POLL-BASED
ODDS
MARKET-BASED
ODDS
ANALYST-BASED
ODDS
THE SENATE
WIZARD ℠
LIVE UPDATES!
OUR TRACK
RECORD
ABOUT
US
CONTACT
Analyst-Based Odds
Method
Our analyst-based analysis uses the qualitative ratings from The Cook Political Report, Sabato's Crystal Ball, Inside Elections, and Real Clear Politics. These analysts rate every Senate and House seat with a qualitative rating for either the Democrat or Republican candidate designated as: Safe, Likely, Lean, or Toss-up. Inside Elections has an additional rating: Tilt which reduces the number of their Toss-up races by about half. Note: Cook and Inside Elections designate their strongest rating as Solid whereas Sabato and RCP use Safe. With its final projection, Sabato moves all Toss-up races to Lean D or Lean R thereby providing greater insight to their expectation. Exception: Sabato left both 2020 Georgia Senate races as a Toss-up on election eve. Perhaps fittingly, both races ended in a run-off. As a result of removing often too-close-to call Toss-up races, Sabato's Lean ratings would be expected to be less accurate than the Lean rating accuracy of the other analysts.

Here is the historical accuracy from 2012 thru 2020 for each analyst's rating:

SENATE HOUSE
Rating
Cook
Sabato
Inside
Elections
RCP
Safe D 100% 100% 100% 100%
Likely D 100% 100% 100% 100%
Lean D 100% 73.9% 100% 100%
Tilt D - - 66.7% -
Toss-up 40.5%D
59.5%R
50%D
50%R*
35.3%D
64.7%R
43.6%D
56.4%R
Tilt R - - 80% -
Lean R 100% 91.7% 100% 91.7%
Likely R 100% 100% 100% 100%
Safe R 100% 100% 100% 100%
Rating
Cook
Sabato
Inside
Elections
RCP
Safe D 100% 100% 99.8% 100%
Likely D 96.4% 98% 96.2% 100%
Lean D 87.3% 67.7% 80% 93.2%
Tilt D - - 66.7% -
Toss-up 43.9%D
56.1%R
50%D
50%R*
47.9%D
52.1%R
46.3%D
53.7%R
Tilt R - - 81.6% -
Lean R 93.7% 84.4% 87.5% 92.5%
Likely R 97.8% 98.9% 96.2% 97.9%
Safe R 100% 100% 99.9% 100%

*As Sabato removes all Toss-up ratings from its final projection, we assume that a race currently listed as a Toss-up to be 50%D / 50%R.

Except for Sabato, the Senate ratings accuracy shows no difference between Safe D, Likely D or Lean D as they all have been 100% accurate. Similarly, there is no difference in Safe R or Likely R - both 100% accurate. Lean R is also 100% accurate for Cook and Inside Elections and over 90% accurate for Sabato and Real Clear Politics.

Not counting the 2018 House NC-09, where the race results were voided and a new election ordered, there are 2,174 House seat ratings over the previous 5 elections which provides a greater sample size. Still, the House ratings show Safe approaching 100% accuracy as does Likely. Lean exceeds 90% for RCP and averages 90% for Cook, a bit less for IE, and of course less for Sabato.

The accuracy for the analyst-based qualitative ratings is much greater than is widely assumed. For example, FiveThirtyEight uses the following scale for its Senate and House qualitative ratings:

 Safe(Solid)  Likely D   Lean D   Toss-up   Lean R   Likely R   Safe(Solid)
 ≥95% D   ≥75% D   ≥60% D   <60% both   ≥60% R   ≥75% R   ≥95% R 



Here is a summary of the current number of races corresponding to each of the qualitative ratings for each analyst:

SENATE HOUSE
Rating
Cook
Sabato
Inside
Elections
RCP
Safe D 8 8 8 5
Likely D 1 2 2 2
Lean D 2 3 0 1
Tilt D - - 2 -
Toss-up 4 3 8
Tilt R - - 2 -
Lean R 3 5 1 3
Likely R 2 3 3 3
Safe R 15 14 14 13
Rating
Cook
Sabato
Inside
Elections
RCP
Safe D 159 154 165 141
Likely D 13 17 13 20
Lean D 15 27 7 13
Tilt D - - 14 -
Toss-up 36 20 34
Tilt R - - 8 -
Lean R 13 34 12 29
Likely R 11 16 9 20
Safe R 188 187 187 178





The Senate


The Senate Wizard ℠ calculates the overall probability for each party to win using the odds for each of the 35 races and evaluating the probability of all possible outcomes. The number of possible outcomes we evaluate is 2^35 or over 34 billion.

For each Senate seat, we use as the odds, the historical accuracy for the analyst's qualitative rating. Here are our current projections for each analyst:











Here is our analyst-based Senate Tracker since July:



Our projection, based on the historical accuracy of each analyst's current qualitative ratings, is mixed with Cook and RCP indicating an edge to Democrats, IE indicating a very slight edge to Republicans, and IE a larger edge to the Republicans. These projections are subject to change as the analysts change their qualitative ratings.




The House

Our probability model, The Wizard of Odds ℠, calculates the overall odds for each party to win by using the odds for each of the 435 races and evaluating the probability of all possible outcomes. There are 2^435 possible outcomes or over 88 duoquadragintillion, a number that's 131 digits long.

For each House seat, we use as the odds, the historical accuracy for the analyst's qualitative rating.

Some correlation between races exists, particularly among Toss-up races which can produce "blue" or "red" waves. Our overall House projection adjusts for the variance from the expected results over the indicated election cycles.

Here is our current projection for each analyst:








Here is our analyst-based House Tracker since July:



Our House projections, based on the historical accuracy of each analyst's current qualitative ratings, indicates a strong edge for the Republicans to take control of the House. These projections are subject to change as the analysts change their qualitative ratings.

HOME
POLL-BASED ODDS
MARKET-BASED ODDS
ANALYST-BASED ODDS
TRACK RECORD
ABOUT US
BLOG