How screwed are the dems really on the house?

 So I've been looking at RCP's election map and they have it where the GOP win is inevitable. Even worse, they have it where there's 227 relatively republican safe seats and 34 toss ups. If the GOP wins half of those toss ups, they will control 244 house seats, which is a rather large majority. Meanwhile 538 predicts something closer to the GOP having somewhere between 246 and 214 house seats within an 80% confidence interval. So probably 230 on average In my model based on RCP's data, a Democratic win isn't even within the 95% confidence model (or 97.5% with a 1 tail result). So what's going on here? Well, let me take a closer look at all of the 34 races that RCP has has tossups.I'm not gonna do fancy charts, I'm just going to get as much of a read on these races as I can.

 AK-AL: D+6
AZ4: D+7
CA9: Tie
CA13: Tie
CA22: D+5
CA47: N/A
CA49: D+6
CT5: R+0.5
GA2: D+3.5
IL6: N/A
IL13:D+3
IL14: N/A
IN1: D+7
KS3: D+14
MD6: R+2
ME2: D+4
MI3: N/A
MI7: D+3
MI8: D+3
NC1: D+6
NH2: D+8.7
NV4: D+4
NY3: D+1
NY4: N/A
NY17: N/A
NY18: N/A
NY19: D+5
NY25: N/A
OH9: N/A
OR4: D+1
OR6: D+3
TX28: N/A
TX34: D+4
VA7: N/A

So, let's count this up. 

Out of 34 races, we have 19 D leaning races, 2 R leaning ones, 2 ties, and 11 with no polling data. Admittedly most of these races have very little polling data, and some of these races have polls from 3-5 months ago. So not a good indication of what's reliable. Still, looking at it, this looks like a bit of a slaughter for the GOP. They're only expected to outright take 2 of these races? They could take more. But even if all of the NAs go GOP, that's still less than half. If we split them 50/50, the dems would win 6 of them and the GOP 7....giving the GOP the extra one to be generous. So that's still 25D wins and 9 R wins. That would give us a house of 236R and 199D. That's still a GOP win, but let's look at the "leans GOP" races to see if the GOP lead is as high as it's forecasted to be. There's 29 of these races:

AZ1: N/A
AZ6: N/A
CA27: N/A
CA41: N/A
CA45: N/A
CO8: N/A
IA1: N/A
IA3: R+8
IL17: N/A
MN2: D+1
NC13: D+1
NE2: R+15
NH1: R+1.3
NJ7: R+8
NM2: R+2.3
NV1: R+6
NV3: R+5
NY22: D+4
OH1: N/A
OH13: R+9
OR5: N/A
PA1: R+7
PA7: D+1
PA8: D+6
PA17: Tie
RI2: R+8
TX15: R+4
VA2: Tie
WA8: D+4

Now, a lot of these races are put in the red column for a reason. The "fundamentals" lean red. Ie, voter base does historically vote red. And again, polls are often scarce on this level. But let's count this up. 11 outright R wins. 6 D wins, far more than there were R wins in the "toss up" column. 2 ties. And 10 unknowns. Based on what probabilities would dictate, we should only see 1-2 Rs in this column. Even if we give the Rs every unknown, we're talking 23 R wins and 6 D wins. 

Now let's look at the 13 "leans Dem" races:

CA25: N/A
CA26: N/A
CO7: D+3
CT2: D+19
FL23: N/A
NJ3: D+6
NJ5: N/A
NJ11: N/A
NM1: D+8.5
NM3: D+18
PA6: N/A
PA12: N/A
VA10: N/A

So here we have 5 D wins and 8 unknowns. And I bet those unknowns will go all D. 

So, let's really break this down now. We can count 161 safe D seats, 198 safe R seats. 

All 13 D leaners go to the dems, bringing them up to 174. 

For the leans GOP, I'll give all the N/As to the GOP, and then distribute the rest based on probability (1 tie for each party). This is 22 R wins and 7 D wins. So the GOP will have 220 and the dems will have 181. With that, the GOP wins with a solid 220 seats. They need 218 to control congress.

 Now let's go with the toss ups. Dems win 19 outright, the GOP wins 2, we can give 1 tie to each, and then for the "no datas", I'll give 6 to the GOP and 5 to the dems. So that means the dems will win 25 here and the GOP will get 9. 

That brings the GOP up to 229 and the dems to 206. 

The GOP wins, but it won't be a HUGE lead. And there's a HUGE amount of uncertainty in these results as there are so many unknowns.

But, that's my best guess to guess the house, and we get the exact same results that 538 does.They seem to think that 230ish is the most likely outcome, and I come to 229. So yeah. I just wish RCP did a better job compiling the data here, they make it looks like the median outcome is gonna be like 244 seats. That's ridiculously high. I expect a GOP win, but yeah, 538's data is more reliable here. And I will generally agree with their prediction.