After my conversation with Andrew yesterday about The Economist election forecasting model I got curious about how G. Elliot, Merlin and Andrew want their prediction to be assessed given the menu of strange contingencies we have in front of us.
I checked Betfair rules for some guidance:
This market will be settled according to the candidate that has the most projected Electoral College votes won at the 2020 presidential election. Any subsequent events such as a ‘faithless elector’ will have no effect on the settlement of this market. In the event that no Presidential candidate receives a majority of the projected Electoral College votes, this market will be settled on the person chosen as President in accordance with the procedures set out by the Twelfth Amendment to the United States Constitution.
This market will be void if an election does not take place in 2020. If more than one election takes place in 2020, then this market will apply to the first election that is held.
10/03/2020 10:30am – Once voting (whether postal, electronic or at the ballot box) begins in the year 2020 for the US Presidential Election 2020, the election will be deemed to have taken place for the purposes of this market. We will then settle the market as per our rules regardless of whether the election process is fully completed in 2020 or beyond.
A correspondent replied: “betfair lawyers r sh*ttin brix”
So what happens if Biden wins, Trump goes ballistic (no if there), Democrats impeach, and the Republican Senate assures an “orderly transition” with 20+ of their senators recognizing the unique opportunity to rehabilitate themselves well before they have to face *their* voters again? Will the next president forecast be correct then?
We can always go further into the tail… a contested outcome, widespread civil unrest, no mechanism for resolution, calls for the military to step in, a second constitutional convention, etc.
People want an oracle
I guess the great thing about this forecast is that they don’t have to shit br*x on the spirit vs. the letter here — nitpickers & well-actuallies can’t take them to court (I know it’s a forecast of “vote intentions for Election Day” but we all know what we do with this forecast)
Incidentally, the Betfair “Trump Specials” market currently has the probability of Trump finishing his 1st term at 90%. This is higher than the current, and still too low, 98% chance of Biden winning Cali that we discussed (here and here). The Trump Specials market isn’t nearly as deep as Cali tho — you can get a 9-to-1 payoff betting against him finishing his first term only up to a $10k stake before you exhaust all sellers based on current limit orders. But still.
I know Andrew doesn’t think much of betting markets: “The market doesn’t produce the forecast so much as it motivates investors to find the good forecasts that are already out there,” and neither does Nate Silver: “Political betting markets are basically just a competition over what types of people suffer more from the Dunning–Kruger effect.”
But these markets do tend to order the probabilities correctly, even in the olden days before polling. Of course this isn’t magic — bettors may employ various simple and useful heuristics like the “nature of the times” (see Philip Converse’s work), or more disciplined algorithms that tally factors for & against (e.g. see Allan Lichtman’s model).
Thoughts?
—
Full disclosure: I have a stake backing YES here, it is not a big one, but with long-shots it doesn’t have to be.
Ps. Looks like there is free money in the popular vote betting market, 16% net return on investment if you back Biden winning the popular vote. Of course, if there is a 10% chance of madness your stake could sit in escrow for awhile.
PPs. If I sound like I have my head in the sand it is because I do. I haven’t checked the news in two weeks and we were just released from a 3.5-month lockdown here in Melbourne, Australia.