The Grant Williams Podcast

Marvin Barth argues polling bias favors Democrats and predicts Republican midterm gains via populist voting shifts.

1h 16m Listen to the original Episode page
US Midterm ElectionsPolling MethodologyPopulismDonald TrumpFiscal PolicyBetting Markets

Summary

Marvin Barth of Thematic Markets argues that standard polling consistently overstates Democratic support because it fails to capture low-trust, right-leaning populist voters who are less likely to respond to surveys. He predicts Republicans will hold the Senate and has a 20-30% chance of retaining or gaining seats in the House, citing historical rarity of incumbent party losses and structural polling errors. Barth also outlines a 'Jacksonian' framework for Donald Trump, suggesting a Republican Congress would empower Trump to pursue fiscal consolidation through tariffs and entitlement reform rather than debt expansion. He concludes by distinguishing between quantifiable 'objective risks' and unquantifiable 'uncertainty,' warning that betting markets fail to price the latter.

Episode sections

  1. 0:10 – 2:57

    Introduction and Context

    Grant Williams introduces Marvin Barth, highlighting his track record of challenging consensus election views.

  2. 2:57 – 8:04

    The Consensus View on Midterms

    Discussion of the prevailing consensus that Democrats are poised for a midterm sweep, based on generic congressional polls showing a 7-9 point Democratic advantage.

  3. 8:04 – 16:27

    Barth’s Track Record and the Populism Hypothesis

    Barth explains his history of correctly predicting election outcomes by identifying low-trust populist voters who are underrepresented in polls.

  4. 16:27 – 22:07

    Defining Populism and Polling Bias

    Barth defines populism as a reaction to disenfranchisement and explains how low-trust voters create a structural bias in polling that is hard to correct.

  5. 22:07 – 25:18

    Left-Wing Populism and Trust Dynamics

    Analysis of why left-wing populist movements may not suffer from the same polling undercount due to their higher-trust, educated demographic base.

  6. 25:18 – 33:51

    Bayesian Process and Data Analysis

    Barth describes his Bayesian methodology, looking for correlations between past polling errors and current sample composition rather than proving hidden voter existence.

  7. 33:51 – 38:05

    Polling Methodology and Bias

    Barth argues that increasing sample sizes does not fix bias driven by low trust and admits his analysis is not peer-reviewable but contradicts consensus views.

  8. 38:05 – 42:16

    Senate and House Election Predictions

    Barth predicts Republican Senate holds and a 20-30% chance of Republican House control, citing gerrymandering and polling errors in key states.

  9. 42:16 – 49:41

    Historical Context of Incumbent Party Gains

    Review of the four instances since the Civil War where the president's party gained House seats, suggesting a rare Republican gain is possible.

  10. 49:41 – 57:47

    Trump’s Power and the 'Jacksonian' Framework

    Barth argues a Republican Congress would empower Trump to pursue a 'Jacksonian' legacy through bold, historic actions rather than legislative checks.

  11. 57:47 – 1:01:30

    Fiscal Consolidation and Tariffs

    Prediction that Trump will push for fiscal consolidation via tariffs and wealth taxes to offset income tax cuts and address US debt sustainability.

  12. 1:01:30 – 1:06:38

    Entitlement Reform and Political Strategy

    Speculation that Trump may cut entitlements like Obamacare to reduce spending, framing it as reform to appeal to lower-income voters.

  13. 1:06:38 – 1:06:58

    Mamdani Rage and Election Risks

    Brief mention of 'Mamdani rage' as a potential risk factor for the election outcome, though details are cut off.

  14. 1:06:58 – 1:15:40

    Polling Errors and Populism

    Discussion of the correlation between populist voting and polling errors, with potential for reversal in Democratic-leaning states.

Claims worth verifying

The consensus view is that Democrats will win a sweep of the midterms, with betting markets pricing in a 92-95% chance for the House and less than two-thirds for the Senate.

Unidentified speaker ·6:55

How to verify: Check current odds on major betting markets (e.g., PredictIt, Kalshi) and financial news reports on midterm election probabilities as of the recording date.

Generic congressional polls show a 7-9 percentage point advantage for Democrats heading into the midterms.

Unidentified speaker ·5:43

How to verify: Review aggregated generic congressional ballot polls from sources like FiveThirtyEight or RealClearPolitics for the relevant election cycle.

Marvin Barth correctly predicted Donald Trump's wins in the 2016, 2020, and 2024 elections against the consensus polling view.

Unidentified speaker ·8:52

How to verify: Compare Barth's published forecasts from 2016, 2020, and 2024 against actual election results.

In 2014, polling errors favored Democrats even more significantly than in 2016.

Unidentified speaker ·12:34

How to verify: Analyze historical polling data and actual results for the 2014 US midterm elections.

Populist voters are 'low trust' voters who do not answer polls but turn out in high numbers when populist issues are on the ballot.

Unidentified speaker ·9:44

How to verify: This is a theoretical framework presented by Barth; verification would involve analyzing voter turnout data against trust metrics in past elections, which Barth cites as his research basis.

David Shore, Obama's 2012 chief data scientist, found that low-trust PhDs voted for Trump and high-trust working-class whites voted for Clinton.

Unidentified speaker ·31:11

How to verify: Search for David Shore's published data or presentations regarding trust metrics and voting behavior in the 2016 election.

Pollsters adjust weights to correct for demographic biases, such as over-sampling older white women and under-sampling young Latino men.

Unidentified speaker ·29:52

How to verify: Review methodology sections of major polling organizations (e.g., Pew, Gallup) which typically describe their weighting procedures.

Left-wing populist movements in Europe (e.g., France's Mélenchon, Germany's De Linka/BZD) have punched above their polling weight in actual elections.

Unidentified speaker ·23:32

How to verify: Compare pre-election polling averages with final election results for the French and German left-wing parties mentioned.

Populism is defined by political disenfranchisement where voters feel center-left and center-right parties ignore their preferences on issues like immigration and trade.

Unidentified speaker ·19:46

How to verify: This is a definition provided by Barth; it can be verified by reviewing his 2016 study 'The Politics of Rage' mentioned at 19:26.

Barth is a former Bush administration appointee.

Unidentified speaker ·24:43

How to verify: Check Marvin Barth's professional biography or LinkedIn profile for government service history.

Polling errors in the 2014 midterms had demographics most similar to the current election cycle.

Marvin Barth ·37:19

How to verify: Compare demographic composition data from 2014 midterm polls with current polling data via Pew Research or Census Bureau datasets.

The generic ballot historically overstates the popular vote for Democrats in House races by about two percentage points.

Marvin Barth ·39:41

How to verify: Analyze historical generic ballot polling results against actual House popular vote totals from the Federal Election Commission (FEC).

There are approximately 75 House races that are potentially in contention.

Marvin Barth ·40:43

How to verify: Review the Cook Political Report’s current House race ratings for 'Toss-up' or 'Lean' categories.

The incumbent president's party has gained House seats only four times since the Civil War: 1902, 1934, 1998, and 2002.

Marvin Barth ·42:24

How to verify: Check historical House seat changes by party for midterm elections from 1865 to present via the Clerk of the House or FEC archives.

Real spending in the United States was cut under Obama, despite nominal increases due to inflation.

Marvin Barth ·58:08

How to verify: Review US Treasury Department data on federal outlays adjusted for CPI inflation during the Obama administration (2009-2017).

The Social Security Trust Fund is projected to go into bankruptcy in the next presidential term.

Marvin Barth ·59:15

How to verify: Consult the most recent Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds.

Donald Trump’s voters listed federal spending and inflation as their two biggest priorities.

Marvin Barth ·59:43

How to verify: Review exit polls or pre-election surveys from major polling organizations (e.g., Pew, Gallup) conducted during the relevant election cycle.

Tariffs fall overwhelmingly on lower-income households because they spend a higher percentage of income on goods.

Marvin Barth ·1:03:30

How to verify: Analyze Bureau of Labor Statistics Consumer Expenditure Survey data on spending shares by income quintile for goods vs. services.

Howard Lutnick repeatedly discussed raising tax rates on highest earners to finance tax cuts for the bottom 90%.

Marvin Barth ·1:02:08

How to verify: Search public transcripts, interviews, or press releases from Howard Lutnick in the first half of 2024/2025.

The US is on an unsustainable fiscal path, though it has more runway than France.

Marvin Barth ·58:50

How to verify: Compare long-term fiscal projections from the Congressional Budget Office (CBO) with those from the French Ministry of Finance or OECD fiscal outlooks.

Polling errors in US states go up with the share of the Republican vote, forming a 45-degree line relationship.

Unidentified speaker ·1:07:51

How to verify: Review historical US election polling data (e.g., FiveThirtyEight or RealClearPolitics) to correlate state-level polling error with state-level Republican vote share.

In Maine, polling shows voters want Democrats to control the Senate, but they vote for Republican Susan Collins.

Unidentified speaker ·1:08:41

How to verify: Check Maine Senate election results and concurrent partisan control polls from the relevant election cycle.

Michigan and Minnesota have narrow Democratic leads of only one or two points, within the margin of error.

Unidentified speaker ·1:09:19

How to verify: Check current polling averages for the US Senate races in Michigan and Minnesota at the time of the podcast release (Nov 2026).

Betting markets are good at pricing non-farm payrolls because there is a lot of data and known statistical processes.

Unidentified speaker ·1:10:00

How to verify: Analyze prediction market accuracy for US Bureau of Labor Statistics non-farm payroll releases.

In December 2019, no one thought 7 billion people would go into lockdown the following year.

Unidentified speaker ·1:10:43

How to verify: Review historical news archives and scientific literature from late 2019 regarding pandemic predictions.

Do your own research

  • Verify the historical frequency of incumbent party House seat gains in midterm elections using Congressional Research Service data or official House records.
  • Review current generic congressional ballot polls from major aggregators (e.g., FiveThirtyEight, RealClearPolitics) to assess the stated Democratic advantage.
  • Check the solvency timeline of the Social Security Trust Fund using the official Social Security Administration Trustees Report.
  • Analyze the composition of recent polling samples to see if they adequately weight for 'trust' metrics or if they rely solely on traditional demographics like age and race.
  • Research the 'Jacksonian' political tradition to understand the historical framework Barth uses to predict Trump's legislative behavior.
  • Compare betting market odds on Polymarket or similar platforms against Barth's probability estimates for Senate and House control.
  • Investigate the specific polling error margins in Michigan and Minnesota from previous election cycles to validate Barth's 'knife-edge' state claims.
Full transcriptexpand
0:10
Unidentified speaker: Before we get going, here's the bit where I remind you that nothing we discussed should be considered as investment advice.
0:16
Unidentified speaker: This conversation is for informational and hopefully entertainment purposes only.
0:20
Unidentified speaker: So while we hope you find it both informative and entertaining, please do your own research or speak to a financial advisor before putting a dime of your money into these crazy markets.
0:30
Unidentified speaker: And now, on with the show.
0:37
Unidentified speaker: Welcome everybody to another edition of the Grant Williams podcast.
0:41
Unidentified speaker: Joining me in a moment, returning to the podcast is a good friend of mine, Marvin Barth of Thematic Markets.
0:47
Unidentified speaker: I've had a couple of podcast interviews with Marvin and actually had a couple of video interviews with him on my now defunct gold tier.
0:54
Unidentified speaker: The silver tier members listening, you can watch those conversations with Marvin.
0:58
Unidentified speaker: He's a fascinating guy.
0:59
Unidentified speaker: Every time I sit and chat with him, every time I read his work, I am left with so many thoughts running through my head.
1:04
Unidentified speaker: And he and I had dinner in person in London a few weeks ago, and he had my head spinning with something that he was talking about, getting ready to publish.
1:12
Unidentified speaker: He's published it this week, and so I thought this is a perfect time to impinge on our friendship and ask him if he'd record a conversation with me about it, which he was very gracefully willing to do.
1:22
Unidentified speaker: And you are about to listen to the fruits of that conversation.
1:25
Unidentified speaker: Marvin's work, as I say, is absolutely spectacular.
1:28
Unidentified speaker: At thematicmarkets, he is on Twitter and thematicmarkets.com.
1:31
Unidentified speaker: You'll find everything he writes there.
1:33
Unidentified speaker: And I would urge you to give it the once over because I think you'll find it very useful.
1:37
Unidentified speaker: He has a sub stack, same thing, thematic markets, you'll find it all there.
1:41
Unidentified speaker: And as you're about to find out, he is a man of great intellect, and he will challenge you in ways you aren't quite yet ready for, I promise you.
1:50
Unidentified speaker: Anyway, that's it from me.
1:52
Unidentified speaker: Please enjoy my conversation with Marvin Bart.
2:04
Unidentified speaker: Marvin, my friend, welcome back to the podcast.
2:07
Unidentified speaker: It's great to see you.
2:08
Unidentified speaker: It's fabulous to be back.
2:10
Unidentified speaker: Thank you very much for having me on, Grant.
2:12
Unidentified speaker: You and I had a very long, overdue, and most enjoyable dinner the other night when I was back in London.
2:17
Unidentified speaker: Thank you.
2:17
Unidentified speaker: And we got to talking as we invariably do.
2:19
Unidentified speaker: And I've realized that I need to confine my social engagement to you to lunch so that I have time in the afternoon and evening to process all the things that you get me thinking about.
2:29
Unidentified speaker: Because I went home from dinner and I'm sitting there trying to get to sleep and I got all this stuff written away.
2:33
Unidentified speaker: Like, curse you, Marvin, you did it again.
2:35
Unidentified speaker: So I think in future lunch, or better still, breakfast, because then I get the whole day to try and process it.
2:40
Unidentified speaker: But seriously, we talked about a whole bunch of stuff, and I'm not sure how much of that we'll get onto today.
2:44
Unidentified speaker: But there's one very specific place I want to start because it was a thing that we talked about at where you most challenged my expectations, challenged what I strongly believe to be pretty much written in stone, and that is the upcoming midterms.
2:57
Unidentified speaker: Now, you have a phenomenal track record of calling these elections when a lot of people are pretty set in their expectations, and you say, well, hang on, let me thought about this.
3:07
Unidentified speaker: My data's telling me something very different's going to happen.
3:09
Unidentified speaker: And your track record has been outstanding with that.
3:12
Unidentified speaker: So when I when we come up to an election and I want to know what I think might happen, I always listen to you, and you did it again.
3:19
Unidentified speaker: You got me really thinking.
3:20
Unidentified speaker: So let's start with how you see the consensus currently sitting and where that consensus comes from.
3:29
Unidentified speaker: And then after that, let's get into your own work in trying to understand the reality beneath the consensus.
3:36
Unidentified speaker: Okay.
3:36
Unidentified speaker: So I think the consensus is pretty straightforward, and betting markets are a great indicator of it, which actually is if we have time, there's a whole wonderful, interesting discussion we could have about when betting markets are a useful indicator and when they're not.
3:56
Unidentified speaker: But let's take them as an indicator of where the consensus sits.
4:00
Unidentified speaker: And certainly I can also tell you that the feedback on my piece uh from all my clients in a wide range of different markets has been wow, that's out there.
4:13
Unidentified speaker: Yeah, yeah, exactly right.
4:15
Unidentified speaker: Thank God.
4:16
Unidentified speaker: All these people not sleeping because of you, man.
4:19
Unidentified speaker: Yeah, so I can pretty safely say the consensus is that this is a democratic sweep.
4:26
Unidentified speaker: I think the most optimistic Republicans out there are that they might be able to hold the Senate or at least keep it to a 50-50 tie where the vice president JD Vance will break the tie.
4:40
Unidentified speaker: But of course, that means he has to stay on Capitol Hill all the time.
4:43
Unidentified speaker: So, where does that come from?
4:44
Unidentified speaker: That comes from, if you look at the polls, have quite consistently been building all year towards increased advantage for the Democrats.
4:55
Unidentified speaker: So you look at the sort of what's called the generic congressional poll.
5:00
Unidentified speaker: And it's generic because you're not asking about a specific candidate.
5:03
Unidentified speaker: You're just asking people, hey, who do you intend to vote for Democrats or Republicans in in Congress this year?
5:09
Unidentified speaker: And by the way, that gives you a different answer than if you ask them about the specific person.
5:14
Unidentified speaker: So there's a big caveat around that.
5:17
Unidentified speaker: But that's been showing an advantage for Democrats that started at the beginning of the year at a couple of points and it's been increasing, increases significantly after the Iran war.
5:28
Unidentified speaker: And then actually in September, it really takes another bump up, interestingly enough.
5:33
Unidentified speaker: And so the latest ones are that it's around, depending on how you measure, seven to nine percentage points in favor of the Democrats.
5:43
Unidentified speaker: And that's for the Congress.
5:44
Unidentified speaker: And the reason why people generally focus on the generic ballot polls is there's 435 house seats, right?
5:54
Unidentified speaker: It's really expensive to go conduct a poll.
5:57
Unidentified speaker: They don't conduct them for very many house seats, or at least not publicly available ones.
6:02
Unidentified speaker: But Senate races, there's only 100 senators, and only a third of them come up for election every two years.
6:10
Unidentified speaker: And only a dozen of those are even in contention.
6:13
Unidentified speaker: So they poll those.
6:15
Unidentified speaker: And same thing there.
6:16
Unidentified speaker: You've seen a movement in several of these states towards Democrats.
6:23
Unidentified speaker: And if you were to just look at polling numbers, it looks like the Democrats would pick up seven seats in the Senate.
6:33
Unidentified speaker: And given that the Republicans only have a three-seat majority, that swings it to the Democrats.
6:38
Unidentified speaker: So this is where this is coming from.
6:40
Unidentified speaker: And betting markets have the House at I think 92 to 95% goes Democrat, and a probability of it's down a bit for the Senate because some of the incoming data from early returns have changed that since I published.
6:55
Unidentified speaker: But now it's down to around a little less than two-thirds chance that the Democrats take the Senate.
7:01
Unidentified speaker: So that's as close as you'll get to quote unquote a landslide in an election of this nature, right?
7:07
Unidentified speaker: When people talk about landslide, that's a pretty decent picture of what they're talking about, which kind of seems weird, particularly for people in the UK where a landslide is normally counted in the hundreds of seats, and you think that seven could be a landslide.
7:19
Unidentified speaker: But as you say, that that's a big shift in terms of control and power.
7:23
Unidentified speaker: So let's start up with the fact that you went into this exercise expecting the Republicans perhaps to lose both chambers.
7:33
Unidentified speaker: And then you started doing the work, your work, that has taken you to conclusions that many people didn't get to previously.
7:42
Unidentified speaker: So I don't know how much you want to talk about the kind of work that you do and the kind of questions you ask the places you go, but to the best extent you can, lay out how you approach producing your own polling, if you like, on the likelihood of either side winning.
8:04
Unidentified speaker: So what am I doing forecasting elections out there?
8:07
Unidentified speaker: And yet I've been doing this for many years.
8:10
Unidentified speaker: And as you kindly noted, my track record on calling these elections has been quite good.
8:16
Unidentified speaker: And in particular, it's been in calling them against the grain of what the consensus was.
8:23
Unidentified speaker: So for 2018 and 2022, the previous two midterms, for instance, I didn't have an out-of-consensus view.
8:31
Unidentified speaker: I've figured that the Democrats were going to pick up seats in both of those and had no real significant reason to challenge that view and came out as expected.
8:43
Unidentified speaker: But 2016, I thought it was more likely than not that Donald Trump was actually going to win, despite his being behind in the polls.
8:52
Unidentified speaker: In 2020, polls said he was nine percentage points down.
8:58
Unidentified speaker: I thought it was going to be a very tight election and that he would lose very narrowly, which turned out to be the case.
9:04
Unidentified speaker: And in 2024, he was down a few points.
9:08
Unidentified speaker: And I thought he's actually going to win the popular vote, and he's going to have what we Americans call coattails.
9:15
Unidentified speaker: He's going to drag along people in the House and Senate.
9:18
Unidentified speaker: And all of those things came to pass.
9:21
Unidentified speaker: And so what was going on there?
9:23
Unidentified speaker: And the issue is effectively this phenomenon of populism, which as you know, I've done a lot of research on for its potential economic effects and market effects.
9:37
Unidentified speaker: And one of the things about populists is they tend to be low trust voters.
9:44
Unidentified speaker: And they are also people want to think of them as sporadic voters.
9:49
Unidentified speaker: I don't think that's the correct way to characterize them.
9:52
Unidentified speaker: I think they are reliable voters when a populist is on the menu.
9:58
Unidentified speaker: So if they don't get a populist on the menu, yeah, they don't vote.
10:03
Unidentified speaker: But if a populist or a populist issue is on the menu, say Brexit in 2016 in the UK, they do come out.
10:12
Unidentified speaker: And that's why you see these surges in turnout around these.
10:17
Unidentified speaker: And in 2016, in both Brexit and in the US election, one of the things that was the clearest signal to me that you were getting all these low trust voters were actually coming out was when you saw in the UK that turnout numbers were high.
10:34
Unidentified speaker: I was like, okay, this is all these people who felt that they've never had a voice before are coming out.
10:39
Unidentified speaker: The second one is in the US, we have this early voting, which we're going through right now, right?
10:46
Unidentified speaker: And you see all the people out there counting up how many of these are people who are registered Democrats or registered Republicans have turned in their vote anticipating that those people are going to vote along party lines.
10:57
Unidentified speaker: And that's what they really want to focus on.
11:00
Unidentified speaker: And that's where all the news focus is.
11:03
Unidentified speaker: What I looked at is what was happening to independent voters, people who are not registered with either party or who hadn't voted in either primary.
11:11
Unidentified speaker: And the reason for that is because nobody is trying to get those people to turn out.
11:17
Unidentified speaker: The Democrats have a big machine to get their voters out, the Republicans have a big machine to get their voters out, but nobody is saying, hey, let's go get those wild card independent people out to vote.
11:29
Unidentified speaker: Nobody wants that, right?
11:31
Unidentified speaker: And yet, independent votes were through the roof in 2016 in the states that were reporting it.
11:37
Unidentified speaker: And I said, there you are again.
11:39
Unidentified speaker: These are all those people who are not really affiliated with either party, but they see some, they see a populist issue on the ballot.
11:48
Unidentified speaker: And at that point it was Donald Trump.
11:51
Unidentified speaker: And there is this perception that in the US it's really only about Donald Trump.
11:56
Unidentified speaker: Because as I mentioned, 2018 and 2022, the midterm elections seemed to be roughly in line with polls.
12:07
Unidentified speaker: By the way, there's some caveats to that.
12:10
Unidentified speaker: But the places where people really started to notice these poll errors, and they're not poll errors, they're poll bias.
12:16
Unidentified speaker: They were in one direction.
12:18
Unidentified speaker: They continuously overstated Democrats' lead was in the presidential elections of 2016, 2020, and 2024.
12:27
Unidentified speaker: But if you go back and look a bit further, you see the same thing in 2014.
12:34
Unidentified speaker: 2014 actually had even bigger polling errors, polling bias in favor of Democrats than 2016 did.
12:44
Unidentified speaker: And what's interesting about that is that was the first year you really start to notice populism show up in candidates and platforms.
12:54
Unidentified speaker: So you had candidates, Johnny Ernst in Iowa, like Tom Cotton in Arkansas, both of whom massively outperformed their polls.
13:05
Unidentified speaker: You also had a lot of populist candidates who lost in the primaries, but effectively dragged the whole party over onto many populist issues in their congressional platform.
13:19
Unidentified speaker: So you had an outperformance of a lot of Republicans across the board, right?
13:24
Unidentified speaker: So the issue is are we seeing that same sort of polling airs this year?
13:31
Unidentified speaker: And how do you identify them?
13:34
Unidentified speaker: And this is a real problem.
13:36
Unidentified speaker: If it were easy, pollsters would have fixed it by now.
13:40
Unidentified speaker: But it's not because what you're effectively looking for is people who don't show up in polls.
13:47
Unidentified speaker: And so that goes back to that point that I made at the very beginning about these populist voters.
13:52
Unidentified speaker: They're low trust, but they vote when populist issues are on the table.
13:56
Unidentified speaker: What that low trust means is that they don't answer polls.
14:02
Unidentified speaker: So they don't show up in people's polls.
14:06
Unidentified speaker: And this, there's always been low trust individuals.
14:09
Unidentified speaker: So if you go back and look and talk to pollsters from 20 years ago, they'd say, oh yeah, we've got all sorts of low trust individuals.
14:15
Unidentified speaker: And it was usually minorities, right?
14:18
Unidentified speaker: It was blacks, Latinos in the United States who felt disaffected by society and didn't answer pollsters' phone calls.
14:28
Unidentified speaker: But it wasn't an issue because they also didn't tend to vote.
14:33
Unidentified speaker: And same thing, these same low-trust populist voters, a lot of them didn't vote in the past either.
14:40
Unidentified speaker: As I said about Brexit and about 2016, they came out of the woodwork when all of a sudden they felt there was something for them to vote on, somebody who was speaking specifically to their issues.
14:53
Unidentified speaker: So in the past, it had never resulted in bias.
14:56
Unidentified speaker: You missed them in the poll, that didn't matter because they didn't turn up to sh to vote.
15:00
Unidentified speaker: But all of a sudden, when you start putting populace on the menu, they start turning out to vote.
15:06
Unidentified speaker: And so that's really the the question.
15:08
Unidentified speaker: Do we face that same problem here?
15:10
Unidentified speaker: Do we see that there is a potential for polling bias this year?
15:18
Unidentified speaker: And if so, what is its magnitude?
15:21
Unidentified speaker: Is it big enough to affect the polls?
15:23
Unidentified speaker: So that's the sort of broad framework that I was approaching this with.
15:28
Unidentified speaker: And going back to your original uh question, yes, I came into this.
15:36
Unidentified speaker: My expectation was because populism, in my view, is on the menu, polling airs would be there.
15:42
Unidentified speaker: But I also thought, ah, the Democrats got a pretty big lead here.
15:46
Unidentified speaker: So they're probably not going to overcome that.
15:48
Unidentified speaker: So the House is almost certain to go, and the Senate is potentially going to go Democrat.
15:54
Unidentified speaker: That was my prior going into it.
15:56
Unidentified speaker: We'll come to why your priors are important in your process in a short while, but let's stay on this for a second because I'm curious as to whether you think that this low trust, populist issue voting cohort, is this now a phase or is this now a permanent feature of Western politics?
16:14
Unidentified speaker: And is it possible that the populist voter can also be, given the right circumstances, a left-leaning Democrat cohort?
16:27
Unidentified speaker: Populism right now is a right side issue because that's just the way the political spectrum has evolved.
16:34
Unidentified speaker: But one can imagine if we have right-leaning government incumbency for some considerable period of time, that the populism shifts to the left.
16:42
Unidentified speaker: So is this populism a phase or a permanent feature of Western politics?
16:46
Unidentified speaker: And how do you think it how big a factor could it be over the next sort of 15 to 20 years?
16:54
Unidentified speaker: So you have opened a whole can of worms there.
16:57
Unidentified speaker: Excellent.
16:57
Unidentified speaker: Hey, this is just getting you back for last Tuesday night.
17:00
Unidentified speaker: There you are.
17:01
Unidentified speaker: So look, it's really interesting, and I've had these debates, and this has come up multiple times, not only in client conversations, but also recently on in social media, uh, that people challenge me on this and say, well, populism is about, and I'm going to paraphrase here, and it's probably an overgeneralization of what people are saying.
17:25
Unidentified speaker: But basically, people seem to assume populism means the mass is asking for more free stuff.
17:33
Unidentified speaker: That is not populism, in my view.
17:35
Unidentified speaker: That is absolutely not populism, and all my research on populism says that's not what it is.
17:42
Unidentified speaker: Okay.
17:44
Unidentified speaker: That is leftism.
17:47
Unidentified speaker: That is social democracy.
17:49
Unidentified speaker: That is socialism.
17:51
Unidentified speaker: That is the masses saying, hey, we want to redistribute from the rich to the poor.
17:57
Unidentified speaker: That is a legitimate political viewpoint that is already out there in mainstream parties, right?
18:06
Unidentified speaker: You don't need to be a populist for that.
18:08
Unidentified speaker: Now, by the way, to exactly your question, populists can be both left-wing and right wing.
18:15
Unidentified speaker: What we have seen is that a lot of the populism that seems to have come to the fore over the last couple of decades in Western democracies has been what we might call right-wing populism.
18:32
Unidentified speaker: Although there's a really interesting characterization part of that, which is it does tend to have a very strong labor element to it.
18:39
Unidentified speaker: Like it is about working class and it is about protecting working class jobs over other things.
18:45
Unidentified speaker: So that's why key concerns have been immigration, right?
18:48
Unidentified speaker: Immigration effectively floods the job market with competition for these workers.
18:54
Unidentified speaker: And of course, trade also exposes these workers to competition from abroad.
19:00
Unidentified speaker: So that's why that's two key components of the whole right-wing populist agenda are actually labor issues, things that traditional labor parties or social democrat parties had often advocated in the past.
19:16
Unidentified speaker: What populism really comes down to, and so I did this study back in 2016, which I think we've talked about on some of our previous conversations, The Politics of Rage, where I went through populist party after populist party across 22 different democracies, what their manifestos actually said.
19:36
Unidentified speaker: And over and over and over, what you keep finding is that it comes down to a sense of political disenfranchisement.
19:46
Unidentified speaker: That the center left and the center right parties, so whether you have a right-wing leaning or a left-wing leaning, neither of those parties, neither of those center parties were offering you choices you cared about.
20:00
Unidentified speaker: So again, I always come back to immigration.
20:02
Unidentified speaker: Immigration is a great example of this.
20:04
Unidentified speaker: In almost every single Western democracy, if you look at polling, going back to the 1970s, significant majorities of people have asked for less immigration.
20:16
Unidentified speaker: Not no immigration, less immigration.
20:19
Unidentified speaker: And yet the opposite has happened.
20:22
Unidentified speaker: Both the center-left and center right parties have promoted more immigration, right?
20:29
Unidentified speaker: Same thing, less clear than immigration, but on the tariff issue.
20:34
Unidentified speaker: Generally, most of these populations have favored tariffs and favored industrial production.
20:40
Unidentified speaker: And yet, neoliberalism has dominated both the center left and center right parties.
20:46
Unidentified speaker: And so effectively, you had a whole set of voters out there who were saying, it doesn't matter which one these guys I vote for, none of them actually listen to me.
20:55
Unidentified speaker: So I've been disenfranchised.
20:57
Unidentified speaker: Yes, we have a democracy, but I don't actually have a choice of what I want.
21:02
Unidentified speaker: That to me is what defines populism, not whether you're left or right.
21:08
Unidentified speaker: And this I'm sorry for the long preamble on that, but it's really critical to answering your question about what we should expect going forward.
21:17
Unidentified speaker: The reason, the fundamental reason why these people are low trust is because government keeps failing them, right?
21:25
Unidentified speaker: So of course, if government is going to be unresponsive to them, they're going to be low trust.
21:34
Unidentified speaker: And so when you ask me, well, are we going to be facing the same problem in 20 years?
21:41
Unidentified speaker: My retort to you would be, Well, are governments going to start to listen to their populace again?
21:46
Unidentified speaker: Is democracy going to start to work again?
21:48
Unidentified speaker: Are governments actually going to do what their people tell them?
21:51
Unidentified speaker: If they do, I would expect trust to rebound and you'll have less of a problem with this.
21:55
Unidentified speaker: If they don't, then yeah, this is going to continue to be a problem.
21:59
Unidentified speaker: But then we get to the second question you posed, which is, well, does this have to be a right-wing populism?
22:07
Unidentified speaker: And that's potentially an issue here.
22:11
Unidentified speaker: And in fact, it's one of the risks I raised to my own analysis of this election is that we have seen a surge in left-wing populism, and not just in the U.S.
22:23
Unidentified speaker: So in the United States, we've seen the Democratic Socialist Party effectively using the Democratic Party as a platform in the same way Donald Trump's MAGA movement has used the Republican Party as a platform.
22:36
Unidentified speaker: And so I would characterize the DSA as a left-wing populist movement.
22:43
Unidentified speaker: They are definitely feeling like they have been disenfranchised.
22:48
Unidentified speaker: Now, the difference is that the people who have been associated with the left-wing populism, at least in the United States, also happen to come from what you might call high trust tribes.
23:03
Unidentified speaker: They tend to be college-educated, upper middle class people who actually have high trust in society.
23:11
Unidentified speaker: And so there is an argument that they are overrepresented in the polls, and that's what's giving you this leftward tilt in the polls.
23:19
Unidentified speaker: But I don't know that, right?
23:22
Unidentified speaker: We're gonna have to see.
23:32
Unidentified speaker: We actually do see evidence that whether it is the French left under Mélenchon or the German left under De Linka and was it BZD?
23:44
Unidentified speaker: I can't remember what the uh other parties, they actually their signs that they too have been punching above their polling weight in actual elections.
23:54
Unidentified speaker: So I think your question is very valid in terms of whether it'll start to affect polls more in terms of undercounting left-wing votes.
24:04
Unidentified speaker: But it's so far it's not clear in the United States.
24:07
Unidentified speaker: We haven't seen any clear evidence of that.
24:09
Unidentified speaker: So it's a definite risk for this election, but it's not one that I think we can count on.
24:14
Unidentified speaker: You mentioned this idea of priors earlier on, and this leads us into your process.
24:19
Unidentified speaker: And I'm gonna get you to lay that process out if I can in a second.
24:23
Unidentified speaker: But just as a kind of setup to that, as you said earlier, you went into this kind of expecting the Republicans to lose both chambers.
24:30
Unidentified speaker: Probably that was the likelihood.
24:32
Unidentified speaker: At least the House and maybe and maybe the Senate, yeah.
24:35
Unidentified speaker: But your analysis, which we're gonna get to now, without wanting to issue any spoiler alerts, shifted that weight.
24:43
Unidentified speaker: And again, I think it's probably important that we state here that you were actually a Bush administration appointee at one point.
24:49
Unidentified speaker: So people will want to frame that as you having a Republican bias.
24:53
Unidentified speaker: But obviously, when it comes down to the data, I know that you put that aside and you're it's all about the data.
25:00
Unidentified speaker: You're not looking for confirmation bias, you're looking for the answers.
25:03
Unidentified speaker: Was there a moment in your process where this turned, or was it more of a case of is it more of the same from those previous experiences you had in in looking through the data and coming up with surprising outcomes?
25:18
Unidentified speaker: I would say it came in in two bumps.
25:22
Unidentified speaker: So by the way, it's probably worth defining some terms here because yeah, if you could hold that that idea about bumps in your head, and then let's go through your process and the terms, and you can drop those in as we come to them along the way.
25:37
Unidentified speaker: So I use this term, and you just used it in your questionnaire prior.
25:41
Unidentified speaker: So it's probably useful for people who haven't studied Bayesian statistics to under understand this.
25:48
Unidentified speaker: So, in what you might call classical statistics, you have a hypothesis, you go out and test it against the data.
25:57
Unidentified speaker: And if the data in a statistical sense show something very different, you reject your hypothesis.
26:05
Unidentified speaker: So it's always about rejecting your initial hypothesis in Bayesian statistics, and I'm bastardizing this a bit, but this is roughly the difference, is you start with a view, and then you say, Do the data as they come in, each new piece of data I get in, update my view in a different way, or does it cause me to just stay where I was?
26:32
Unidentified speaker: And so my view, as I said, my prior, that's your view that you enter with was that we did have polling errors, but I wasn't sure whether they would be big enough to overcome what seemed like a big democratic advantage.
26:51
Unidentified speaker: Okay, and so then you go out and look at the data and you're looking for evidence that shifts your prior and creates a new view, your posterior view.
27:03
Unidentified speaker: That's what it is.
27:04
Unidentified speaker: So prior and posterior.
27:07
Unidentified speaker: And the reality is that there is a statistical significance to that too.
27:13
Unidentified speaker: But let me go back to the problem that we're trying to solve here.
27:17
Unidentified speaker: So I use a Bayesian process, but I wouldn't call it Bayesian statistics, if that makes any sense.
27:23
Unidentified speaker: Because remember the issue is we have people who don't show up in polls.
27:31
Unidentified speaker: Now, they might there might be a few who sneak into polls, we don't know, right?
27:37
Unidentified speaker: And so maybe it's just that they're under-sampled in in polls.
27:42
Unidentified speaker: But let's suppose in the extreme case, they don't actually even show up.
27:47
Unidentified speaker: Well, how would I ever statistically test whether something is in the poll or not if I know it's not in the poll?
27:59
Unidentified speaker: It doesn't exist, right?
28:01
Unidentified speaker: You can't do that.
28:03
Unidentified speaker: So the onus on proving statistical significance that they exist is extreme here.
28:12
Unidentified speaker: You're almost never going to get that.
28:16
Unidentified speaker: And so the only way you ever see this is ex post.
28:20
Unidentified speaker: After the election, you say, Oh my gosh, actually, there were way more populist voters than I thought.
28:28
Unidentified speaker: Okay, well, that tells you that indeed you were missing all sorts of people, but you don't know that uh beforehand.
28:35
Unidentified speaker: So, what I'm looking for in the polls is any kind of sign of that biases that existed in the poll still exist after the pollsters have adjusted those polls.
28:54
Unidentified speaker: And why would they adjust it?
28:56
Unidentified speaker: This is very straightforward because there are some obvious biases that we know exist in polls.
29:03
Unidentified speaker: So this is obviously a stereotype, but your some 70-year-old little old white retiree woman sitting by the phone is much more likely to answer a pollster's call and complete their survey than a 25-year-old Latino man.
29:24
Unidentified speaker: Okay, this is just a statistical fact, by the way, but it also fits everybody's, I think, priors.
29:31
Unidentified speaker: I was gonna say it wouldn't come as a surprise to many people, I don't think.
29:35
Unidentified speaker: Exactly.
29:36
Unidentified speaker: So what happens is whether you're doing a telephone poll or an online poll, you're getting a bias sample coming in.
29:44
Unidentified speaker: The pollster knows that you're gonna get way too many little old white women and not enough young Latino men.
29:51
Unidentified speaker: Okay.
29:52
Unidentified speaker: So what you do is you adjust the weights of those.
29:56
Unidentified speaker: So if you know that little old white women are 10% of the population, but you get 40% of your sample from little old white women, well, guess what?
30:07
Unidentified speaker: You divide their result by four.
30:10
Unidentified speaker: And suppose that you only get 5% Latino men, but you know that they're supposed to be 10%, then you multiply the results you get by two.
30:21
Unidentified speaker: And voila, you have now a supposedly unbiased poll.
30:26
Unidentified speaker: And to deal with this polling bias problem that we've talked about, pollsters have gone through a variety of different contortions to try and tease out whom they're missing.
30:39
Unidentified speaker: So, one of the things that they noticed pretty early on is that oh gosh, we seem to be missing a lot of working class white men.
30:48
Unidentified speaker: Those seem to be the people who have voted for Trump and we were missing.
30:52
Unidentified speaker: Now, it turns out, and one of the guys who's been like the best on this is a Democratic data scientist pollster.
31:00
Unidentified speaker: He was actually Barack Obama's chief of data science in the 2012 election at the age of 19.
31:06
Unidentified speaker: The guy's pretty off the charts.
31:07
Unidentified speaker: Uh, his name is David Shore, and he was one of the first people to highlight this trust issue.
31:11
Unidentified speaker: And he actually goes through the data ex post, he found trust surveys and things like this, and he shows that those white working class men, if they were high trust, voted overwhelmingly for Hillary Clinton.
31:25
Unidentified speaker: Whereas people with PhDs were low trust voted overwhelmingly for Donald Trump.
31:32
Unidentified speaker: So actually, it's not working class versus educated.
31:36
Unidentified speaker: It goes back to that trust issue.
31:39
Unidentified speaker: And this is the problem is that I can slice and dice the data on all sorts of demographics, what your income is, what your religion is, whether you're rural versus urban, whether you're black versus white, all of these different things, but none of them get to this trust issue.
31:57
Unidentified speaker: And how do I even find out what your trust is if you don't even answer my poll?
32:04
Unidentified speaker: If you're not answering my poll, I can't ask you the question.
32:06
Unidentified speaker: Hey, by the way, are you high trust or low trust?
32:09
Unidentified speaker: So this is the problem.
32:11
Unidentified speaker: But we do know that from past, as I said, there seems to be a correlation with white working class men.
32:20
Unidentified speaker: There's also a correlation, by the way, with black and Latino men in terms of their Trump votes versus what they poll as.
32:29
Unidentified speaker: We've seen this in various elections.
32:32
Unidentified speaker: And so we know that there may be larger numbers of these low trust voters hiding in those groups.
32:40
Unidentified speaker: That they can be in any group, but there might be larger ones in those.
32:44
Unidentified speaker: So then you can go through and start to look and see: well, hey, do I see a relationship between polling airs and the input sample, that is, before they do any statistical adjustments to it in pollsters, polls with the final results in past years, and look for whether those correlations or those over or under representation in their samples exist again this year.
33:13
Unidentified speaker: And what you find is that indeed there is a very strong correlation between past polling errors and represent over or under-representation across these different demographics, and that those same under and over-representation problems occur this year.
33:29
Unidentified speaker: Now, there's some additional wrinkles, which I get into in the piece, this gets a little bit more complex, but if you think about it, the polsters aren't stupid.
33:40
Unidentified speaker: They say, Oh, okay, we know that these particular groups seem to have a bigger problem.
33:46
Unidentified speaker: So what we're really going to do is we're going to try and massively oversample it.
33:51
Unidentified speaker: So even though we only need out of a hundred, say, 10 Latino men, we're going to sample 50 just to really nail that down.
34:03
Unidentified speaker: Well, if only the high trust people answer, the fact that you increased your sample size to 50 didn't change it.
34:13
Unidentified speaker: It just actually, if anything moved this.
34:16
Unidentified speaker: Yes, it made your bias problem worse.
34:20
Unidentified speaker: This is one of the things that I was wrestling with after our conversation and then reading the piece, right?
34:25
Unidentified speaker: Because if this idea of that reweighting makes things worse rather than better, then if you identify the unrepresented populist group and you upweight them, you amplify the error.
34:38
Unidentified speaker: So if the fix is the actual problem, right, is there any kind of methodological route out of this for the pollsters?
34:47
Unidentified speaker: Or are we just in a period where even if temporarily the industry is actually structurally broken for the time being?
34:55
Unidentified speaker: So I've thought a lot about that question.
34:57
Unidentified speaker: I don't see a good solution.
35:00
Unidentified speaker: Right.
35:00
Unidentified speaker: Because what I did is considered highly unkosher.
35:06
Unidentified speaker: And you read the piece, you'll know I throw lots of caveats down.
35:11
Unidentified speaker: I say, hey, look, nothing I have here is statistically significant.
35:16
Unidentified speaker: This is not proper Bayesian statistics.
35:19
Unidentified speaker: This is I come in with a view and then I look.
35:22
Unidentified speaker: Does anything in the data contradict my view?
35:27
Unidentified speaker: And nothing did, by the way.
35:30
Unidentified speaker: So I keep my view.
35:32
Unidentified speaker: That's not proper Bayesian statistics.
35:35
Unidentified speaker: Like this would not pass any peer review process, right?
35:40
Unidentified speaker: Right.
35:41
Unidentified speaker: And pollsters are caught in the situation where they're using statistical analysis.
35:47
Unidentified speaker: That's the basis of their being.
35:49
Unidentified speaker: So they kind of have to stick to it.
35:52
Unidentified speaker: So they're not going to stick their neck out there and say, hey, you know what?
35:55
Unidentified speaker: Our polls have a problem and it's just big.
35:59
Unidentified speaker: That's not a big selling point if you're a pollster.
36:02
Unidentified speaker: No, this is true.
36:03
Unidentified speaker: But that's not my job.
36:05
Unidentified speaker: Going back to what you said at the beginning, where you rightly identified my past Republican service uh as a potential bias here.
36:15
Unidentified speaker: Today, my clients pay me to get the right answer.
36:18
Unidentified speaker: I don't give a damn, right?
36:20
Unidentified speaker: So, what I do, and not only do they pay me to get the right answer, they want me to give them an answer.
36:29
Unidentified speaker: They are not going to pay me to say, yeah, nothing's statistically significant, so I don't really have a view here.
36:36
Unidentified speaker: So then I have to actually make judgments.
36:40
Unidentified speaker: And this is my judgment.
36:43
Unidentified speaker: But when I go through and look, the fact that nothing contradicts this view that polling analysis exists.
36:52
Unidentified speaker: And then I do some much more sophisticated analysis, really detailed analysis, where I actually look at how some of the composition of these voters is correlated with polling errors across elections through time, and what the composition of polls today looks most like.
37:11
Unidentified speaker: And guess what?
37:13
Unidentified speaker: Over and over again, it keeps coming up with the years that had some of the biggest polling errors.
37:19
Unidentified speaker: And in particular, by the way, that 2014 midterm that I highlighted as looking the demographics of the underlying poll look most like that, not like 2018, not like 2022.
37:35
Unidentified speaker: And then I go through and I look at cross-state examples about how polling errors seem to be magnified by the percentage of the population that it normally votes Republican and different other Democrat factors.
37:50
Unidentified speaker: I put all these things together, and what I keep coming up with is A, there's almost certainly significant polling errors this year.
37:59
Unidentified speaker: And B, they are large enough to swing the election for the Republicans.
38:05
Unidentified speaker: Now, part of that also has to do with what elections really matter.
38:12
Unidentified speaker: So there's two, you need to think separately about the House versus the Senate in this.
38:17
Unidentified speaker: And the Senate's pretty straightforward.
38:19
Unidentified speaker: Most of these seats they're talking about potentially flipping to Republicans.
38:24
Unidentified speaker: And oh, by the way, some of the ones that Democrats are supposed to hold, like Michigan and Minnesota, are within one or two percentage points in the existing polling.
38:35
Unidentified speaker: So if I find that those states have polling bias that might be on the order of three to four percentage points, you can see that the Senate ain't flipping, right?
38:46
Unidentified speaker: Right away.
38:47
Unidentified speaker: And in fact, actually, my baseline result, even though yeah, my baseline result is that the Republicans may actually pick up a seat in the Senate.
39:00
Unidentified speaker: They may lose North Carolina, but pick up both Michigan and Minnesota and thus actually end up with a 54 rather than a 53 seat majority.
39:11
Unidentified speaker: Okay.
39:12
Unidentified speaker: So that's the Senate.
39:15
Unidentified speaker: And again, that's highly dependent on how big is that polling air.
39:19
Unidentified speaker: But if my analysis suggests that roughly on average across most of these races, it's going to be on the order of three to four percentage points, that that takes you there.
39:27
Unidentified speaker: The house is a different kettle of fish.
39:29
Unidentified speaker: One of the things, first of all, think about the generic ballot.
39:33
Unidentified speaker: One, recall what I said about generic ballots are different from asking about actual people.
39:41
Unidentified speaker: And historically, what we found is that even before polling bias became a thing, the generic ballot tended to overstate the popular vote in the house races for Democrats by about two percentage points.
39:54
Unidentified speaker: So take that off the top.
39:56
Unidentified speaker: Then you look at the fact that because of gerrymandering, this practice of shaping districts around specific parties in different areas to get a specific outcome.
40:11
Unidentified speaker: You have places in California and New York and places like that that are massively like there's no question this district is going to go for Democrats or in other places that it's going to go for Republicans.
40:26
Unidentified speaker: They're going to have a 70 or 80 percent to 30 or 20% majority.
40:33
Unidentified speaker: What really matters is close races.
40:36
Unidentified speaker: And if you look at close races, we don't have very good data on that because as I said, they don't take very many polls of this.
40:43
Unidentified speaker: But from some of the few organizations like the Cook Political Report that do poll these things, there's about 75 house races that really are potentially in contention.
40:56
Unidentified speaker: And if you look at those, in September, they had these guys at a two percentage point margin on average, not seven to nine, two percentage points.
41:06
Unidentified speaker: Okay.
41:07
Unidentified speaker: Well, now we're in polling air territory, right?
41:10
Unidentified speaker: Right, right, exactly.
41:12
Unidentified speaker: Yeah.
41:13
Unidentified speaker: And so again, look, I'm making some big assumptions here about what the actual distribution of those look like and using some fancy statistics.
41:24
Unidentified speaker: And I look at this and I run it through a variety of different assumptions, including, by the way, that the polls have increased another percentage point since then, which is what the generic polls did since Cook did their original polling in September.
41:40
Unidentified speaker: So let's bias it against, let's just say the Democrats are even further ahead.
41:43
Unidentified speaker: It's three points, not two points.
41:46
Unidentified speaker: Even with that, what I find is that with normal size polling airs, actually, shockingly, and you've read my piece, so you know I don't even lean in and say, I think they're gonna keep the house.
42:00
Unidentified speaker: I say, I think they got a good chance of keeping the house, but actually the numbers come out, they keep the house.
42:06
Unidentified speaker: One of the few times in history they actually gain the president's party actually gained seats in the house and potentially a record number.
42:16
Unidentified speaker: Let people understand how rare that is, Marvin, because when you gave me the data, I was shocked at how rare it is for an incumbent to actually gain in the midterms.
42:24
Unidentified speaker: Yeah, so it's only actually happened four times since the Civil War.
42:28
Unidentified speaker: Pre-Civil War, the US was a little bit different, parties were a little bit less farm.
42:31
Unidentified speaker: So 1902, you should probably toss that one out because that was actually an expansion of the house.
42:36
Unidentified speaker: So you added, I think it was a hundred new seats to the house, and so yeah, the the Republicans got some.
42:43
Unidentified speaker: They didn't actually come out in percentage terms, they just gained overall.
42:46
Unidentified speaker: So that's one of those.
42:48
Unidentified speaker: Then you have 1934, when in FDR's first year of the New Deal, you have 1998, when the Republicans were trying to prosecute Bill Clinton for the Monica Lewinsky affair, and voters said, Hey, we care about other issues, not about your damn partisan politics.
43:13
Unidentified speaker: And then 2002, right after the 9-11, George Bush did that.
43:18
Unidentified speaker: Those are the only times the incumbent president's party has gained seats, and the biggest of those was nine seats.
43:27
Unidentified speaker: That was that 1902 one.
43:29
Unidentified speaker: So when I do these numbers and I get numbers on the order of nine to 12 seats that Republicans could actually pick up.
43:39
Unidentified speaker: This is what I'm saying.
43:40
Unidentified speaker: I it's hard for me to discount that, even though it's not statistically significant, right?
43:48
Unidentified speaker: So that's where I come out on this.
43:51
Unidentified speaker: A fanalist absolutely fascinates you.
43:53
Unidentified speaker: And you are explicit throughout the piece that none of this is, as you just said there, statistically significant, right?
43:58
Unidentified speaker: Your your regressions come back insignificant, and the the whole framework is unable to establish bias in a classical sense, right?
44:10
Unidentified speaker: But it still leads you to make what, in my reading over the last several months, is one of the loudest out-of-consensus calls I've seen, which is great.
44:19
Unidentified speaker: But I'm curious how you how do you decide when consistency with the model is enough to act on?
44:26
Unidentified speaker: Because for a lot of people, the data that you're seeing will come back and they'll go, well, this can't be right.
44:33
Unidentified speaker: But you're not contrarian for the sake of it.
44:36
Unidentified speaker: You've been out of consensus at points where it mattered, and you've been right.
44:42
Unidentified speaker: So, how do you guard against that whole, well, I've been out of consensus, I've been right.
44:48
Unidentified speaker: This comes back out of consensus, I'm gonna go with it again because it's been right before, not necessarily because this particular set of numbers really scream to me that this is a great call to make.
45:02
Unidentified speaker: Does that make sense?
45:03
Unidentified speaker: It's a bit of a roundabout.
45:04
Unidentified speaker: No, it makes sense.
45:05
Unidentified speaker: And the simple answer is just what everybody in our industry does, right?
45:12
Unidentified speaker: There are people out there who are quantitative investors or quantitative research shops, and they exclusively rely on statistically significant models that continuously, on average, produce a winner, right?
45:29
Unidentified speaker: Or if they don't, they go out of business.
45:30
Unidentified speaker: Whatever.
45:31
Unidentified speaker: Everybody else, like anybody else who's using a subjective metric, is making a judgment.
45:39
Unidentified speaker: It's saying, look, I don't have enough data to prove this, but the data are consistent enough with my judgment and experience to suggest that this is what's going to happen.
45:51
Unidentified speaker: And that's all that's happening here, is that I don't see my view on populism not going away.
45:58
Unidentified speaker: Still very strong, still an important driver of elections in Western democracies.
46:06
Unidentified speaker: Two, have pollsters figured out a way to deal with this.
46:09
Unidentified speaker: We just discussed it's basically impossible to deal with if trust is really the underlying variable, which uh I firmly believe it is.
46:19
Unidentified speaker: So as a result, I know for a fact the data will never say that this is going to be a bias problem.
46:31
Unidentified speaker: I can only say, are the data consistent with it with their potentially being a bias problem?
46:39
Unidentified speaker: And if I think that A, populism is around, and B, that pollsters have not been able to solve this problem, then it leads me to an answer.
46:48
Unidentified speaker: And then it's really about sizing it.
46:51
Unidentified speaker: And that's where statistics comes in.
46:54
Unidentified speaker: And again, there's enough variability here that some of those things in terms of the margins are statistically significant, but most of them are not.
47:03
Unidentified speaker: So that's why I used, if you recall in the piece, a wide variety of different measures of what could potentially drive this.
47:10
Unidentified speaker: And by the way, the one that is the least approved by the data is the one that gives the Democrats the best result.
47:21
Unidentified speaker: Right, right.
47:23
Unidentified speaker: So yeah, that's where I come from.
47:26
Unidentified speaker: It really is judgment.
47:28
Unidentified speaker: And at some point, you have to recognize you're not going to get statistical significance and you still have to make a call.
47:35
Unidentified speaker: And so here's my call.
47:37
Unidentified speaker: There's two other pieces of this that I want to dig into, but before we get to those, let's just be very clear with people what that call is.
47:45
Unidentified speaker: Because we've talked about the method and we've talked about the surprising result, but let's talk about what the actual call is that you believe may end up coming to pass.
47:54
Unidentified speaker: So, first and most important is I can largely preclude the idea that we are going to have a landslide for the Democrats.
48:04
Unidentified speaker: Okay.
48:05
Unidentified speaker: So I don't think that's happening.
48:08
Unidentified speaker: My base case is that the Republicans will hold the Senate.
48:14
Unidentified speaker: Maybe barely, but they will hold it.
48:18
Unidentified speaker: And that they have a decent chance of holding the house.
48:23
Unidentified speaker: Although, as I said, my prior going in is that they would lose the house.
48:30
Unidentified speaker: And so, even given my analysis, it's hard for me to really stomp my foot that they're actually gonna take the house.
48:39
Unidentified speaker: That said, I do think there is a significant, and I'd define significant as say a 20 to 30 percent chance, right?
48:47
Unidentified speaker: So, not small, that actually the Republicans do not only hold both houses, but actually increase their leads in both.
48:57
Unidentified speaker: And that, if they do that, this has massive consequences, aside from the fact that it'll shock everybody, but it has really big consequences.
49:08
Unidentified speaker: This brings us to the things I wanted to get into because that because again, and you lay this out, and at the end of this, I'm hoping everyone is going to actually sit and read this piece because it's just such a thought-provoking and brilliantly argued piece.
49:21
Unidentified speaker: Let's get into those consequences because again, when I read through them, I was sitting there going, man, I didn't think of that.
49:29
Unidentified speaker: I didn't think of that.
49:30
Unidentified speaker: So just walk us through.
49:32
Unidentified speaker: Let's say that your outlier prediction is right and they gain in both.
49:37
Unidentified speaker: Let's talk about what that potentially unleashes in the aftermath.
49:41
Unidentified speaker: Well, I think the most important thing is that Donald Trump goes from being a lame duck the final two years of his final term, we assume.
49:53
Unidentified speaker: Yeah, yeah, yeah.
49:54
Unidentified speaker: You have to put that little asterisk there.
49:56
Unidentified speaker: I know it's crazy, right?
49:57
Unidentified speaker: Oh, it's a little asterisk there with that man, unfortunately.
50:01
Unidentified speaker: So he goes into the final two years of his uh likely final term.
50:08
Unidentified speaker: Usually that means the president is a lame duck.
50:11
Unidentified speaker: They have lost most of their power because nobody's really that concerned about him.
50:16
Unidentified speaker: Quite the opposite here.
50:17
Unidentified speaker: Donald Trump could be at his most powerful in the following two years if this happens.
50:24
Unidentified speaker: Because remember how this happened.
50:27
Unidentified speaker: He specifically went out in the primaries and took out people who disagreed with him.
50:35
Unidentified speaker: At the threat of everybody saying, you are going to lose our majorities in both houses because these people are unelectable.
50:44
Unidentified speaker: And the clearest example of that is North Carolina.
50:47
Unidentified speaker: Even under with my polling bias estimates, North Carolina is, by the way, there's a small caveat to that based on incoming early votes.
50:55
Unidentified speaker: But if we go by the polls and my polling heirs, they're gonna lose North Carolina, right?
51:01
Unidentified speaker: And it's because he's Tom Tillish, you're out of here, right?
51:06
Unidentified speaker: And in Texas, Texas is considered competitive because they got rid of John Cornyon and brought in a populist who has some unsavory parts or some questionable parts of his past and has turned off a lot of other voters.
51:21
Unidentified speaker: So, despite that, if they win, who is the Republican in either chamber who is going to stand up to this man because they know he's gonna be in their districts campaigning against them in the next election, whether he's on the ticket or not.
51:40
Unidentified speaker: One of our prior conversations, we talked about my La Casa Nostra Americana framework for thinking about American foreign policy under Donald Trump.
51:50
Unidentified speaker: Apply that domestically, Don Trump is gonna take out any capo who tries to rise up against him, right?
51:59
Unidentified speaker: And so you better do what the Don says, right?
52:04
Unidentified speaker: So so he is becomes extraordinarily powerful, in my view.
52:10
Unidentified speaker: And I think whether you think that I am Republican or pro-Trump or anti-Trump or whatever, or you have Trump derangement syndrome, I think we can all agree he will use it.
52:25
Unidentified speaker: Yeah.
52:27
Unidentified speaker: Yeah, I think that's a fairly safe bit.
52:29
Unidentified speaker: Well, but again, when you talk about what he might use that for, once again, I didn't see this stuff coming.
52:36
Unidentified speaker: The Andrew Jackson comparisons that are diametrically opposed to things like the big beautiful bill.
52:42
Unidentified speaker: So walk us through that because again, we all know Trump is a man of immense mystery and intrigue and contradictions.
52:51
Unidentified speaker: He's a fascinating character.
52:53
Unidentified speaker: If you can study the character and get past the man from whichever side, you either love him or hate him, if you can find a way to put that side and objectively study him.
53:02
Unidentified speaker: He is a fascinating character.
53:05
Unidentified speaker: But again, what you talk about what lies beneath all the bluster, again, I found extraordinary at first blush, but then the more I read about it and thought about it, I was like, man, again, you've got another point here.
53:21
Unidentified speaker: So so talk a little bit about that.
53:23
Unidentified speaker: To exactly your point, he is a fascinating figure.
53:27
Unidentified speaker: Whatever you think of him, like you really, whether you're a Trump lover, hater, or just agnostic on him.
53:35
Unidentified speaker: You can't be, right?
53:35
Unidentified speaker: You can't be agnostic.
53:36
Unidentified speaker: It's impossible.
53:37
Unidentified speaker: Yeah, yeah, it is kind of hard to be agnostic, but I throw that out there.
53:40
Unidentified speaker: But whether you're a lover or a hater, you gotta throw that aside and just as an intellectual, just enjoy wow, the character development, the political uh savvy, or in other ways, craziness of this man.
53:59
Unidentified speaker: There is so much complexity to understand here.
54:02
Unidentified speaker: And one of those complexities we've talked about before, that he very consciously does try to distract people with misinformation or to divert their attention elsewhere.
54:18
Unidentified speaker: There's also, I cannot rule out a lot of the stuff he says and does.
54:22
Unidentified speaker: I'm just like, no, that's just plain erratic.
54:26
Unidentified speaker: I don't see any method to that mad madness.
54:28
Unidentified speaker: That is just madness.
54:30
Unidentified speaker: Madness.
54:30
Unidentified speaker: Yes.
54:32
Unidentified speaker: But if you slice through all of that and look at it, I think there are two key things that show up again and again, particularly in the second term.
54:45
Unidentified speaker: One is that his instincts and the way he almost always ends up going, and some of this might not just be him, a lot of it might be he's shaped by the political movement that he's created, right?
55:00
Unidentified speaker: Which has deep roots in American history.
55:03
Unidentified speaker: This is very clearly Jacksonian populism, back to Andrew Jackson.
55:09
Unidentified speaker: And so one of my new rules for life is, or not rules for life.
55:16
Unidentified speaker: I don't apply this in my personal life, but rules for understanding Trump is WWJD, but not what would Jesus do?
55:25
Unidentified speaker: It's what would Jackson do?
55:28
Unidentified speaker: Because whatever Jackson would do is often the right answer for what Trump will do when you're confused.
55:35
Unidentified speaker: Okay.
55:36
Unidentified speaker: So that's one rule, and I'll come back to how that applies to some of these things.
55:40
Unidentified speaker: The other is that, particularly in his second term, I think we talked about this in one of our previous conversations, that this is a man who wasn't interested in being president the first time.
55:50
Unidentified speaker: It was just an ego kick.
55:52
Unidentified speaker: And then he starts to get interested by the end of his presidency, and then he gets kicked out of office, and he is pissed.
56:00
Unidentified speaker: He is really mad.
56:02
Unidentified speaker: And he's very competitive and he wants to come back.
56:05
Unidentified speaker: And credit to him.
56:07
Unidentified speaker: He figured out how to do it and he did come back.
56:10
Unidentified speaker: And now this is a big mistake I think a lot of people make.
56:13
Unidentified speaker: They think that he is about the short term and about, oh, I'm going to do this to get this win or whatever.
56:20
Unidentified speaker: No, this man is all about he wants himself to go down in the history books as a Jackson, as an FDR, as a Lincoln, one of the big name presidents in history.
56:33
Unidentified speaker: And the only way you do that is by doing really big things.
56:36
Unidentified speaker: So I think those are two fundamental rules you need to use whenever you evaluate Donald Trump.
56:43
Unidentified speaker: And it doesn't mean that they'll always be right, but I think they're very helpful.
56:47
Unidentified speaker: And that gets me to two of the big predictions I have that I think are very counter-consensus is one, with the leeway and time that he will now be afforded by a Republican Congress that he is all powerful in front of, is he is going to bring the Iran war to a conclusion in terms of forcing their submission.
57:14
Unidentified speaker: Because ultimately, this is really about the blockade.
57:17
Unidentified speaker: And the blockade is working.
57:18
Unidentified speaker: And actually, the blockade itself is not very costly to the US.
57:23
Unidentified speaker: This is three and four decimal places to the right on the federal budget to continue the blockade, and it's being very effective.
57:33
Unidentified speaker: So then it's just a waiting game.
57:35
Unidentified speaker: And if he has the Congress, he has the waiting time.
57:37
Unidentified speaker: So I think that's one.
57:39
Unidentified speaker: The second is, and this one I think is much more controversial because everybody thinks he's a debt guy, he likes big debt.
57:47
Unidentified speaker: And even you just said something, you compared it.
57:50
Unidentified speaker: Well, this is against uh Oba, the one big beautiful bill.
57:54
Unidentified speaker: And I'm gonna correct you there because that's not true.
57:57
Unidentified speaker: Despite what everybody said, real spending in the United States under Oba was cut.
58:08
Unidentified speaker: Nominal spending did go up because inflation went up, but real spending for one of three times in the last 60 years that we have clear data on this was cut.
58:21
Unidentified speaker: Okay, that is a rarity.
58:24
Unidentified speaker: You can also look at the primary balance in the United States.
58:26
Unidentified speaker: It was it has not increased under Donald Trump yet.
58:29
Unidentified speaker: Now, if you look for in future years, it does under Oba, but my presumption has always been Oba was just to get them through here.
58:38
Unidentified speaker: And if he gets a Republican Congress, then he'll do something else.
58:42
Unidentified speaker: And so this comes back to what would Jackson do and also his legacy.
58:48
Unidentified speaker: It is absolutely clear that the U.S.
58:50
Unidentified speaker: is on a definitely unsustainable path.
58:54
Unidentified speaker: Now, its unsustainable path is has more runway than, say, France's does, but it is an unsustainable path.
59:04
Unidentified speaker: And the next president for sure is going to have to deal with the fact that the Social Security Trust Fund goes into bankruptcy in the next term, right?
59:15
Unidentified speaker: So, one, reflect back on there is only one president in U.S.
59:20
Unidentified speaker: history who has ever not just balanced the budget, but actually paid down all of U.S.
59:25
Unidentified speaker: debt.
59:26
Unidentified speaker: Only one.
59:32
Unidentified speaker: If you look at Donald Trump's voters, what are the two biggest things that they wanted in terms of, if you poll them, what were their two biggest priorities?
59:43
Unidentified speaker: Number one was federal spending, number two was inflation.
59:47
Unidentified speaker: When we've talked about this before, on the inflation front, nobody expected him to appoint a hawk.
59:53
Unidentified speaker: Nobody expected him to appoint Kevin Walsh.
59:56
Unidentified speaker: That was one of my big calls last year.
59:58
Unidentified speaker: He was going to appoint the Kevin Warsh, and Kevin Warsh would turn out to be a hawk, right?
1:00:03
Unidentified speaker: So far, he's mostly proving me correct.
1:00:07
Unidentified speaker: We'll see.
1:00:07
Unidentified speaker: I still have questions, but he's mostly proving me correct.
1:00:11
Unidentified speaker: Same thing.
1:00:12
Unidentified speaker: I think we're going to be surprised on the same thing on the fiscal front.
1:00:15
Unidentified speaker: His voters expect and want fiscal consolidation.
1:00:19
Unidentified speaker: And this is his path to being in the history books, is that he actually gets us there.
1:00:27
Unidentified speaker: And so I do think that you're going to see a huge new focus on fiscal consolidation in the second term, which, especially given the backdrop of what we're talking about taking place in Europe with France and other countries, is going to increasingly look attractive to the American electorate.
1:00:47
Unidentified speaker: People are going to say, well, we don't want that.
1:00:50
Unidentified speaker: And that's when tough choices start to be made.
1:00:54
Unidentified speaker: Does the American public think that?
1:00:56
Unidentified speaker: Because I think everybody everywhere in the West can point to Froud and say, I don't want that.
1:01:01
Unidentified speaker: But then when somebody says to them, okay, well, here's how you avoid it, they go, Well, hang on a second.
1:01:07
Unidentified speaker: Hang on a second.
1:01:08
Unidentified speaker: I don't want to give up all the good stuff.
1:01:09
Unidentified speaker: I don't want to give up the goodies.
1:01:11
Unidentified speaker: I just don't want that.
1:01:12
Unidentified speaker: Find me another way of doing it where we don't have to be fiscally conservative and we don't have to rein in the spending.
1:01:18
Unidentified speaker: That's the thing.
1:01:18
Unidentified speaker: Because when I read the Jacksonian stuff after I got past my holy cow moment, I'm like, yeah, actually that makes it make sense, right?
1:01:26
Unidentified speaker: It makes sense.
1:01:27
Unidentified speaker: I get it.
1:01:28
Unidentified speaker: But where does that consolidation come from, right?
1:01:30
Unidentified speaker: And by raising tariffs and, God forbid, taxes on the wealthy, because a lot of his supporters, his real hardcore financial supporters, are not going to like that, and cutting entitlements, which is essentially the only real way to do this, practically speaking.
1:01:48
Unidentified speaker: Does that survive contact with a Congress that's looking already at 2028?
1:01:53
Unidentified speaker: I don't know.
1:01:54
Unidentified speaker: I think that that is the the the biggest issue, right?
1:01:58
Unidentified speaker: So now I will say that there are very clear hints in his language and in other members of his administration.
1:02:08
Unidentified speaker: So in the early months of this term, we go back to the first half of last year, Howard Lutnick in particular kept repeating this.
1:02:18
Unidentified speaker: And I thought that that was interesting.
1:02:20
Unidentified speaker: But Howard Lutnick, you can discount a lot of stuff he says, whatever.
1:02:24
Unidentified speaker: But then Trump himself said it, right?
1:02:27
Unidentified speaker: And they talked about raising tax rates on the highest earners to finance a tax cut for the bottom 90%.
1:02:41
Unidentified speaker: And so I went out and I just did some back of the envelope calculations.
1:02:45
Unidentified speaker: And actually, yeah, you could finance with reasonably modest tax rises on the top 10%, you could finance removing income taxes on the bottom 90%.
1:02:59
Unidentified speaker: Now that doesn't balance the budget, but it at least leaves you neutral there.
1:03:04
Unidentified speaker: And why are you doing that?
1:03:06
Unidentified speaker: Well, you're doing that because you actually are going to raise tariffs a lot.
1:03:10
Unidentified speaker: And by the way, when you have a Republican Congress to now affirm all of those tariffs that you did through executive action before, that some of which have been overturned by the courts, they actually do produce a significant amount of revenue.
1:03:26
Unidentified speaker: And you can imagine they could produce a lot more, but who do they fall on?
1:03:30
Unidentified speaker: Like overwhelmingly, because they fall on the cost of goods and because lower income households are the ones who predominantly spend their income on goods rather than services.
1:03:43
Unidentified speaker: This is exactly raising taxes of the bottom 90%.
1:03:47
Unidentified speaker: So if you think about it that way, the Lutnick Trump talk last year is basically hey, we're going to raise taxes on everybody, but to offset the fact that we're raising taxes so much on the middle and lower classes, we're going to raise income taxes on the rich that would otherwise get away from them.
1:04:09
Unidentified speaker: So I think that's part of it.
1:04:11
Unidentified speaker: And then spending cuts are going to be significant.
1:04:14
Unidentified speaker: And that is going to have to happen through entitlements.
1:04:18
Unidentified speaker: And nobody really wants to do that.
1:04:21
Unidentified speaker: But I think this is a one part of the whole putting all this fraud that keeps popping up in front of people.
1:04:31
Unidentified speaker: Yeah, absolutely.
1:04:32
Unidentified speaker: People make the right point that when you look at this, oh, $80 million here, $100 million here, those seem like big numbers, but when you add them up, this is rounding air on the federal budget.
1:04:46
Unidentified speaker: That's not the point.
1:04:47
Unidentified speaker: The point is it says, look, federal programs are ripe with corruption and need to be reformed.
1:04:53
Unidentified speaker: Well, reform is a way to cut things.
1:04:57
Unidentified speaker: And oh, by the way, what is the thing that keeps coming back into focus?
1:05:01
Unidentified speaker: I see more and more reports highlighting this, that one of the big drivers of this is Obamacare costs, right?
1:05:09
Unidentified speaker: And that is a very clearly partisan issue.
1:05:14
Unidentified speaker: So I think we'd love to take an X to that.
1:05:17
Unidentified speaker: Absolutely.
1:05:19
Unidentified speaker: And so I think when you put it all together, maybe he doesn't solve the whole problem, but he starts the process and does it in a way that effectively reaffirms to middle and lower class voters that, hey, I'm looking out for you, right?
1:05:40
Unidentified speaker: I'm doing this in a way that takes care of you.
1:05:44
Unidentified speaker: And you can say, well, yeah, but you're also undermining maybe Social Security and Medicare and things like that.
1:05:50
Unidentified speaker: But those programs are bankrupt anyway.
1:05:52
Unidentified speaker: I wonder how many voters at this point actually are thinking, yeah, that's going to exist in 10 years.
1:05:58
Unidentified speaker: So if you put them on more solid footing, that might actually be a positive.
1:06:03
Unidentified speaker: That's speculative on my part.
1:06:05
Unidentified speaker: No question.
1:06:06
Unidentified speaker: If you told me I had to do this, how would I do it?
1:06:08
Unidentified speaker: Or if you told me that he is definitely going to do this, what would be my supposition of what he's going to do?
1:06:14
Unidentified speaker: These are the things that would come to mind.
1:06:16
Unidentified speaker: This is the path that seems like the most realistic path.
1:06:21
Unidentified speaker: But this goes back to this is a bit of an out there call, but again, comes back to this.
1:06:27
Unidentified speaker: What would Jackson do and what is going to put him in the history books?
1:06:30
Unidentified speaker: I think we can all agree that for anybody listening to this who doesn't want to be France, take a look at social screen.
1:06:38
Unidentified speaker: That's the easy win.
1:06:39
Unidentified speaker: Just start there, right?
1:06:40
Unidentified speaker: But let's listen, you there's there's one risk that you cite.
1:06:43
Unidentified speaker: Well, there's a few, but the one that I wanted to talk about, you called it Mamdani rage.
1:06:47
Unidentified speaker: It's a very real risk.
1:06:48
Unidentified speaker: Let's just walk through what Mamdani rage is and the problems it might actually present.
1:06:53
Unidentified speaker: And I guess how we would know before election night if that was going to be a factor.
1:06:58
Unidentified speaker: So this is exactly what we were talking about at the beginning of the podcast about this idea that the populism could be on the left or the right.
1:07:06
Unidentified speaker: And is low trust an equal feature of left-wing and right-wing populism?
1:07:11
Unidentified speaker: And as I said, there are some signs that maybe it is equal, or at least that it somewhat affects the left in Europe.
1:07:19
Unidentified speaker: I've seen very little sign that's the case in the United States so far.
1:07:25
Unidentified speaker: But this election may prove that.
1:07:28
Unidentified speaker: Where are we going to see that?
1:07:30
Unidentified speaker: I would look in the same way that if you look across states at their polling error.
1:07:37
Unidentified speaker: So just draw a little graph in your head.
1:07:39
Unidentified speaker: On one axis, put how Republican versus Democrat that state is.
1:07:46
Unidentified speaker: And on the other axis, put polling errors.
1:07:49
Unidentified speaker: And it's basically a 45-degree line.
1:07:51
Unidentified speaker: Polling errors go up with the share of the Republican vote.
1:07:56
Unidentified speaker: That's that right-wing populism.
1:08:00
Unidentified speaker: What I would expect is you'll see the opposite in the opposite direction as you go down towards Democrats, that you'll start to get polling errors in the opposite direction as you go towards more and more Democratic states.
1:08:14
Unidentified speaker: So the places where that's going to show up are probably places that people aren't really watching.
1:08:21
Unidentified speaker: I can't remember any of the safe Democratic seats up this year, but it's going to be those.
1:08:27
Unidentified speaker: It could show up in a place like Maine, although it's really interesting.
1:08:31
Unidentified speaker: Maine, as Democratic as it is, this is a really interesting point, is that Susan Collins is wildly popular.
1:08:41
Unidentified speaker: So one of the things that shows up in the polling is that if you ask people in Maine who they want to control the Senate, they will overwhelmingly tell you the Democrats.
1:08:50
Unidentified speaker: If you ask them who they're going to vote for, they will say Susan Collins, the Republican, right?
1:08:56
Unidentified speaker: This does suggest candidates do make a difference.
1:08:59
Unidentified speaker: But so that's a place that I'd look for it.
1:09:01
Unidentified speaker: The other places that I look for it, they're much more knife edge because they're not clearly Republican or Democrat anymore, are the Rust Belt states, and two of them are up right now, which is Michigan and Minnesota.
1:09:19
Unidentified speaker: And the leads for the Democratic candidates in those two states are narrow.
1:09:23
Unidentified speaker: They're only about one or two points.
1:09:25
Unidentified speaker: Which, as you say, is well, well, well within the margin of error.
1:09:29
Unidentified speaker: Before we close, you tease us at the beginning with polymarket and you get to explain when they're good and when they're bad.
1:09:35
Unidentified speaker: So let's just close with that because I'm fascinated by that.
1:09:37
Unidentified speaker: So just run us through that if you wouldn't mind if you've got time.
1:09:39
Unidentified speaker: I wrote something about this in one of my seriously Marvin publications, which are free.
1:09:45
Unidentified speaker: Anybody can read those.
1:09:47
Unidentified speaker: The issue is that betting markets, like financial markets, are really good at objective risks.
1:09:54
Unidentified speaker: That is, those things where we can clearly see the probabilities.
1:10:00
Unidentified speaker: So, for instance, non-farm payrolls.
1:10:03
Unidentified speaker: Markets are really good at pricing non-farm payrolls because we have a lot of data and it's a pretty well-known process, and people have models for it, and they know roughly what the statistics are.
1:10:12
Unidentified speaker: Where markets don't do well is what is defined as uncertainty.
1:10:19
Unidentified speaker: And uncertainty in mathematical terms is very different from what I think people's layman definition of it is.
1:10:27
Unidentified speaker: The layman definition is uncertainty is anything I don't know, or anything that could be random.
1:10:32
Unidentified speaker: That's not what uncertainty is in mathematical terms.
1:10:36
Unidentified speaker: Mathematical terms, uncertainty is a non-quantifiable risk.
1:10:40
Unidentified speaker: You have no way of assigning a probability to that thing.
1:10:43
Unidentified speaker: So, what's the probability that 7 billion people are going to go in lockdown next year in December 2019?
1:10:50
Unidentified speaker: Well, nobody even thought it was on the menu then.
1:10:53
Unidentified speaker: But of course, it did happen.
1:10:55
Unidentified speaker: What's the probability that we have massive polling errors in three straight presidential elections that are all lined up in one direction?
1:11:03
Unidentified speaker: Probably pretty low under classical statistical terms.
1:11:06
Unidentified speaker: And yet it happened because there's uncertainty, a non-quantifiable risk in this.
1:11:13
Unidentified speaker: And so this is one of the things that I keep trying to press people on is that a lot of the world that we see today around us is associated with uncertainty.
1:11:25
Unidentified speaker: It's stuff that has either never happened before or happens so rarely in history that we don't have good statistics on it.
1:11:32
Unidentified speaker: And so as a result, you need to use theory, you need to use that judgment we talked about earlier to assess what the realistic risks of these things.
1:11:43
Unidentified speaker: So everything that I did in this piece is about assessing uncertainty over something that cannot be statistically arbitrated.
1:11:52
Unidentified speaker: And that's why this is a risk for betting markets.
1:11:56
Unidentified speaker: And when people say, oh, but the betting markets say this, so this must be the right answer.
1:12:00
Unidentified speaker: I'm like, well, my first question that comes to my mind is is the betting market on an objective risk with known probabilities, or is it on an uncertain risk, one that we cannot define the probabilities of?
1:12:18
Unidentified speaker: Fascinating.
1:12:18
Unidentified speaker: Marvin, I've kept you longer than I planned to, but that's your own damn fault, frankly, stop being so interesting.
1:12:23
Unidentified speaker: This has been every bit as much fun as I did the other night.
1:12:26
Unidentified speaker: I only a lot more people get the benefit of the contents of that head of yours this time around.
1:12:30
Unidentified speaker: So for the people listening, just let them know how they can follow your work.
1:12:33
Unidentified speaker: You mentioned seriously, Marvin, there, but there are plenty of other ways.
1:12:36
Unidentified speaker: And I'm as keen now as I was the first time we met to introduce you to as many people as possible because I just find your work incredibly thought-provoking and incredibly useful for me when I'm trying to figure out what the hell is going on.
1:12:47
Unidentified speaker: That's very kind of you to say, and thank you.
1:12:51
Unidentified speaker: The best place to find me is thematicmarkets.com, and there you can find all of my research and seriously marvin as well.
1:13:02
Unidentified speaker: So my research is Thematic Markets.
1:13:04
Unidentified speaker: That is institutional grade research.
1:13:06
Unidentified speaker: That's what this piece was published on, and that I charge a fee for.
1:13:10
Unidentified speaker: And I also do these free pieces called Seriously Marvin, which are generally my unconventional thoughts that I don't really have time to devote to do significant research or are off-pieced, maybe not quite market issues that everybody's going to be interested in, but those are all free.
1:13:31
Unidentified speaker: And I also do have a much smaller, less professional, less good podcast than yours, the thematic edge, which you can access for free on there as well.
1:13:40
Unidentified speaker: And then if you're a Substack fan, you can also find both or all three of those things on Substack.
1:13:45
Unidentified speaker: So look under thematicmarkets uh.substack.com or seriously marvin.substack.com.
1:13:51
Unidentified speaker: The usual social media, X, LinkedIn, wherever.
1:13:55
Unidentified speaker: It's all thematic markets for the at thematic markets, right?
1:13:58
Unidentified speaker: Fantastic markets.
1:14:00
Unidentified speaker: Marvin, my friend, always a pleasure.
1:14:02
Unidentified speaker: I look forward to our next in-person meeting, but this will have to do for the meantime.
1:14:05
Unidentified speaker: And again, thanks for your time and thanks for your thoughts.
1:14:07
Unidentified speaker: And I've got half a day to process this conversation.
1:14:09
Unidentified speaker: So hopefully I will sleep a little bit better tonight.
1:14:11
Unidentified speaker: Thank you.
1:14:12
Unidentified speaker: It was my pleasure.
1:14:13
Unidentified speaker: Thank you very much.
1:14:15
Unidentified speaker: Well, there you have it, folks.
1:14:16
Unidentified speaker: My sleepless nights are now your sleepless nights.
1:14:18
Unidentified speaker: So I'm sorry to dump that upon you.
1:14:19
Unidentified speaker: Hopefully you'll listen to this early in the morning and have a whole day to process what Marvin just said, and you'll sleep a little bit easier than I did when I had it all whizzing around my brain the other night after dinner.
1:14:27
Unidentified speaker: It's fascinating stuff.
1:14:28
Unidentified speaker: And again, there is no telling whether Marvin's right about these elections, but his track record is good.
1:14:34
Unidentified speaker: His predictions are decidedly non-consensus this year.
1:14:37
Unidentified speaker: And if he is right, it is going to cause all kinds of carnage.
1:14:41
Unidentified speaker: So I think you're best prepared by listening to it and at least having a sense in your head about what might happen if the perceived wisdom goes squirrely and the opposite happens.
1:14:51
Unidentified speaker: All Marvin's work is available at thematicmarkets.com.
1:14:54
Unidentified speaker: He's at the same handle at thematicmarkets on Twitter.
1:14:56
Unidentified speaker: And I again would urge you to check out his work because I think it's absolutely best in breed.
1:15:01
Unidentified speaker: That's it for me for another conversation.
1:15:03
Unidentified speaker: I enjoyed that one immensely, and I'll be back with another one in the not too distant future.
1:15:06
Unidentified speaker: In the meantime, as always, thank you so much for listening.
1:15:29
Unidentified speaker: So, while we hope you find it both informative and entertaining, please do your own research or speak to a financial advisor before putting a dime of your money into these crazy markets.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3 and interpreted by Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.

More from The Grant Williams Podcast

The Grant Williams Podcast

The Grant Williams Podcast Ep. 137 - Frank Giustra FULL EPISODE

Episode discussion: Frank Giustra discusses memoir, Lionsgate struggles, Goldcorp success, and philanthropy

Speakers’ claims: Frank Giustra recounts his journey from a difficult childhood to becoming a senior partner at Yorkton at age 24, then launching Lionsgate. He details the near-bankruptcy of Lionsgate due to ego and lack of operational experience, contrasting it with his later success in creating Goldcorp by identifying undervalued gold assets. Giustra also addresses political backlash surrounding his philanthropy and the Clinton Foundation, and concludes with reflections on longevity and his memoir, The Money Dilemma.

EntrepreneurshipMiningGold
59 min2 assets8 claims to verify
The Grant Williams Podcast

The Grant Williams Podcast Ep. 136 - Marvin Barth FULL EPISODE

Episode discussion: Marvin Barth analyzes 2026 midterm election shifts, Fed hawkishness, and fiscal risks

Speakers’ claims: Marvin Barth explains his failed prediction of a Republican wave, attributing the error to a backlash against the Iran war and Trump's administration rather than policy preferences. He argues the electorate is highly fluid and rejects Republican personalities while still supporting conservative ballot initiatives. Barth predicts Federal Reserve Chair Kevin Walsh will adopt a hawkish stance to establish a legacy, leading to higher real interest rates. The speakers discuss the fiscal unsustainability of current debt levels and the potential for equity market caps due to AI investment restrictions and rising rates.

Election AnalysisFederal ReserveFiscal Policy
55 min0 assets18 claims to verify
All insights