I mean, if anything, he's not polling *that* badly.
CA polling overestimates GOPers, like Cox in 2018. My prediction for 2020 is 65-31% Dem.
Not really, unless you're only talking margin. The final Atlas polling average for R's with the actual in parentheses and the R over-prediction to the right:
2018 Gov: 40 (38): 2%
2016 Pres: 30 (31): -1%
2014 Gov*: 37 (40): -3%
2012 Pres: 38 (37): 1%
2012 Sen: 36 (37): -1%
Average: -0.4%
On average, the final polls actually tend to nail the R share in CA. What they usually do is under-predict the D share of the vote. If Trump is averaging around 34% in the polls, we shouldn't really expect him to do much different than that.
*I threw out two high undecided outliers, which actually helps the R over-prediction argument.