Polling companies are going to have to restructure their methodologies,
I believe, after tomorrow's election. I think their sampling is no longer valid. Demographically, it is not measuring things as well as they might think.
Tomorrow, after all the votes are counted, I believe we're going to discover that from 3-4% of the Clinton vote was not accounted for in the aggregate of polling. Where that missed vote comes from isn't completely clear, but I think we'll see it reflected in the results in most states, if not all.
Hispanics, for example, appear to be under-sampled. I also believe that there is an poorly sampled set of women who are nominally Republican.
In 2016, the dynamics of the electorate have changed, I think. Those changes were not fully anticipated and the accuracy of polling is probably not what it should be.
Will I be right? I think so.