Presidential predictions at Predictit -- plausible?

According to Predictit there’s a 27% chance Trump will be gone before January 2020, but a 66% chance that he’ll be the 2020 nominee. If we take those numbers as correct, this leaves only 7% as the total chance for any of

  • Trump being ousted (or too sick to run) during 2020
  • Trump voluntarily retiring after 1 term
  • Trump losing the primaries to, e.g. Kasich or Romney.
    (Of course this assumes that out-before-2020 and 2020 nominee are mutually exclusive. I suppose his Cabinet could do a 25th Amendment on him and then his loving base vote him back in anyway.)

Predictit also shows chances for the Democrats. When I checked a few hours ago, the market prices were as follows. Read this as Trump 66% to be nominee, 38% to be the November winner. I divided 38 by 66 to show 58% as the chance that Trump will win if he runs. I used (100 - No) rather than Yes to get the percentages. Numbers shown in parentheses are numbers I derived. (The “(5 Other 3)” was added by me to get that column to total to 100.

66 Trump 38 (58%)
8 Pence 3 (38%)
7 Haley 3 (43%)
5 Kasich 2 (40%)
4 Romney 0
2 Cotton 0
2 Cruz 0
2 Ryan 1
1 Sasse 0
1 Paul 0
1 Rubio 0
1 Cuban 1
100 Total (48) --> 42

17 Harris 11 (65%)
14 Biden 10 (71%)
13 Sanders 10 (71%)
11 Klobuchar 6 (55%)
10 Warren 7 (70%)
9 Booker 5 (56%)
7 Gillibrand 4 (57%)
4 Winfrey 0
3 Cuomo 1 (33%)
1 *Zuckerberg 1
1 Delaney 0
1 Kaine 0
1 BrownJ 0
1 Murphy 0
1 JohnsonD 0
(5 Other 3)
100 Total (58) --> 58

No Clinton ticket is shown for nominee despite that she is 19% to run. There are Zuckerberg tickets in both the D and R sections. There’s a ticket for Jerry Brown, but none for Sherrod Brown.

If Predictit numbers are to be believed, it is the “oldsters” – Biden, Sanders, and Warren – who are most likely to win if they’re the nominee.

Any comments on the Predictit predictions? (Should we start a Karachi auction in Thread Games? :slight_smile: )

I’ve been wondering what could happen in the next few months to make a Republican nomination campaign against Trump viable (assuming he runs). The last credible such campaign I can remember was Ted Kennedy against Carter, and that didn’t get very far.

Haley, Kasich and Romney are all far more centrist than Trump. Maybe these numbers indicate the potential of a moderate revolt from within.

No comment on the specific predictions, just a general comment that it’s very common for PredicIt odds to be contradictory. They are based on what people are currently betting in each individual market, and different markets can give different odds for the same event. Sometimes there are arbitrage opportunities.

For example, for long periods there were better odds that the Democrats would take control of the House than that they would take over 217 seats, even though that was exactly the same bet.

Yes; I forgot to mention this in OP. What are good bets at Predictit now? Nobody’s going to get rich — they have a $850 maximum per player per contract — but can we identify some smart bets?

Trump has overwhelming support from the Republican base, nearly 90% last i checked. They are not going to turn on him, they are entirely conditioned to dismiss anything Mueller finds out of hand. Trump has the GOP by the balls, the nomination is his if he wants it. Being president is also the best protection he has from legal trouble so he is not going to quit voluntarily.

On this day after G H W Bush’s death it is worth remembering that he was not only popular among Republicans midway through his term but that he was hitting 80% approval overall in early '91. He dropped fast as the economy softened. I am hoping our economy stays strong but cycles are cycles and if a downturn hits? The party and public wouldn’t have turned on Nixon if the economy had been strong. I would be shocked if there were no Republicans who were preparing, quietly, to step up to challenge, possibly with success, in that possibility.

I thought we were well past the point of believing that the prediction markets have any special insight or knowledge. They are 98% reinterpreting available poll data.

Cite? Is this based on Trump-win at Predictit being only 36% on 2 November 2016? “Had the predictions been sound they would have shown Trump 100%”?

And yes, prediction markets interpret available data. What’s the alternative? Studying sheep’s entrails?

Trump will be the nominee in 2020 and will run.

He may be impeached on something (but not removed), but it will be lightweight and carry no substantive support, particularly in a GOP-dominated Senate.

That’s my prediction.

You may as well read entrails for all the value PredictIt gives you. They’re a trailing indicator, not a leading one, and mostly exist as a self cannibalizing way to give people the illusion of predictability.

I think the odds for the major candidates are generally reasonable educated guesstimates, which is all that can be reasonably expected at this point. I think it overrates the longshots; I’m not going to put any money on Oprah, even at 25-1, or on the Rock at 100-1.

I’m not sure about the wide disparity between the chances of the various Dems winning if nominated; at this point, I would say the safest assumption is that all would fare about equally well.

I can see giving Biden and Sanders a bit of an edge over the others, since they already have near-total name recognition and have hypothetical polls showing them beating Trump by wide margins. I don’t take those polls very seriously at this point, as I think they’re about 90% just a proxy for name recognition, and there is plenty of time for a now-obscure candidate to catch up to them in that regard. Still, it is a piece of evidence in their favor that the others can’t currently match.

But I can’t see any earthly reason to put Elizabeth Warren in a group with them rather than with Klobuchar/Gillebrand/Booker.

Another thing I want to say: After watching PredictIt over the last few election cycles, I have come to the conclusion that it’s a myth that the aggregate of bettors is somehow smarter than the conventional wisdom. Rather, it simply reflects it.

58% sounds far too high for Trump’s reelection chances of winning if he runs. I would guesstimate it at…5.8%. (My IMHO, based off of gut feeling only, no data or analysis to offer at all)

Shouldn’t you be citing that they DO have predictive power? Do you think all these people are betting based on anything meaningful other than polling data? From what I’ve read, they are not and useful mostly in races that have limited polling.

The OP asked whether the odds posted on Predictit were “plausible” (which, IMO, they are), not whether they were superior to all other possible methods of predicting elections (which they aren’t).