19 Comments
author

Must be reading the German Press….

https://x.com/wallstreetsilv/status/1847413058982822193

(Had to erase the tracking codes to that tweet, everything to the right of the “?”, and including it.)

Expand full comment
Oct 18Liked by Jeffrey Carter

Hi Jeff, check out the website Kalshi. I put money on Kamala to win. I’m hedging my happiness. If that bitch wins at least I make some money. If Trump wins it will be the best money I ever lost. Wojo

Expand full comment
Oct 18·edited Oct 18Liked by Jeffrey Carter

Trump's chances are good, I agree. While I am not working for any candidate this cycle, am noting that none of the social medial sites I was running have any restrictions this year. Part of having Elon on Trump's team and Mark Z on his heels, perhaps actively managing his business against the creeps who were censoring people the last two cycles.

Expand full comment
author

Good point, but still have the left wing media. Hunter's laptop turns the election last time and this one would be wide open. Probably DeSantis against Newsome or Whitmer.

Expand full comment

I read earlier today that poly market had Trump at 60% following last nights face off at fox.

I immediately thought about the difference between prediction markets and voter polls - specifically finding a way to price in potential fraud.

For polymarket to be at 60, one begins to wonder what Trumps real lead (before fraud) must be the majority swing states.

Expand full comment
author

The market is all 50 states. So, if Trump wins by the slimmest of margins, at the extreme 1 vote in each swing state, he still wins and 60% is correct! Theoretically past fraud is already priced in. Is their new fraud the market does not anticipate?

Expand full comment
Oct 18Liked by Jeffrey Carter

To participate in Polymarket, you can NOT be a "US Person" nor "located in the U.S.". So the market is everywhere BUT the US' 50 states. Terms per the log in for the account I just tried to set up (via VPN).

"By Trading, you agree to the Terms Of Use and attest you are not a U.S. Person and are not located in the U.S."

So it's everyone else, Except US Residents. Interesting dynamic. And ripe for cheating via VPN for those soo inclined.

Expand full comment
author

I don't think the study I linked to took into account computers, but think it probably was assumed since it was done in 2012

Expand full comment

Past fraud???

no such thing... just ask Kin zinger

Expand full comment
Oct 18Liked by Jeffrey Carter

While I believe (based on a fair amount of research for a class on polling and prediction markets I recently taught at a senior center) that prediction markets are more accurate than any public polling, in the short run there is still this. https://www.wsj.com/finance/betting-election-pro-trump-ad74aa71?st=VpHBPo&reflink=desktopwebshare_permalink

Prediction markets would do well to make size and timing of bets clearly public.

Expand full comment
Oct 18Liked by Jeffrey Carter

I’m very curious as to what the betting markets had Trump at in 2020 the day before the election.

Expand full comment
Oct 18Liked by Jeffrey Carter

Whale bets draw piranhas who want a piece of the other side. Always. It only bumps the vol of the bets. Would be a lot of fun if there were options on top of these mkts. Not 0DTE though, I’m too old for that carp.

Expand full comment
Oct 19Liked by Jeffrey Carter

A few tangents. When I was in grad school in the '90s, poli sci colleagues who studied American politics practically only did quant analysis, surveys and econometrics. It wasn't my field but surveys during races were generally zeroed in, matching exit polling and final results. (As I recall) But! The driver of predictions was almost entirely voter ID (pretty constant) with the 'likely voter' variable factored in. I was doing comparative work, and always had a hunch that party ID was a dummy variable for all the stuff they didn't understand, since I was studying places where that wasn't known, relevant, or stable.

When I was working at a giant data crunching place about 20 years later, reliance on surveys for real science (in the US) was already in the toilet. People don't answer their phones or send back forms, and they lie a lot more (respond strategically). Results are much more susceptible now to good/bad assumptions up-front and slick tricks with the numbers on the back-end to make up for deficiencies in the model and the gathered data.

All of that to say: Seems to me there are (at least) three different pools here that are different enough to be disconnected for analyzing their level of prowess in winning bets. Trading futures in commodities is highly selective, not something the average man in the street is going to bet a buck on, participants are highly motivated and moving (I'm guessing) lots of money, so they invest heavily in information. Voter surveys these days are reaching people who want to be reached, want to respond, and among that pool likely voters will be prone to strategic lying since 1. you can tell whose poll you're answering, and 2. polls are merely grist for the propaganda press, not even the campaigns use them. Voter ID is a less powerful predictor these days, and many races have more than two options. Even sure-thing voters have only weak incentives to invest in marginally better information on each race, and just about no incentive to offer their honest opinions.

I suppose none of us know truly who's in the election markets but I'd say the demographic has to be closer to a relatively highly motivated, higher information voter, in terms of how invested or solid their 'bets' are, than a pit trader. But are they betting their preferences or betting to make money? How representative of the voting pool are they (if most voters are low-information)? All of these pools (really four, counting actual voters rather than survey respondents) are self-selected, and all may choose strategically instead of honestly revealing real preferences.

As for betting markets, this one is legal for now but may get yanked at any time: "Predictit.org is an experimental project operated for academic purposes under permission from the CFTC."

Expand full comment
author

do you think the people in the betting markets might not be motivated voters, but actual speculators just trying to make a buck?

Expand full comment

Yes, at higher levels. I'd guess that on the whole, most (by numbers of people) are likely to be dabbling in both (voting & betting) without too much skin in either game. I can't see the incentive for low-propensity voters. I doubt someone not interested in politics, not following at least some races, is going to think one day, Hey maybe I'll strike it rich by betting on something to which I pay no attention! That's more my point above. People in the low-to-middling betting ranges likely think they know enough to make a buck because they're paying some attention, but not to the point of doing research like a real trader. Whether or not they're right is a different question.

If looking at participation by number of dollars, I'd guess more of the big money comes from speculators than primarily motivated voters. Is it naive to think rich motivated voters are more likely to 'speculate' by donating to the candidates of their choice, hoping for a payoff by that route?

I don't really understand how these markets would work as a preference revealing mechanism. It's too specialized, seems to me. Not like millions of barrels of oil or millions of shares across a whole sector. I haven't spent that much time thinking about it, I'll admit.

Expand full comment
Oct 18Liked by Jeffrey Carter

Internal Democratic polling shows 60 trump 40 Harris that’s why they are panicking

That’s polling sample if 100,000

That substantiated the betters

Expand full comment
author

If you are correct about Democratic internal polling, it's showing up in the prediction markets as exactly that.

Expand full comment
author

where are you seeing that exact data?

Expand full comment