If I had to guess right now, I'd predict GOP loses 40-50 House seats in November after things move away from them in Sept-Oct like the president's party in the last 3 wave midterms. But even still, I could see gerrymandering costing Dems >20 seats based on my own nonpartisan maps 
Unless 2018 turns into a Dem tsunami the likes we haven't seen since 1974, I just don't see an outcome where gerrymandering doesn't cost Dems a significant number of seats. Yes, some gerrymanders will fail in November. But they won't *backfire* & help Dems
There's assuredly a level of a Dem wave that would actually cause those GOP gerrymanders to backfire. But it's historically large & still doesn't negate the fact that the GOP likely would've lost the House in 2012 & possibly 2016 too withot gerrymandering
