Our analyst-based analysis uses the qualitative ratings from The Cook Political Report, Inside Elections, Sabato's Crystal Ball, and Real Clear Politics. We adjust their current ratings based upon their historical accuracy.
When Cook, Inside Elections, and Sabato all agree that the seat rating is Safe for the Democrat or for the Republican, they've been 100% accurate in each of the last 4 elections: 2010, 2012, 2014, and 2016. When a seat is Unanimously Safe or Solid, we're giving it the special rating: U-Safe, instead of simply Safe D or Safe R. Here is a summary of each analyst's current ratings for the Senate and House:
Note: Inside Elections generally has a lower number of Toss Up ratings due to their additional rating of Tilt which is between Toss Up and Lean. For their final ratings, Sabato moves all Toss Up ratings to either Lean D or Lean R. As a result, Sabato's Lean rating has a lower probability of accuracy than the other analysts.
From the Senate totals, assuming that every seat is in the correct column, the Democrats would have difficulty getting to 51 seats with any of the analysts. With Cook's ratings, they would need at least 7 of their 8 Toss Up races. Inside Elections, Sabato and RCP each have 50 seats in the Republican column, so the Democrats wouldn't be able to get to 51.
In the House, assuming that every seat is in the correct column, Cook has the Democrats ahead 208-196 with 31 Toss Up races; Democrats needing only 10 of the 31 seats to take control of the House. Inside Elections has the Democrats ahead 213-206; Democrats needing only 5 of their 16 Toss Up races to take control of the House. Sabato has the Democrats ahead 212-201; Democrats needing only 6 of their 22 Toss Up races to take control of the House. RCP has the Democrats ahead 205-198 needing only 13 of their 31 Toss Up races to take control of the House.
Of course, just because a race is rated Likely or Lean or Tilt, it doesn't mean it will definitely be won by the party designated. What does it mean when Cook indicates a race is rated Lean D? Does it have the same meaning as when Sabato rates a race Lean D? What are the odds that a race rated Tilt R by Inside Elections will actually be won by the Republican? How accurate have these ratings been since 2010? Below is the historical accuracy of the final rating by each analyst over the last four biennial elections:
* The % shown for the Toss Up rating is the percentage of the time that the Democrat wins.
With the historical accuracy of the rating for each race, we have the probability associated with each race. Here is the expected value for each analyst, weighted by the historical accuracy of their race ratings, assuming an equal Democratic/Republican split with Sabato's current Toss Up ratings:
From the historical race odds, we calculate each party's overall odds of winning by evaluating every possible combination of outcomes. For the 35 Senate races, there are over 34 billion possible outcomes; for the 435 House races there are roughly 88 duoquadragintillion possible outcomes.
Note: To be clear, the above shows our calculation of the Democratic probability for controlling the House for each analyst, using their historical percentage accuracy for each qualitative rating as the probability of that rating. These results do not necessarily reflect the analyst's current or future projections. Here are recent quotes from these analysts:
Inside Elections - 10/12: "Most likely outcome: Democratic gain of 25-35 seats, with larger gains possible."
Sabato - 10/4: "Our best guess in the House right now remains a Democratic net gain in the low-to-mid 30s, with enough uncertainty that we would not rule out the Democratic gains sputtering out short of the 23-seat net gain they need."
For each Senate seat, using the strongest of the 4 current ratings from Cook, Inside Elections, Sabato, and Real Clear Politics, we get the following table, with blue designating a seat currently held by a Democrat, and red a seat currently held by a Republican:
Assuming all the seats are won by the party designated, the Republicans lead 50-45 with 5 Toss Up seats. Referring back to the historical accuracy of the seat ratings, when a Senate seat is considered Likely D or Safe D, the Democrat has won the seat 100% of the time; when the seat is considered Likely R or Safe R, the Republican has won the seat 100% of the time. There are 2 Lean D and 2 Lean R seats, each rated as such, by 2 or more analysts. How the Toss Up seats split will decide the margin by which the Republicans win the Senate.
For each House seat, using the strongest of the 4 current ratings from Cook, Inside Elections, Sabato, and Real Clear Politics, we get the following table, with blue designating a seat currently held by a Democrat, and red a seat currently held by a Republican:
Assuming all the seats are won by the party designated, the Democrats lead 216-208 with 11 Toss Up seats. Referring back to the historical accuracy of the seat ratings, when a House seat is considered Likely D or Safe D, the Democrat on average, has won the seat about 95% of the time; when the seat is considered Likely R or Safe R, the Republican has won the seat 100% of the time. Most, if not all, election day "surprises" will come from the Lean and Tilt ratings, which are almost equally split between the parties. How the Toss Up seats split will likely decide which party will control the House.