I think a ten points national margin of victory is looming for the Democrats in the House.
There's still a lot of time for the environment to change, but if it doesn't this is probably in the ballpark.
The article shows that in 2006, the average special swing was 17 and the final margin was 8. Ten seems like a fairly good estimate for this year barring something dramatic
.
The difference between specials performance and midterms performance in 2006 is a large anomaly compared to other midterm cycles, so it's not particularly wise to use it as a baseline of expectations. Which isn't to say that a 10% point PV win is an unreasonable estimate for Democrats come November, I think arguments could be made for anything between a 7-17% PV win, but the 2006 special/midterm difference is not a good framework to base guesses on.