they are an internal. I'd guess it depends on what you want to believe about the national environment
Most Democratic internals were (obviously) pretty accurate in 2012, just like most Republican internals were in 2016. Tells us little about 2018.
Do we even know which candidates they polled? Generic "Republican candidate" or maybe some random state legislators who aren't going to run anyway? This entire poll is quite suspect, like most multi-state surveys are in general.
"Generic [party]" tends to over-perform compared to specific candidates.
