They also had Bullock ahead by 12 in October 2016, and he won by 4 on election day. This does look believable though, and Tester +3 is basically my current prediction as well. Lol@ the idea that this race is Likely/Safe D or less likely to go R than NV/WI/ME/etc. because of Candidate Quality™, though.
Does this pollster typically overrate Democrats in general, or is it just all over the place?
Their final 2012 poll has Romney+14, Hill+2, Rehberg+3 (compared with actual results of Romney+13, Bullock+2, Tester+4), so I think it's just not a great poll rather than being skewed in Democrats' favor a lot.
Also, my perception is that Montana is a late-breaking state broke late Republican in 2016 and late Democratic for state races in 2012, given the general high levels of fluctuation the state sees and the general fact that polling averages included some pretty prestigious pollsters in both years and were still off consistently.