I was thinking about a potential swing state (Georgia) and the fact that the Democrats got 8 percentage points more votes than the Republicans. I was thinking about the historical accuracy of this as a metric for partisanship. How accurate were Senate primary totals in predicting statewide winners in presidential election years since and including 2008? What I mean is the party with the most votes in Senate primaries going on in the presidential election and winning the electoral votes.
In general, there is a decent correlation between the margin of victory in the Senate primaries - in terms of the total number of votes - and the margin of victory in the Presidential election. However, it is questionable whether this metric is a good one to use when predicting the result, for the same reasons Joe W laid out in his answer to a related question. The data I used to examine this is from the FEC election results found here.
The graph below shows that there is a fairly clear overall relationship between the two variables, with a correlation coefficient of ~0.68. However, we can see also that the relationship varies quite significantly by year. For example, in 2008, six states recorded higher participation in the Democratic Senate primary than the Republican primary - but voted for McCain in the presidential election. On the other hand, four years later in 2012, seven states recorded higher participation in the Republican primary - but supported Obama in the presidential vote.
To take your example of Georgia - in 2008, the Democrats had far higher participation in their Senate primary - 812,280 votes compared to 392,902 in the Republican primary. In 2016, this was the other way around, with 577,660 votes in the Republican primary compared to 310,053 in the Democratic primary. On both occasions, the margin of victory in the presidential election remained the same, at ~5 percentage points in favor of the Republican candidate.