Beto will win this race on election day, not by daily tracking polls, in which failed to predict the chances of Gillespie against Warner. Which it will be an upset if Beto were to win
Even in the 2016 presidential election most of the polls were predicting that Hillary Clinton would win before the whole world was stunned when Donald Trump won the election thanks to Wisconsin, Michigan and Pennsylvania. So even there, the polls failed to predict the outcome of the presidential election.
So we should expect such surprises for the 2018 midterms.
We *should* expect surprises in the 2018 election, as with all elections, but doing so on the basis of polling error in 2016 is unfounded.
The polls in 2016 showed Clinton with a 1-3 point lead in the popular vote and she ended up with a 3 point lead in the election. It wasn't the polls that failed in 2016, it was the interpretation of them by forecasters, most of whom did not appreciate how tenuous Clinton's lead was in the states you mentioned.
Anyway, back to Texas: is there any possibility that Cruz could depress his own turnout by being insufficiently hard line about the ICE child refugee prisons and family separation policy? It seems like the sort of thing that Texas Republicans would take the most extreme positions on, but Cruz has been somewhat soft on the issue.