On August 14, the day before Trump’s economic speech, signifying the start of his PIVOT, his performance in 538’s polls-only model (which smooths out day-to-day bounces in random state polls) was at an August low of 10.8%. Today, two weeks after the start of the PIVOT, it has now risen to 19.1%. In the NowCast (which is more accurate at stating the exact present state of the polls, smoothing to adjust for outliers be damned), his performance was at 9.5% on August 14, and is now 19.3%. If he continues this performance of growing his chance of winning by 8 to 10% per two weeks, he should be at 27% on September 11 and 36% on September 25. However, I expect his speed to improve a bit over time, as people get more used to his inclusive message. So I expect Trump to be polling at ~40-45% just before the first debate.