I’m not aware it’s possible to project the outcome with any confidence based on early voting patterns. I’m not disputing what you say but I’m curious as to where it comes from. Is there a statistical correlation model which produces such a prediction? Is it a computer model? Is it a methodology you developed? Just asking.
Other states, such as NC, also have heavy early voting, but in 08 Ds turned out higher on Election Day, so it's important the Rs LEAD in absentee/early voting (which they have in NC).
IA and NV are just the opposite: there the Ds build up big early voting leads.
We've explained the science here dozens of times. Have you been watching the threads? Let tell you what sold me on this reality. In 2004, late at night they were still counting ballots in IA. Michael Barone was on FOX looking at just one or two counties. He said, with the counties still yet to report, Bush has won. The counties were composed in such a way that if they voted at traditional percentage splits, Bush would win.
That's all we do here. We look at which counties turned out in what percentages, and just draw trend lines. If you have a red county turning out 20% higher than in 08 when it lost by 10% it doesn't take a genius to see that the Rs will win that county.
Every state has key counties that tell you all you need to know. If the early vote is held own in Clark Co. NV. , we win. If the absentee/early vote is heavy in Warren Co. or Hamilton Co. it's pretty likely we win.