Betting markets >>>>>>> polls

I don’t know the outcome of the election, but several points are already clear:

1. Betting markets are more reliable than polls. The 538 poll average ended at about 8.5% for Biden. I believe they made some adjustments that brought it down to 7.8%. But this is obviously going to be a very close election—similar to 2016—and Trump might well win. The betting markets understood it was close, albeit favoring Biden. If Biden were to win based on a 1% or 2% lead in Pennsylvania, that would be right about what the betting markets thought was the median outcome.

2. A very close loss for Trump would set-up a rematch in 2024. He might be the first president since Cleveland to have two terms separated by the opposition party.

3. I feel good about my wishy-washy election forecast post. In September, I had forecast a Trump win but then later backed off when he had a horrible month. But some of the specific points seem accurate:

In addition, the polls in the US and UK have pretty consistently understated the support for unfashionable right wing populists. I noticed this with Brexit in 2016, but foolishly ignored this in my 2016 US forecast. Just as with Brexit, Trump did a couple points better in the popular vote than in the polls. This pattern repeated recently with Boris Johnson’s landslide win, and I expect a similar pattern here in November. Thus Biden needs to lead in the polls by about 5% on election day in order to win the popular vote by 3%. And that’s not even accounting for the high rate of invalid mail-in ballots, which some feel will hurt the Dems this year.

Commenters told me that polls had been fixed to account for less educated white males. I doubt it.

I also feel very good about my claim that if Biden’s lead in the polls had been less than 5% then he’d lose. Indeed 5% was too low—he might lose with an 8% national lead!

4. People should have paid more attention to news stories of Hispanic and black male voters shifting somewhat toward Trump. And also polls that said most people thought Trump would win.

Again, I don’t yet know the final outcome, but whoever wins it’s a real black eye for polling. And a win for betting markets. This is a very close election.

Update: New Jersey legalized pot! New York and Pennsylvania will likely be forced to follow suit, or else they’ll lose tax revenue to NJ.

Update#2: Arizona also legalized pot. Oregon decriminalized small amounts of hard drugs. DC decriminalized magic mushrooms.

Update#3: South Dakota legalized pot.

Overall, a very good night for ending the war on drugs.



36 Responses to “Betting markets >>>>>>> polls”

  1. Gravatar of Mark Z Mark Z
    3. November 2020 at 19:39

    It is pretty surprising to see his performance among black and hispanic voters. One question I have is: are these groups (especially males) trending rightward in general, or is Trump doing well among them *because* he’s Trump? That is, because his ‘bravado’ and other idiosyncrasies of his policies appeal more to that demographic than, say, Mitt Romney? It would be both troubling and ironic if the GOP’s best shot at winning minority votes is nominating someone like Trump.

  2. Gravatar of Peter Peter
    3. November 2020 at 19:55

    > account for less educated white males

    TBH I find it fails more on the middle to upper middle class white males. Those that have to be “woke” for social status reasons but really aren’t.

    That said I loved a talking head (woman, GOP shill) on NBC earlier that said “Look, most people are more the single issue voters. Women care about more than abortion, blacks care about more than BLM. Elderly care more about than COVID. People are complicated”.

    The part that will amuse me is if Trump takes the popular vote but loses the EC (to your #2 point) or takes EC and popular. I don’t think it will happen as the remaining popular vote is overwhelming Biden (West Coast) but will amuse me.

    Not going to lie, going into day (changed over about a month ago) I was thinking with 60% confidence Biden but as the night goes I’m thinking Trump might pull this off thankfully. Still early but most websites I’m looking out, outside NBC who is the outlier (no surprise), are projecting Trump.

    PS: No I didn’t vote for him, Brock Pierce got my vote because Hawaii prohibits write-ins.

    PSS: Curious how many states prohibit that. I’ve lived in a couple and didn’t realize that was even legal / anybody did until I moved here.

  3. Gravatar of ssumner ssumner
    3. November 2020 at 20:14

    Mark, Remember when conservatives argued that we can’t take immigrants from Latin America because they’ll vote for authoritarian demagogues? Ironies just don’t get any more ironic.

    Peter, I don’t think Trump gets the popular vote–that seems out of reach.

  4. Gravatar of Michael Sandifer Michael Sandifer
    3. November 2020 at 20:38

    At least three betting markets have now turned strongly toward Trump winning, giving Trump an average chance of about 70%.

    If Trump wins, this is a huge repudiation for my view of the electoral landscape.

  5. Gravatar of Mark Z Mark Z
    3. November 2020 at 20:39

    Yeah that idea that the GOP can’t win hispanics, especially 2nd+ generation, is taking a beating.

  6. Gravatar of Cartesian Theatics Cartesian Theatics
    3. November 2020 at 21:08

    This may be terrible for the country, but it’s amazing news for prediction market PR! Maybe it’s an idea who’s time has finally come. Next up ngdp, unemployment, climate, and everything else.

  7. Gravatar of Peter Peter
    3. November 2020 at 22:00

    @Scott: Probably sleeping but a PS to your NJ update, even better the results in Oregon with Measure 110 (Hard drugs). Hopefully that starts to expand even more.

    @Mark: Yeah the problem with the Democrats is it’s always been a big tent which at some point collapses. At some point the woke left starts to pressure the conservative labor right and at that point the GOP vastly expands as long as Duke doesn’t take over. Love or hate Trump but he did his damnedest to shrug off “old white guy party” from Kayne West / Kardashians to the most (non-black/Hispanic) minorities in key rolls (Indian, East Asian, Jew, etc) as well as didn’t overly concentrate on the “woke” right (Religious Right, etc) which past GOP Presidents have focused on which drives out the moderate/liberal right equiv. Most Democrats, or at least a significant majority, are Republican in all but name from the Asians in Hawaii / West Coast to blacks and Mexicans. Generally they hate Tyrone the gang banger (BLM), homosexuals (BLM), Feminists, and well pretty much anything AOC and crew would champion. The only thing that keeps them in the DNC is the GOP toxic name brand.

  8. Gravatar of Cartesian Theatics Cartesian Theatics
    3. November 2020 at 22:14

    Wow, electionbettingodds still locked at 58%. And seriously new Tweeting lows for Trump. God this guy is the worst.

  9. Gravatar of Michael Sandifer Michael Sandifer
    3. November 2020 at 22:25

    Correction: This is already a huge repudiation of my view of the electoral landscape.

  10. Gravatar of ssumner ssumner
    3. November 2020 at 22:57

    I’m going to bed, but if Fox News is right in calling AZ (not a given in this crazy year), It’s very good news for Biden. He’d no longer need PA (or Georgia).

    (I have to think that Biden will still win WI and MI.)

    This is the second race in my life not determined on election night, along with 2000 obviously.

  11. Gravatar of Peter Peter
    3. November 2020 at 23:10

    @Scott: Thinking one at most, not both though regardless I’m back to leaning Biden.

  12. Gravatar of Benjamin Cole Benjamin Cole
    3. November 2020 at 23:46


  13. Gravatar of Benjamin Cole Benjamin Cole
    3. November 2020 at 23:49

    The latest odds in London show a 67% chance of Trump winning. About 2 am NYT.

    “2020 Election Odds & Predictions: Trump’s Odds Back Above 65% as Betting Market Favors Pennsylvania & Georgia Wins”

    The link seems to get my comment banned.

  14. Gravatar of Christian List Christian List
    4. November 2020 at 03:29

    My prediction seems to be 100% correct at the moment. Every single state. I have Wisconsin and Michigan for Biden. I assume they will go to Biden, so in the end it’s 268:270 for Biden.

    With just one go I would have bet all my money on this result, but in Germany one can hardly bet on elections. How were the odds for 268:270, certainly a nice sum.

    Mabye I have one fluke, I have Georgia for Trump but it could go to Biden as well.

  15. Gravatar of sty.silver sty.silver
    4. November 2020 at 04:44

    You are drawing precisely the opposite conclusion from this that I’m drawing. I think this case was strong (almost slam-dunk) evidence that the markets were badly wrong.

    Well, markets > polls is still true. But models > markets, at least in this case.

    After Trump won Florida and Biden won Arizona, 538’s model had Biden at about 85% to win, whereas markets had him at 33%. Then, markets changed for about 33% in favor of Biden which made NO FUCKING SENSE because almost nothing changed in bewteen. Maybe there was an issue with the market and people couldn’t bet, or maybe electionbettingodds didn’t update correctly or something. But if it’s legit, it basically shows that the entire market was horribly mis-calibrated and took several hours (during which there was barely any new information) to fix that.

    I’m also neither convinced that this will be all that close, nor that 90% was an incorrect estimate. This looks to me like it’s somewhere in the most pessimistic forth of the forecast or something, about a 25% chance.

    Full disclosure, I bet about 3500$ on Biden because I thought the markets were mis-callibrated. I think that reasoning was basically correct.

  16. Gravatar of sty.silver sty.silver
    4. November 2020 at 04:49

    I should say that

    1. maybe I’m wrong in saying that there was no new information. I was checking races and following the 538 crew on twitter, and they didn’t think there were any big news in that time.

    2. Here is the chart I’m talking about (set to showing the previous day):

    The way it initially plummeted might be defensible, I think that’s not clear. But the way it stayed low and then went up way too late makes no sense.

  17. Gravatar of Christian List Christian List
    4. November 2020 at 05:29

    I no longer see a good path for Trump. I strongly assume Michigan is going to flip to Biden sooner or later, Trump would need a miracle there, which won’t happen, so he would have Arizona, Wisconsin, Nevada as theoretical last outs, but they are all basically gone already.

    It will be the 268:270 that I feared most. I wish I would have been wrong. A landslide for Biden would have been better, cleaner. And 269:269 would have been too funny.

    The last remaining trick would be lawsuits and Trump’s jokers from the Supreme Court, all of these tricks not really clean in my eyes, hitting and a stabbing, like in the Middle Ages.

  18. Gravatar of Michael Sandifer Michael Sandifer
    4. November 2020 at 05:32

    Sharp disparity in performance for Republicans in this election versus 2018 suggest that Republican voters don’t really care much about the Party. It’s all about Trump. Not surprising, I suppose, but where do Republicans go now with Trump likely defeated?

  19. Gravatar of zephito zephito
    4. November 2020 at 05:42

    Biden certainly seems to be in the driver’s seat. I had it 291 for him; he might get that or 305 or 306. The pop vote margin should shift towards him as well. If he wins by 6 points and with 306 EV, would the polls really be that far off?

    Also, the betting markets were completely unjustified in giving Trump 70-80% last night after the Florida call. Not that Trump couldn’t still have won (or could still win), just that there was no way it should be at 80% at that time. I thought I was going crazy when I saw that, with Biden performing better vis a vis HRC in some of the suburban midwestern counties. Now, of course, it’s shifted back.

    I do agree that the perception of the polling industry will take a beating, whether that’s fair or not. I would also agree that if it ends up being Biden +3 overall, that would be pretty far off. We’ll have to see what the final total is.

  20. Gravatar of Michael Rulle Michael Rulle
    4. November 2020 at 06:14

    Polls have proven to be biased even when they announce Biden’s victory. Re:538, he should never have done his “auto-correlation” model—-and kept to the old model. Any model which changes a 10% probability to a 70% probability with a one state change in a toss-up state is obviously flawed. This is what happens when pollsters “root”.

    Paranoia strikes deep—-on both sides and with good reason. I will make the case for GOP paranoia. PA, which ironically was the lodestar state but may not matter, was already playing counting games last night—-as if they were trying to figure out how many votes they needed from Philly to win—-if any—-if Biden won without it

    Also, mail in counting in Wisconsin and Michigan also was done late—-as were cities. The opposite of what was expected.

    Trump will have to hold the lead in all the current states he is ahead in as of 8am Wednesday——given that includes Michigan and PA—-good luck.

    Trump had a 300k vote lead in Mich with a 68% count. At that point in time 100% of the remaining Detroit vote would have to to go to Biden to almost tie. And low and behold at 92% state count it is almost tied.

    I am sure Dems will come up with their own paranoia stories—-all likely valid. The randomness of mail in rules is a recipe for distrust. The popular vote count was called again by Rasmussen more accurately than all others——1 % for Biden versus 1.7% actual so far.

    Worst of all——-is the loser will feel ripped off even if for moronic reasons. This is like 1960—huge turnout two presumed rip-off states——and a virtual tie. Then, Nixon just let it go. One thing we do know is neither of these two guys will let it go.

  21. Gravatar of Michael Rulle Michael Rulle
    4. November 2020 at 06:38

    As of now, if Trumps wins states he is ahead in it is 268——I cannot imagine a worse outcome—-maybe 269 or 270. If he holds leads and gets Nevada he wins. I assume betting markets are heavily Biden. Have not looked and don’t care anymore.

  22. Gravatar of Thomas Hutcheson Thomas Hutcheson
    4. November 2020 at 06:48

    And for the war on Uber

  23. Gravatar of Big Al Big Al
    4. November 2020 at 06:57

    I too wish it is more certain, but I actually think that Biden will pick up PA in the end which should give him a more comfortable margin. Plus GA may still swing his way too.

  24. Gravatar of Student Student
    4. November 2020 at 06:58

    Indeed they are… they seem to be moving sharply to Biden this morning… predictit has Biden at 80cents and trump at 17cents as of 10:00am.

  25. Gravatar of sty.silver sty.silver
    4. November 2020 at 07:40

    I think Trump never had a chance in Michigan. He had a 10 point lead in the polls. Errors happen, buy errors by 10 points are exceedingly rare.

    The only evidence for Trump having been ahead is that he was ahead in early counts. But 538 warned us ahead of time that this would happen because of the mail-in vs. in-person votes. I would be surprised if Biden didn’t win Michigan comfortably, and extremely surprised if he lost it.

    Although, he won’t win it by 10 points. There was a sizeable error in the polls (but less than 10%).

  26. Gravatar of zephito zephito
    4. November 2020 at 07:46

    I’m not sure I agree with you Michael R. Peoria county in IL went for Biden pretty early in the race. So goes Peoria, so goes the rest of the upper midwest (other midwestern counties also showed Biden out-doing HRC). Considering that Trump won MI and WI by the barest of margins last time, he had no room for error and this did not bode well for him. AZ was the uncertain factor for me; once that came in I felt there was an extremely good chance Biden would get it and went to bed. As expected, he’s now ahead. At this point there is absolutely no hard evidence for major cheating and no circumstantial evidence either. I suspect Biden will catch up in PA too.

  27. Gravatar of zephito zephito
    4. November 2020 at 07:47

    BTW, Trump carried Peoria last time so the evidence was the shift in vote to Biden, not the fact that Biden won outright.

  28. Gravatar of Todd Kreider Todd Kreider
    4. November 2020 at 08:08

    If Biden wins, Trump goes to a talk show and won’t get the nomination in 2024 as a 78 year old. Given Biden’s cognitive decline, Harris could easily take over as President in 2023 or 2024.

  29. Gravatar of janice janice
    4. November 2020 at 08:26

    It was a beautiful and wonderful night for communism world wide.

    1. Oregon’s policy allows you to consume small amounts of cocaine and meth legally, but apparently you cannot sell it. Interesting, because in order to consume the product you actually have to buy the product from a SELLER. Yeah, amazing concept that even a third grade moron could grasp. What’s really happening is that these politicians, like in other leftist/Marxist/communists states, are actually involved in the drug dealing. So if they remove your ability to sell and make consumption legally, they can keep their monopoly and increase demand at the same time. You watch what happens in Oregon in the future. Tremendous violence and crime will overtake that state.

    2. If conservatives don’t hold the senate or presidency, your communist party will now be an authoritarian party (which is precisely what they want). And they will ensure they hold onto power by packing the courts, engaging in more ballot harvesting, buying votes, and adding a bankrupt and corrupt Puerto Rico and D.C to your list of communist states. Changing the constitution is a communist ideal. To them nothing is absolute. Only relative.

    So congratulations to the baby boomer generation for destroying the American dream. They were the spoiled kids of American hero’s who then raised spoiled degenerate morons.

    The very term “build back better” is a term used by Klaus Schwab to define the “great reset”. In other words, you now live under the CCP. And you have now forced future generations to fight a civil war to regain their freedoms.

    You should be proud =/

  30. Gravatar of ssumner ssumner
    4. November 2020 at 08:41

    sty.silver and zephito, Maybe the short term moves in the market last night were wrong, but its performance over the past month was impressive. A 65% odds of a Biden win is like a point estimate that he’ll win the tipping point state by 1% or 2%. And it looks like he’ll win Wisconsin (or possibly Pennsylvania, or both) by 1% at most. The polls had him at 5% in the tipping point state. Sorry but even if Biden wins by 5% in the popular vote, this was a very close election.

    Michael, You said:

    “PA, which ironically was the lodestar state but may not matter, was already playing counting games last night—-as if they were trying to figure out how many votes they needed from Philly to win—-if any—-if Biden won without it”

    And so it begins . . .

  31. Gravatar of D.O. D.O.
    4. November 2020 at 08:53

    Isn’t American politics wonderful. Florida is a red state (legislature+governor+senators) and it went for Trump stronger then expected presumably because Cuban Americans don’t like socialism… and minimum wage amendment won 60/40. Minimum wage is probably the most “socialist” thing there is in US. Culture war keeps raging…

  32. Gravatar of Christian List Christian List
    4. November 2020 at 09:07

    It’s true, apparently Biden can also flip PA. That would be really important, because then the result would be safer.

    If he doesn’t win PA, then he needs every single Electoral Vote, right? Trump will then attack and sue in multiple places.

    He probably wouldn’t be in the right, but as they say in soccer, it is goal only when the referee decides on goal. And the final referees are on the Supreme Court.

    But even if it came to that, I don’t think that the judges would make a decision that would be clearly illegitimate. They are not the lackeys of Trump, even if some fake news press keeps claiming so.

  33. Gravatar of Michael Rulle Michael Rulle
    4. November 2020 at 11:28

    Scott–well cheating “COULD” happen—not saying it is—but if it were—what better state could it occur than where the AG says “if all the votes are counted Trump is going to lose”—- 2 days before the election. Is it the function of an AG to tell the public who is going to win, 2 days before an election? No. They represent a fair election and a fair process—not going out on a limb to say one candidate wll lose–How we he know that—-yet he he asserted that Trump will lose—-“if we count all the votes”

    Proves nothing except at best he is a sleaze bag—who COULD cheat

  34. Gravatar of ssumner ssumner
    4. November 2020 at 12:19

    Michael, You said:

    “Proves nothing except at best he is a sleaze bag—who COULD cheat”

    Wait, are you saying Trump is a sleaze bag because he says outrageous things every day? Stuff very similar to what the AG says? I thought you supported Trump.

  35. Gravatar of xu xu
    4. November 2020 at 15:58

    Massive win for Donald Trump
    These corrupt officials refused to let conservatives observe the polls in a city that is #1 in corruption in the United States. That was part of the deal! And supreme court will absolutely rule against them.
    Plus there are videos now of burning trump votes in Michigan.
    This is the end of the democratic party stranglehold on the rustbelt states.
    Not only will they lose face, and now be known as the party of corruption. But Hunter Biden will almost certainly be arrested within the next 3-6 months.

  36. Gravatar of Luc Mennet Luc Mennet
    5. November 2020 at 10:25

    One thing that the betting markets seems to have messed up is interpreting the first results that came in on election night, where going into November 3rd had Biden at ~65%, then at about 10pm PST had trump at ~75% to win, then switched to Biden at ~80% the morning of the 4th once it became obvious that there would be a massive swing towards Biden as mail ins got counted. People like 538 on the other hand had been correctly predicting this swing since early October.

Leave a Reply