Alex Tabarrok has an excellent post at Marginal Revolution this morning explaining why he had and has confidence in prediction markets. It’s very hard to argue, while predicting, against people who are putting their own money on their own predictions.
Like Alex, I tracked those markets closely, which is why I was telling friends that I expected Donald Trump to win the presidential election and the Republicans to take the Senate.
I also had my own personal prediction machine that told me Trump would win. Admittedly, it was after the polls had closed in the eastern time zone, but it was only a little while after.
If you followed the election, you know that one of the big issues was which way Pennsylvania would go. If Donald Trump were to win Pennsylvania, he would likely win the national election. We all knew that it would be close but we were also told that it could take hours to count the Pennsylvania vote, as it did.
But New Jersey abuts Pennsylvania. Why not, I thought, use New Jersey as a leading indicator of the vote in Pennsylvania? I said to my wife, while we were watching the results live, that if Donald Trump gained at least 4 more points in the popular vote in New Jersey than he gained against Biden in 2020, he would win Pennsylvania. Why? Because in 2020 Trump had lost Pennsylvania by only 1.2 percentage points. So with a gain of at least 4-points in New Jersey, relative to his share in 2020, he would likely get at least a 1-point margin in Pennsylvania. We found out early that Trump beat his 2020 New Jersey percentage by about 5 percentage points. He ended up getting a 2-point margin in Pennsylvania.
I’m not saying that my method was better than the prediction markets: my method was clearly worse because it gave me results so much later. But it was way better than sitting there in the early evening PST or late evening EDT, wondering, like millions of Americans, who would win.
Note: By the way, I won $40 from a friend on Facebook and $10 from a neighbor, betting that Trump would win. I made these bets 2 to 3 days before the election, and what gave me confidence was the prediction markets. These are the ones I followed.
READER COMMENTS
Craig
Nov 7 2024 at 4:58pm
“Why not, I thought, use New Jersey as a leading indicator of the vote in Pennsylvania?”
I’m former North Jersey so yeah, maybe (I get it) but you are crossing the pop/country line there. West has Rockies which are more impressive than Apalachains but make no mistake about it the Apalachains are a sub-cultural line…..
David Henderson
Nov 7 2024 at 6:47pm
That’s a good point. I get it. That’s why I built in a fudge factor: a 4-point lead in New Jersey had to be consistent with only a 2-point lead in Pennsylvania. And remember that Pennsylvania includes Philadelphia, which is much like New Jersey. Sure enough, I think I recall that the data nerds found that Philadelphia’s numbers were a few points less unfavorable to Trump this time than in 2020.
Craig
Nov 7 2024 at 11:13pm
I specifically decided not to watch the results until a bit later in the evening since really its not a sporting event, its Schroedinger’s cat, its either dead or alive, we just haven’t opened the box[es] yet. So I decided to watch the betting odds because I found that interesting and before the polls closed it was a slight edge for Trump. As the evening progressed the odds got worse and worse for Harris. I finally look at a map at 11pmish and I see GA, PA, MI, WI, NV and AZ still out there and of course people had been discussing PA so I ‘zoomed’ into PA and at that specific hour I did not understand why the odds were ballooning against Harris because it was close and a significant portion of Philadelphia still hadn’t reported. And so I checked Philadelphia vs Philadelphia in 2020 and it was 2 points better but I still didn’t see why the odds turned so heavily at that particular time. Nevertheless at 11pm the betting markets had it figured out that the election was effectively over, it was 35:1 by 1130p and from what I was seeing, it looked like it still needed to be decided and that things still hung in the balance.
My gut was telling me Trump because of the odds but also because of FL. He won Dade, Hillsborough (Tampa), Pinellas (St Pete) and Duval (Jax) which still seems impossible to me.
raja_r
Nov 11 2024 at 3:02am
” its Schroedinger’s cat, its either dead or alive, we just haven’t opened the box[es] yet”
You might want to reread the thought experiment- the cat is both dead and alive until you look (superposition).
Mark Barbieri
Nov 8 2024 at 9:28am
I found that ElectionBettingOdds cratered during much of the evening. I switched to using Polymarkets. I also like the variety of bets there. For example, I can see that there is an 88% chance of 53 Republican seats in the Senate and a 12% chance of it being 54. In the house, you need 218 seats for a majority and there is an 80% chance that the Republicans will control between 220 and 224, but a 19% chance that it will be between 215 and 219.
I can also see that there is only a 3% chance that Democrats will control the house, implying that there is a 16% chance that Republicans will have 218 or 219 seats. No matter how you slice it, it is very doubtful that Republicans will have more than a 6-seat margin in the House, which is pretty tight and unlikely to survive during the 2026 election. In fact, the Republican control of the House will probably look a lot like the Democrat’s loose control in 2020.
steve
Nov 8 2024 at 10:23am
Or, you could just note that in every country which had an election in 2024 the incumbent lost vote share. This has been a long term trend but was uniform this year. I think the difference is stronger this year because of the lingering effects of inflation.
Steve
Comments are closed.