Polls vs betting markets
I had an email exchange a couple days ago with Josh Marshall over at Talking Points Memo about polls (which he’s written a lot about recently) and the election betting market (which he had never mentioned). Yesterday, he used our exchange as a jumping off point to explain why he doesn’t believe the betting market is reliable and certainly no improvement over polling. The money grafs:
“First of all, as I said, bets are largely made on the basis of polls. But let’s go a bit beyond that. In theory at least in equity markets you have armies of industry analysts studying industries and providing insights into the future challenges and profitability of businesses. Same in commodities, currencies, bonds, etc. Investors make investments on the basis of this and other kinds of information. To the best of my knowledge there’s really nothing like this informing political betting markets. Again, it’s mainly polls and the “analysts” who you see in the media. If we’re talking about calling most races right within ten points of the result keeping informed on those fronts is a good idea. But those aren’t the kind of races we’re talking about. If you want to place your money on the basis of political commentators’s predictions about the future, on the basis of the roundtable on the Sunday shows … well, good luck to you.
“Then there’s another point. Efficient markets need broad and diverse participation. There’s very little evidence these markets have that. Just to note one example, one of the most prominent of these ‘markets’ this year has been Peter Thiel’s Polymarket. But Polymarket only takes bets in crypto and can’t accept bets within the United States. So we’re talking the insights about US election outcomes of foreigners or expats trading in crypto. I am, shall we say, skeptical of the decisions and prognostications of that group.
“The final point is that betting markets, in my experience, tend to disproportionately be made up people who like to play the market. Which means they tend to be fairly right-leaning.”
Read the whole thing at the link.
polls vs political betting markets
Never-the-less, there are those using that model, good bad or indifferent
I’ve seen it mentioned two or three places over the past couple of weeks more in the sense of “mentioned in passing”, as in “Hey that bozo is playing the Vegas odds” and I’m not sure I’d dismiss it. At least any more than I dismiss any other polling method
Still just “best” guesses …
Nearly all polls are entertainment products and should be evaluated on how entertaining that have been. That’s not easy to do, but a proxy could be the amount of paid distribution polls get. The reason they are entertainment is that the fundamental question asked can never be confirmed: if an election were held today, who would you vote for? Even simultaneous polls on the same race whose results are so far apart that their reported margins of errors don’t overlap can never said to be right or wrong, because there is not an election today until there really is one. So if both entertain an audience, both served their purpose. Betting markets entertain gamblers.
@Eric,
LOL! I can see you know nothing about polling.
Polls distributed publicly are fundamentally for entertainment. News distributors conduct or buy them because they believe their audiences will find the material interesting. There are private polls that are likely used to help campaigns allocate resources and adjust messaging. We don’t see those. The fundamental question of pre-election polls just can never be checked by results. Again, polls conducted within the voting period are some kind of hybrid with some respondents answering an intention query and others answering what they have already done. So I am happy to restate that the biggest purpose of publicly available pre-election polling is entertainment. Also polls and betting markets do not deal with the same question really. Polls are who do you intend to vote for and bets are on who you believe will win.
@Eric,
So repeating over and over that polls are just for entertainment is your argument?
LOL!
Nobody said polls and betting deal with the same question. What Josh said is that they’re not independent and that bettors are using polls to place their bets.
Yes, that’s my argument. Election polls – excluding exit polls (which have been very complicated by increased early voting) – are not conducted to predict anything. You are free to use them as if they were, but they weren’t. If you choose to do that, it would be more rigorous to not use anything other than the final poll any organization conducts prior to the event and its actual result. I’d add that for this Presidential election it seems all the polls have states with enough EVs whose numbers are within the published margin of error that they are not even predicting a winner. Harris wins: We called that! Trump wins: We called that! Betting markets can have the same problem based on the bet structure. Imagine that in state “X” the bet is Trump by 53,000. You bet Harris. Trump wins by 48,000. Trump wins that election but you win your Harris bet.
@Eric,
As I posted earlier, you don’t know anything about polling. Thanks for proving my point.
Ten Bears and Joel:
People do stupid things. They will bet on a shiny object just because it looks good. There is no clear foundation (some times, many times, all the time) which can support their choices.
Oh. but this has to be the result of this poll or that poll. It appears, at least to me; polls shift the responsibility of making a decision for yourself.
Do I follow this crowd, or should I follow that crowd. I do not like his bigotry, yes but the other guy is old and is feeble minded. Most could not read, much less understand what they mean, the results of an economy or why it happened the way it did, etc. Much of the political appeal is based on commentary, appearance, a person’s past; rather than reality.
We have a nut job versus the old man specter. A liar and fraud versus reality. You can not change what you look like or how you acted but you can tell a different version of the past. Many people will not look at what took place and how they acted. They like the shiny object and will get angry when they have to give it up even though the alternative is better.
They believe what you want to believe and disregard the rest.
Lots of money riding on those polls based on beliefs. So, what is a little lie or slightly obscuring reality to up the ante?
Maybe it’s not Vegas that’s being played, aeh? I’m firmly convinced that if we’re not living in one of Ludlum’s better novels it’s some crankpot’s interpretation of the NAZI science of mass manipulation because sometime somewhere along the line we forgot the lessons of the Cold War and nineteen thirties’ Germany. Not sure that “herded like cats” is such a joke …
Polls have value but also have bias. Even with the best statistics there is a margin of error. There are good and bad poll takers. Doing an analysis of predictions and measuring results lets us know the bad actors are and they screw to democrats. Only one poll matters. The rest are what you make it. Entertainment or a tool to sell an idea or advertising. If you are in the propaganda game or sales go for it. Betting is expressing value with money. Yes I believe bets or free.
@Strong,
As a scientist, I’m in the prophecy business. Science is all about predicting the behavior of things, not for entertainment.
The serious polling outfits have a product to sell–prophecy. To the extent that their prophecies prove accurate, they’ll be hired next cycle. To the extent they’re not, they’ll go out of business.
I’ve been following polls for 40 years. I don’t follow individual polls, I follow meta-analysis of aggregate polls. In 2016, Nate Silver and Sam Wang were pretty accurate. In 2020, they somewhat underestimated Trump’s support, but he still lost by a wide margin.
Nobody knows who is betting at the betting sites. A couple of right-wing billionaires, (Musk, Putin) could dump a couple hundred million to shift the betting markets significantly. A fool and his money, etc.
Joel, I don’t know who does the censorship on angrybearblog but it is quite out of hand. Anything that is not pro-Harris will not get posted in the comments. Censorship is a big issue for Americans in this election and this blog plays right into that.
So let me just throw you a softball question in order to get this comment approved (low chance).
Putin has said that he wishes Harris to win the 2024 election. So why would he be spending billions in pushing betting markets in Trump’s favor as you say? Putin wants to keep doing whatever he wants to in Eastern Europe, and his best chance at that is with Biden and Harris leadership.
@Giles,
LOL! We don’t have a problem with people who are not pro-Harris. We do have a problem with right-wing trolls, who are quite out of hand. You are not entitled to comment here, and the 1st amendment doesn’t protect your right to post on blog threads. Get over yourself.
If you believe that Putin wants Harris to win, and that his best chance to conquer Ukraine is with Harris/Walz, you haven’t been paying attention. That’s nonsense. We don’t support nonsense here at AB.
Let me just throw you a softball question in order to get an honest response (low chance). Why do you believe repeating right-wing propaganda deserves an audience here at Angry Bear?
I found that comic here when I was looking up who actually said it. The author of the page finds it in print prior to Twain or Disraeli. He includes an anonymous judge who referred to “a simple liar, a damned liar, and an expert witness” wherein “the point lies in the fact that expert witnesses are allowed to give evidence as to what is their opinion, and hence are out of the reach of an indictment for perjury.”
Even though people been taught to distrust statistics, if done by someone who knows statistics and wants to learn something, polls can be useful.
The flip side is that even someone trained in statistics, (I learned mine doing process control for ink jet manufacturing), may not be able to discern whether any particular poll is being done properly.
@Arne,
Over time, biases become obvious (e.g., Rasmussen skews to the right). Any responsible pollster discloses their methods, so that strategies that might lead to bias (e.g., overuse of land lines) can be discerned.
All this time I thought it was Sam Clemens …