I’ve been doing analyses of house races in which I fairly conservatively evaluated each of the close races. Here, I use a slightly different approach. I use only the most recent non partisan poll to estimate the chance of a Democrat vs. Republican winning the race.

When I do this, I get just enough Democrats winning to remain a minority but by only one seat. However, that puts seven seats NOT in the Democratic column because they are estimated to be at exactly 50-50.

These seats, clearly key, are FL26, KY06, MI11, NM02, NY19, OH12, and UT04.

If those are really 50-50, then a conservative two or three seats among them will be Democratic at the end of election night, and the Democrats will have a slim majority in the House.

If we insert a magical blue wave, raising all the chances of Democrats winning by one percentage point, then the Democratic majority is much larger, to the tune of about 27 seats. A two percent blue wave gives the Democrats, interestingly, just a few more, to reach 35 seats.

However, that is unlikely, if the Blue Wave is already factored into the polls.

We might think of there being two Blue Waves. Or, a Blue Wave and a Blue Tide added together. Or a Blue High Tide with a Blue Storm Surge on top of it. Whatever. Point is, there is the general, expected, electorate, which I assume the polls cover, and that may be enough. But added to this is the added votes from the ever mythical “they never vote but they might come out this one time,as if” vote. That is where the one or two percent extra could come from. I suspect the effects of that extra spring tide may be seen in only some districts.

Here is my raw data, ranked from highest proportion to Democrats to lowest. Feel free to argue.

PA05 0.60

PA06 0.60

MN02 0.57

VA10 0.57

CA49 0.57

PA07 0.56

PA17 0.56

AZ02 0.56

KA03 0.56

CO06 0.55

NJ07 0.55

FL27 0.54

IO01 0.54

MN03 0.53

KA02 0.53

IO03 0.51

IL06 0.51

CA10 0.51

MN01 0.51

CA45 0.51

NJ11 0.51

NY22 0.51

CA39 0.51

ME02 0.51

FL26 0.50

KY06 0.50

MI11 0.50

NM02 0.50

NY19 0.50

OH12 0.50

UT04 0.50

CA25 0.50

TX32 0.49

IL14 0.49

PA01 0.49

TX07 0.49

NJ03 0.49

NC09 0.49

CA48 0.49

FL15 0.49

MI08 0.48

VA02 0.48

VA07 0.47

NC13 0.47

MN08 0.41

PA14 0.40

Addendum:

Looks like New Mexico 02 is going to the Republican. That is one of the 50-50 districts in the above chart.

As of this morning, fivethirtyeight.com gives the dems an 85.1% chance of taking control of the house.

The Economist has a model, https://www.economist.com/graphic-detail/2018/05/24/whos-ahead-in-the-mid-term-race , gives the dems a 69% chance of taking control of the house.

PredictWise, https://markets.predictwise.com/politics/2018-congress-house , gives the dems a 67% chance of taking the house.

These numbers are just for a majority, not the size of the majority.

Those are all sources that gave Clinton a massive likelihood of wining over Trump. I gave that race a 50-50.

I’m not using the same model here as I did for Clinton-Sanders or Clinton Trump, but I’m also not making the same mistake the pollsters might be making now.

As I understand the above, Greg is saying that if the dems win all, but only, those seats listed above as having a probability above .5, then they remain a minority by one seat. There are 24 seats listed above with probabilities greats than .5; thus to win a majority the dems need to win at least 25 of those seats listed above.

Using the probabilities above, Greg seems to be predicting, by my calculations (a Monte Carlo type program based on 1,000,000 simulated elections), a probability of about 39.7% of winning the house.

We can note that this a substantially less that the probabilities others are giving in my previous comment.

At this time, I’d give the dems just under a 50% chance of winning the house with sure seats. In other words, those are not in a probabilistic model, they are simply seats they will win, as is the case with most seats the Dems and GOP will win.

Of the remaining seats, if they win only one or two, they get a majority.

Keep in mind that out of hundreds of reps, a certain number die, go to jail, or bail, so you need a majority of more than one to be meaningful.

Of the remaining tossup seats, they are all likely to go Republican, but it is unlikely they will all go Republican. As it were. So this post is predicting a Democratic majority but a small one.