The conventional wisdom and the betting markets

It seems like the conventional wisdom is that the GOP took the House by a narrow margin and the Senate is still up in the air. That’s what I hear in some media reports. But that’s not exactly the current view of the betting markets:

Can that be right? Are the Dems actually more likely to take the Senate than the GOP is to take the House? The 91.5% figure seems to be based on the fact that the Dems are strongly favored on all three outstanding Senate races (especially Nevada and Arizona), and they only need two of them. I’m not sure why the House GOP odds are not higher, perhaps because the Dems currently lead in most of the unresolved races. To be sure, an 83% chance of taking the House is still excellent, but I see absolutely no speculation on what would happen if the Dems ended up with 51 seats (not unlikely), and somehow also took the House. Senator Manchin would become much less influential.

Again, this is not likely, but I see at least a mild disconnect between pundit talk and actual election odds.



38 Responses to “The conventional wisdom and the betting markets”

  1. Gravatar of foosion foosion
    10. November 2022 at 11:09

    Pundit talk is probably the least reliable indicator.

    Did you ever read Expert Political Judgment by Tetlock?

  2. Gravatar of anon/portly anon/portly
    10. November 2022 at 11:22

    Are the D’s strongly favored in Nevada? I had guessed they were in Arizona, where the D lead is large, but Nevada depends on the remaining mail-ins breaking strongly D, I think. I guess I could find an analysis of this somewhere.

    My thought would have been that Georgia might be thought of as a lock, that Walker has no chance of winning without Abrams on the ballot also.

  3. Gravatar of ssumner ssumner
    10. November 2022 at 12:46

    Foosion, No.

    Anon, Yeah the Dems are strongly favored in Nevada (80%), as mail-in ballots tend to be heavily Democratic in that state. But it’s not locked up.

  4. Gravatar of George George
    10. November 2022 at 13:07

    Define ‘psyop’.

    Trump endorsements had a 93% success rate, GOP retakes House, is about to retake the Senate, assuming remaining counts are legal, all this DESPITE the Democrats cheating, and yet the enemy of the people propaganda arm of the D party fake news mafia still wants everyone to believe the narrative ‘Trump lost’ or ‘Trump lost, DeSantis won’ (despite Trump getting 1.1 million more votes in Florida during Presidential election than DeSantis got for his re-election).

    Blame him for what, firing Nancy Pelosi? On the verge of retiring Schumer? In an off cycle year we are taking the house majority and likely the senate, all the while stopping Biden’s disastrous agenda. Now think how big the actual losses would’ve been had Trump stayed in his basement like Biden.

    The funny thing about the legacy media is that they believe everyone else is as stupid as they are, LOL.

    For AZ, there are 384,000+ mail-in ballots that have yet to be counted, where people hand delivered them due to distrust in drop boxes and the USPS. Kari Lake estimates she will get 60%-80% of those votes. NO WONDER the ballot counting went stagnant this morning. They are stalling because they know the ballots coming in will mostly be for Lake.


    Out of Brazil, what we have now is an official press release from Defense ministry clarifying that, contrary to what the press and the Electoral Court have said, the military vote audit did NOT exclude that fraud may have happened due to the many grave problems with the system.

  5. Gravatar of George George
    10. November 2022 at 14:11

    Florida banned mass mail-in ballots, banned ballot harvesting, requires voter ID, and DeSantis created an election police force.

    Florida also just so happened to have had historic win margins across the state for Republicans, while Democrats somehow won close races elsewhere across the country during a horrific economy.

    And Florida has three times more people than AZ, yet Florida counted everything and was done election night and AZ is still using their fingers and toes and corrupt machinery.

    Do the math folks, it’s obvious what’s going on.

  6. Gravatar of foosion foosion
    10. November 2022 at 14:49

    Scott, see

    Generalists with a good feel for probabilities tend to be better at forecasting than confident narrow experts. The author later organized a team that won a US intelligence agency forecasting contest (which is the subject of his book Superforecasting).

  7. Gravatar of Michael Sandifer Michael Sandifer
    10. November 2022 at 15:09

    The pundits I listen to, who have actually led successful national election campaigns in the past, say that most of the political commentators in the popular media don’t actually study the numbers closely. They’re casual observers, even when experts/former insiders. They advise not to take such comentators seriously, just as many polls that appear in the media are not to be trusted much.

    Pundits I listen to include David Axelrod, James Carville, Mike Murphy, and Robert Gibbs.

  8. Gravatar of JMCSF JMCSF
    10. November 2022 at 15:44

    George, you’re right! It’s a scam, don’t vote!!

  9. Gravatar of Student Student
    10. November 2022 at 18:03

    Scott there is a lot of wiggle room in the betting market odds. I got the numbers for AZ, NV, and GA at 8:35 from predict it.

    They were

    A = P(AZ dem) = .94
    B = P(NV dem) = .82
    C = P(GA dem) = .71

    That means a there 4 possible combinations of dems getting at least 2 out of 3

    P(A ∩ B) = 0.7708
    P(A ∩ C) = 0.6674
    P(B ∩ C) = 0.5822
    P(A ∩ B ∩ C) = 0.5472 (dems 51 seats).

    Probability of democrat control senate should be:

    P(A ∪ B ∪ C) = P(A) + P(B) + P(C) − P(A ∩ B) − P(A ∩ C) − P(B ∩ C) + P(A ∩ B ∩ C).

    2.47 – 0.7708 – 0.6674 – 0.5822 + 0.5472 = 0.9968

    But the market question is giving dems an 89%. Thats a lot wiggle room. Republican control of the house is at 87%. They are basically equal with that much wiggle.

    BTW, the probability of 51 dem seats should be 0.5472, better than 50/50.

  10. Gravatar of steve steve
    10. November 2022 at 18:11

    Being markets, like many other markets, are subject to emotion driven swings in the short term. In the long term they are probably pretty good but there is that old chestnut about the market staying irrational longer than you can stay solvent.


  11. Gravatar of Sara Sara
    10. November 2022 at 19:26

    One commenter says the “election lead is large.” (in arizona)

    Based on what? The ticker on the bottom of Fox? CNN? MSNBC?

    I feel like most of you are amatuers.

    Folks, Arizona has 600,000 outstanding ballots, 300,000 that were delivered on election day by people concerned with fraud. In other words, people who lean republican.

    Kari Lake will win easy. Blake Masters will close the distance, and may win if those ballots are 60% or greater. I think at the moment he needs 57% to catch up. It will be close.

    The Georgia race is up for grabs; it just depends how the libertarian votes split; I presume Walker will get more, but he would need at least 60/40, assuming everyone votes the same on December 5th.

    Nevada will be very tight. 50/50 at this point.

    I will even help Sumner make a prediction; he’ll like this one, because he’s viciously anti-trump.

    I am almost certain that Trump will now destroy the entire MAGA movement — which actually started as the Tea Party movement, and which is really a classical liberal movement, predicated upon fiscal and individual responsibility — and he will destroy the party because he’s a very competitive man who will refuse to take a back seat to DeSantis. The infighting between those two will create a schism that might help the corrupt establishment class stay in power.

    Good news for Sumner and the apparatchiks. Bad news for those who value freedom.

    Maybe we ought to convince some of those radical democrats and MAGA supporters to unite, so we can form a Chomsky style anarcho syndicalism party — a fair compromise between the two who really seek many of the same things — because even that would be better for humanity than the global undemocratic institutions Sumner envisions.

  12. Gravatar of ssumner ssumner
    10. November 2022 at 19:40

    Student, You said:

    “2.47 – 0.7708 – 0.6674 – 0.5822 + 0.5472 = 0.9968”

    You made a mistake somewhere, but I’m too lazy to find it. Maybe someone else will help me out.

  13. Gravatar of Andrew C Andrew C
    10. November 2022 at 20:51

    Over the last 12 hours it’s gotten clearer that Nevada will go Blue. The drops coming in are favorable enough for CCM to win. The real question is the Georgia runoff in December, which my guess is Warnock wins. What’s gonna be interesting is a house with a <6 seat Republican majority and a 51 democratic seat senate, and what the democrats can pull off with a few Republican representatives going overboard. I can totally imagine a bill that protects abortion rights up to 15 weeks getting passed in the next two years.

  14. Gravatar of Babak Babak
    10. November 2022 at 22:58

    Scott, I didn’t double check the probabilities, but the algebraic sum is correct.

    However, as someone who’s dabbled in predicit, I would caution against taking it too seriously. It is very far from a liquid market. The maximum bet is low and the transaction costs are prohibitive. I once did the math, and I needed 15-20% gain on my bet to break even if I were to deposit $850 (the maximum at the time), buy a position, sell it and cash out (the biggest fee is on cashing out).

  15. Gravatar of George George
    11. November 2022 at 04:30

    JMCSF, you said:

    “George, you’re right! It’s a scam, don’t vote!”

    Yes, I’m right about voter fraud, but in no way do I agree with the fake patriots, fake maga, RINO sources trying to convince people the best way to deal with it is to not vote and allow 100% of all votes to be fake.

    It is people voting that helps reveal the scams, so the best thing to do is vote! And that’s what the country did. Red Wave happened.

    Everyone, check out what Trump retruthed this morning:

  16. Gravatar of Student Student
    11. November 2022 at 04:37

    Oh man I am stupid lol. Don’t do election probability when you are playing poker. There are 4 possibilities (including the sweep). I computed it as if there were 3. Feel free to delete that idiocy lol.

  17. Gravatar of Student Student
    11. November 2022 at 05:14

    Redoing using the correct union this time lol.

    A = P(AZ dem) = .94
    B = P(NV dem) = .82
    C = P(GA dem) = .71

    4 possible combinations of at least 2 out of 3. Relabeled for the 4 cases.

    1 = P(A ∩ B) = 0.7708
    2 = P(A ∩ C) = 0.6674
    3 = P(B ∩ C) = 0.5822
    4 = P(A ∩ B ∩ C) = 0.5472 (dems 51 seats).

    Probability of democrat control senate should be:

    P (1 U 2 U 3 U 4) = P(1) + P(2) + P(3) + P(4) – P(1 ∩ 2) – P(1 ∩ 3)- P(1 ∩ 4)- P(2 ∩ 3) – P(2 ∩ 4) – P(3 ∩ 4) + P(1 ∩ 2 ∩ 3) + P(1 ∩ 2 ∩ 4) + P(1 ∩ 3 ∩ 4) + P(2 ∩ 3 ∩ 4) – P(1 ∩ 2 ∩ 3 ∩ 4)

    2.5676 – 0.5144 – 0.4488 – 0.4218 – 0.3886 – 0.3652 – 0.3186 + 0.2995 + 0.2815 + 0.2456 + 0.2126 – 0.1639 = 0.9855

    The original point still stands tho. The total probability doesn’t match what it should be based on the individual race probabilities.

    That’s such a weird result, maybe I did something else wrong such as converting the predictit prices to odds

  18. Gravatar of Ricardo Ricardo
    11. November 2022 at 05:33

    Why would anyone want Manchin to have less influence.

    He’s the only one that is moderate.

    Is it not becoming more clear that Sumner wants a one party state?

    1. He advocates “packing the court” as a way to push through legislation that was unconstitutional. I.e., any legislation that would siphon more power from the community and states where people live.

    2. He has difficult time understanding the nuassance of the Supreme Courts decision. Like a simpleton, he boils everything down to buzz words like “pro-choice” or “pro-life” without asking WHO has the right to determine when the right to life begins. These are the profound questions that are dealt with by the courts. I.e., states, federal government, communities, etc. In other words, you could be pro-choice and still believe that the state legislators and people living within those states have a right to determine their abortion laws.

    3. He believes Pelosi, McConnell, Schumer, Cheney and Biden are gold standard politicians; I wonder how many economists would agree? Not many!

    He wants NATO to be global, which would require the homogenization of culture and force to crush any opposition, including the Muslims, Christians, South Americans, Africans, etc; it’s a new form of “white man savior” disease, which is predominant of the left.

    4. He said in 2020 that there is “no such thing as voter fraud.” So don’t ask questions about voting machines that can calculate fractions of a vote, and don’t mention the Carter Center, because they are stupid for advocating for “voter ID” and “No mail-in ballots” because apparently the Carter Center has no experience in these matters, despite covering elections worldwide. Jimmy must be one of those cultists; or maybe he’s just a really smart, really educated, moderate left wing guy who understands the obvious. If you vote, you need an ID. You need to actually prove you’re a resident of the state.

    5. Sumner and his radical policies will lead to WW3. And as you all know, this is the guy who asked me to go to war for him in Ukraine. He told me to “…do more” because we “need to do more…” to stop Putin, so when we go to War Sumner expects to be the guy pulling the strings and bankrolling the profit. Of course, when you ask him to grab his helmet and buy a gun he finally shuts his mouth and becomes a pacifist again.

    I’m afraid that until we can finally remove these corrupt politicians, we can expect more outbursts from the old man Sumner war hawk and his establishment thugs.

  19. Gravatar of veritas veritas
    11. November 2022 at 06:41


    Perhaps the greatest mental barrier for those who devoted their adult lives unconsciously sanctioning/apologizing for/intellectually assisting a criminal conspiracy of taxation without representation, i.e. “inflation”, is to admit that they are not merely wrong about their ‘models’, their ‘variables’, their ‘equations’, but about the TRUTH of the INTENTIONS of those controlling central banks worldwide.

    Why? Because it would be to admit that they were INVOLVED as promoting people who are in fact evil.

    Who are the Silent Thieves?
    Why are they manipulating you?
    How are they stealing your wealth?
    What is inflation?
    Monetary manipulation.
    Taxation without representation.

  20. Gravatar of John S John S
    11. November 2022 at 08:36


    Is there still good online action? I figured the games (at least NLHE) were dead due to bots and the proliferation of training sites.

  21. Gravatar of Student Student
    11. November 2022 at 08:45

    It looks to me there is ~10 pts of error in the prediction markets. I don’t know which way it goes or why. Prolly random. I don’t know if there are bots or large single bets on something (sports betting market odds move in two ways, seemingly for this reason). All i know is I expected the above to match the senate odds for dems more closely. Maybe I made a mistake. The work is shown, so someone could point it out.

  22. Gravatar of John S John S
    11. November 2022 at 08:51


    Side notes: the ranked-choice instant runoff in Alaska has Murkowski (91,200 votes) in good position to win vs. Tshibaka (94,100). (The 4rd place finisher, Buzz Kelley, endorses Tshibaka, but he only got 2.9% of the votes, so most of the 3rd place Dem votes should go to Murkowski.) If Murkowski wins (96% on PredictIt), the Trump faction definitely won’t control the Senate.

    Non-partisan open primaries + Final Five ranked choice passed in Nevada (though it needs to be reapproved in 2024 to take effect).

    Also, a ranked-choice instant runoff would save the cost/time of having a runoff election in GA (Warnock vs. Walker).

  23. Gravatar of Student Student
    11. November 2022 at 08:51

    John, you mean poker lol. Training sites require training. Not everyone does that. Bots are banned on good stuff. But I was playing with friends last night.

  24. Gravatar of John S John S
    11. November 2022 at 09:09


    Oh ok, was just wondering. I still keep up a bit (Galfold has put out some great Youtube vids on using solver results for decision making), but my impression is that grinding out a decent wr above micros is pretty tough with all the solver-using nits out there.

    Thought about getting into live PLO, but this whole Garrett-Robbi fiasco gives me a sick feeling (even a few players card-signalling the dead aces could leave me vulnerable on 3-flush boards). I’ll prob just screw around with low stakes NL live now and then.

  25. Gravatar of Student Student
    11. November 2022 at 09:20

    Def just a hobby these days unless you get lucky/and or sponsored. But it’s a fun game.

  26. Gravatar of ssumner ssumner
    11. November 2022 at 09:59

    John, The Trump faction won’t control the Senate regardless, as there are a number of other anti-Trump Republicans (Romney, Collins, etc.)

  27. Gravatar of anon/portly anon/portly
    11. November 2022 at 10:43

    Student: “The original point still stands tho. The total probability doesn’t match what it should be based on the individual race probabilities.”

    That’s based on the probabilities being uncorrelated, of course. But anyway you’re obviously doing something wrong.

    You should immediately see that .9855 (or .9968) is wrong since .18 * .29 is roughly .2 * .3 = .06 so right away the R’s have a roughly 6% chance of winning both Nevada and Georgia.

    A = P(AZ dem) = .94
    B = P(NV dem) = .82
    C = P(GA dem) = .71

    P(R’s win A only) = .06 * .82 * .71 = .0349
    P(R’s win B only) = .94 * .18 * .71 = .1201
    P(R’s win C only) = .94 * .82 * .29 = .2235
    P(R’s win none) = .94 * .82 * .71 = .5473

    Total probability of D’s winning 2 or 3:

    = .0349 + .1201 + .2235 + .5473 = .9258

    When you calculated P(A U B U C), that’s the probability of the Democrats winning at least one of the three, which is also equal to the complement of the R’s winning all three.

    .06 * .18 * .29 = .0031

    1 – .0031 = .9969

    So that first one, .9968, was correct, either because you did the calculations correctly or a lucky accident. It wasn’t what you wanted to calculate, though.

    I hope this is reasonably clear.

  28. Gravatar of John S John S
    11. November 2022 at 11:10


    Good point. But getting Murkowski in over Tshibaka is still preferable, right? And if she does win, it will be entirely due to ranked choice voting, as she is still behind in 1st place votes.

  29. Gravatar of ssumner ssumner
    11. November 2022 at 11:25

    anon, Yeah, I was too lazy to look into his “correction”, but as you say it also seems wrong.

    John, Good point. To be clear, I favor RCV.

  30. Gravatar of veritas veritas
    11. November 2022 at 12:05

    The ultimate litmus test for knowing who is and who is not truly ‘pro free market’, is not about roads or hospitals or t-shirts or cars.

    It is information.

    You look to see how yourself and other people react to the idea, and now a real world becoming, of a free market in information itself.

    Musk tweets:

    What is your immediate gut feeling reaction to the notion of ‘citizen journalists’? Or, translating it, a ‘free market in information collection, analyzing and dissemination’?

    Is it ‘oh no’ in any way? If it is, that is an ANTI-free market logic within you being implemented as your conceptions and your speech.

    Is there any notion that is itself an introduction of a smear of humanity, that a priori ‘perceives’ humanity as such as characteristically unfit or incapable of researching, collecting, analyzing, and disseminating information in an open and free environment?

    That there ‘must’ instead be ‘philosopher kings’ who must smother and dominate the information bandwidth ‘for the sake of humanity’ that would otherwise destroy itself from ‘disinformation’?

    Who would ‘fact check’ the ‘fact checkers’?

    Check out this Musk tweet.

    “Game Over”….for who? IMO, for the ‘philosopher kings’ game of information domination. It has been conclusively empirically falsified. It turns out that when 90% of the information is controlled by just 6 corporations, 6 CEOs, that not only mass corruption ensues, but the exact thing that was pushed as a reason for having ‘philosopher kings’, to eliminate threat to humanity, WAS ITSELF the greatest threat to humanity in human history.

    To divide humanity into two, into those with the ‘keys to the kingdom’ and those without, has resulted in the spread of literal satan worshipers who rape and murder children in a genocidal ritual of evil.

    How many ‘philosopher kings’ here knew the truth that 800,000 children go missing EVERY YEAR in this country? What happens to all those children?

    The truth isn’t going to be for everyone, but many of us have been preparing ourselves for when the horrible truths become ‘public’ as the narrative itself is freed from the very perpetrators.

    Question: If Democrats won AZ and NV, it already would have been announced. But it hasn’t, because they didn’t. The corrupt state election ‘officials’ are slow walking because they know they lost.

    How can Florida count 8.5 million ballots and have an answer same day, but AZ and NV with a fraction of the population of FL, keep delaying and need days and days? Answer is obvious, they’re trying to reduce the impact of the red wave by splitting up the House and Senate victories, with RINOs and legacy media try to divide the GOP by pushing fake stories of ‘Trump failed’ and ‘Trump is a liability’, and they’re trying to give themselves more time to find a way to cheat. It’s obvious. All the excuses are to cover up what they’re really trying to do.

    Those of you in this thread conducting probability analyses, FYI, you are using faulty source data. Garbage in, garbage out.

    Lake won, Masters won, GOP won at least Senate 51 seats, more if cheating was reduced to zero. Did anyone see the report, and the election official’s excuse, for why the Washoe County NV cameras on the ballot counting ‘suddenly went dark’…for 8 hours throughout the night? Nothing to see here, don’t worry nobody went in or out of that building with any fake ballots. Trust them, there is no cheating ever in elections.

  31. Gravatar of Steve Steve
    11. November 2022 at 14:32

    Doesn’t House control depend on 14 toss-ups in California, where the ballots won’t be counted until Christmas?

    Entirely possibly that the GOP ends up with a 1 seat majority, and that majority control requires consent of both Valadao and Boebert.

    California, discrediting democracy around the globe…

  32. Gravatar of ssumner ssumner
    11. November 2022 at 16:08

    Veritas, Love to see you conspiracy nuts squirming.

    Steve, Don’t forget Greene.

    California: The best place to live, and also the worst governed state.

  33. Gravatar of Student Student
    12. November 2022 at 06:03

    Anon, no it’s the second one. The first one is wrong because you are computing p(1 or 2 or 3 or 4). The sweep case is it’s own possibility so it because a text book case of a union of 4 cases.

  34. Gravatar of Student Student
    12. November 2022 at 06:07

    Notice – P(1 ∩ 2 ∩ 3 ∩ 4) is subtracted out because it’s already included because 1 n 2 n and 3 is case 4.

  35. Gravatar of Student Student
    12. November 2022 at 06:14

    “You should immediately see that .9855 (or .9968) is wrong since .18 * .29 is roughly .2 * .3 = .06 so right away the R’s have a roughly 6% chance of winning both Nevada and Georgia.”

    Yes and you also notice .9968 and .9855 does not equal the market for dems control senate which was 89% at the time.

    This is my entire point. The betting markets are not unbiased because because these things should all match if it were.

  36. Gravatar of Student Student
    12. November 2022 at 06:27

    P(R’s win A only) = .06 * .82 * .71 = .0349
    P(R’s win B only) = .94 * .18 * .71 = .1201
    P(R’s win C only) = .94 * .82 * .29 = .2235
    P(R’s win none) = .94 * .82 * .71 = .5473

    Total probability of D’s winning 2 or 3:

    = .0349 + .1201 + .2235 + .5473 = .9258

    This is very wrong. This is more wrong than what I did first lol. You haven’t addressed many situations lol. A and B is not A+B.

  37. Gravatar of Student Student
    12. November 2022 at 06:39

    Notice yours does not sum to 1. The reason is because you left out many cases (the unlikely set but still worth 6-7 pts of the total.

    Notice in mine, before you subtract out p(1n2n3n4) it sums to greater than 1. This is because there is over lap between the case p(1n2n3) and p(1n2n3n4). Bottom line, you need to compute entire thing.

  38. Gravatar of ssumner ssumner
    12. November 2022 at 08:21

    Student, I don’t think anyone is understanding your comments. I don’t even know what you are trying to say here. Your figures are clearly wrong.

Leave a Reply