
This is how it goes with prediction
science. There are no crystal balls.
It reminds me of an unhappy week I spent at the happiest place on earth – Disney World. TV ratings came in for new broadcast premieres and
we started comparing them to predictions for their program development concepts. It ruined my vacation.
For this election, the methods with the best track record mostly missed and
some of the newer betting-related sources did better. Most had some predictions that panned out and others that didn’t.
Following is how each performed based on Tuesday’s
predictions:
The Polls
FiveThirtyEight’s polling average had Harris up by 0.9 (Harris 47.9 vs. Trump 47.0). The numbers looked to be within
FiveThirtyEight's error band, forecasting a tie or a no call.
advertisement
advertisement
If you read the results as a “no call,” you could say it wasn’t wrong or right. If you took it as a
tie, you could say it was wrong. I prefer to go with “no call.”
The Prediction Markets
Iowa Electronic Markets (IEM) favored a Democratic victory, although, its vote-share market was so close, I listed it as a “no call” based on Media Predict’s
rules. IEM’s Winner-Take-All market clearly missed. it favored Democrats 74% to 24% for Republicans.
The Keys
The 13 Keys prediction system experienced a rare miss. This is the first time the keys were wrong, and
not due to a discrepancy between the popular and the Electoral College votes.
Professor Allan Lichtman may want to go back to his research and add a key or two. There isn’t a
key for the candidate being male or female. As there are only two instances, he wouldn’t have much historical data to establish a basis.
The Betting Market
From a prediction perspective, BetOnline’s 2024 presidential election odds to win were right (Trump -147, Harris +127). Its election-winner-to-lose-popular-vote odds were wrong (No
-170, Yes +130). Its popular-vote-winner odds were also incorrect (Harris -425, Trump +315).
The Stock Market
The NASDAQ share price for Trump
Media Technology Group Corp. (DJT ticket symbol) proved to be a prescient signal (+75.13% year-to-date).
The Dow Jones Industrial Average (DJIA) Model missed. It predicted a 70%
probability that the incumbent White House party would win. As it’s solely based on stock market performance, it doesn’t account for other variables.
The
Misery
The Misery Index missed. The fact that it wasn’t correct says a lot about why the Democrats lost.
The Models
The Political Economy Model (Charles Tien, Hunter College & The Graduate Center, CUNY; Michael S. Lewis-Beck, University of Iowa) and the Prospective Financial Future & Length of
Time in the White House Model (Brad Lockerbie, East Carolina University) were both no calls. Neither right or wrong.
The Voting
Early voter
registration favored the Democrats. It proved to be an unreliable signal for this election.
There will be a lot of talk about why the Democrats lost and much of it will be
speculation or myth.
It has come up that if Harris would have chosen Josh Shapiro instead of Tim Walz as her running mate, she might have won Pennsylvania. Shapiro is the popular
governor of the state.
This is true, but PA only accounts for 19 electoral college votes and she’s currently short 46.
Another one is that if Biden would have
stepped down earlier and the Democrats had had a real primary contest they may have come up with a better candidate.
Harris performed a little worse than Biden in a lot of counties
where she needed to do as well or better. The underlying assumption is that gender, the fact that she’s female, could have caused her results to be lower than his. A male candidate might not
have experienced the same decline.
Historical data shows that candidates who are highly contested in the nomination process usually lose.
There will be some people
that blame inflation and the economy for the outcome.
However, the Misery index considers both inflation and unemployment and it was below the threshold for when the incumbent party
would be predicted to lose an election. Therefore, we can surmise that Harris likely lost for other reasons.
Another interesting point about this election is that Trump won without
the help of Cambridge Analytica (CA). Although, it’s likely that some of my former colleagues worked on this campaign while at other companies.
For those of you who supported
Trump, congratulations. For those that didn’t, I’m sorry for your loss. 2028 will be here before you know it.
Note: This op-ed assumes the final popular vote outcome
doesn’t change as of this writing (Harris 47.6% vs. Trump 50.9%).