Those percentages are, of course, based on assumptions, either that everyone is an average hitter (and runner) in the specific situation or some additional assumptions that haven't been documented. Is the 43.8% based on an average hitter's likely outcomes, or Stott's specific likely outcomes against that pitcher at that time? And Bader's and Kepler's and Turner's likely outcomes afterwards? And Castellanos being the runner on 3B or 2B? Does it consider that Castellanos has a lower probability of scoring from 2B on a single than an "average" runner? Considering all that (which probably requires information that is not really known), the change in win probability might have been positive--or more negative.
Sorry for the mini-rant, as someone who develops mathematical models for a living, I've seen too many people ignore the impacts of the simplifications required in the assumptions. It makes things worse when we express these probabilities so precisely. 