Some people think that political betting markets are more accurate at predicting elections than pollsters are. One such betting market, with real money, is British casinos
There is much noise here, and maybe I made a math mistake, but the overall punters think outsiders match up better against Trump than DC Democratic Party insiders. This is not surprising if you believe Martin Gurri's well-argued thesis.
Here is a snapshot of the various odds of the presidential winner on Jan 20, 2019:
What those a number X means is a bet of 1 pound pays off X pounds. So you can read it as a probability of 1/(X+1) if the bet is "fair". So (1-1/(X+1))*(-1) + (1/(X+1))*X = 0. There are several funny things about this if you want to read these as the probability the punters think candidates will win. First is that different casinos have different odds. Second is that those odds don't add up to 100%; the casinos make money but taking a cut. So here are the betting odds for the winning party:
Let's go with the median of the different casinos. If we convert into percentages of the median odds we get:
Dems 48%
Reps 58%
Ind 2%
Tot 108%
So the house percentage is represented in the approximately extra 8%. I don't expect precise answers out of any of this. But let's proceed and see what we can infer. Here is the odds for individual candidates winning the Dem primary:
To tease out how the punters think the various candidates match up against Trump we can note the betting odds give N, the probability of winning the nomination, and G, the probability of winning the general. We can note that the probability of winning the general is:
G = N * P
where P is (N | P) the conditional probability that the candidate wins the general given they win the primary, so:
P = G/N
So what are those G and N for the various candidate? If we assume that all probabilities are distorted by the same multiplier, they will cancel in the equation above. So we can just take the Y in the payoff on nomination, and the X on the payoff on general election, and we get:
P = (X+1)/(Y+1)
Using the medians for each X and Y, then P: probability of a nominee, if they win the primary, to win the election for the candidates above not including Tyson (note I would vote for him!) are:
Yang 0.63
Klobuchar 0.61
Bloomberg 0.54
Sanders 0.50
Biden 0.47
Buttigeig 0.45
Warren 0.39
Clinton 0.37
