The House models were a disaster this year, the Economist had the GOP house seat range with their 99% confidence interval between 171-208 seats, the GOP could get up to 214 seats when all is said and done, the GOP had an under 1% chance of winning 213 or 214 seats with the dem chances being over 99%.
In the 538 model, error margins were wider, this helped their model be more accurate but still the GOP winning 214 seats was on the 5th percentile of outcomes, and 213 seats was the 6th percentile.
The people who made the model for the Economist are ~idiots
https://twitter.com/gelliottmorrisbut all models will suck if there will be 5+% polling error>>>