Guests discuss measuring market consensus via sentiment data and argue contrarian positioning signals opportunities.
Summary
Host and guests Greg Blaha and Jim Bianco explore how consensus forms and why it often leads to collective errors in forecasting. They examine tools like the CFTC COT report, Consensus Inc. surveys, and media sentiment to gauge extreme positioning. Bianco highlights missed recession predictions and persistent inflation myths as examples of herd behavior. The discussion suggests that extreme consensus, whether bullish or bearish, often marks turning points in asset prices.
Key takeaways
- 1 Consensus Inc. quantifies sentiment by crawling the web for financial newsletter writers, finding that in early October, only 16% were bullish on bonds, indicating extreme bearishness.
Unidentified speaker (speaker_2) · 9:31
Episode sections
- 0:00 – 2:04
Introduction and the Value of Contrarianism
Host introduces the episode's topic: determining the value of a contrarian opinion. The speaker notes that the current environment is post-COVID, characterized by supply chain issues and remote work, leading to sticky inflation.
- 2:04 – 6:36
How Consensus Forms and the Herd Echo Chamber
The host uses the 1906 ox-weighing contest to illustrate how independent guesses average out to accuracy, contrasting this with markets where participants read the same data and each other's research, creating an echo chamber.
- 6:36 – 12:12
Gauging Sentiment: Covers, Surveys & The COT Report
Guest Greg Blaha discusses methods to measure consensus, including anecdotal evidence like The Economist cover stories, quantitative surveys from Consensus Inc., and hard data from the CFTC Commitments of Traders (COT) report.
- 12:12 – 16:09
The Danger of One-Factor Models
The speakers argue that relying on a single indicator (like the COT report) is a flawed one-factor model. They emphasize that sentiment data is most useful when combined with other factors to identify extreme positioning.
- 16:09 – 18:53
Consensus in Polling & Magnitude Errors
Jim Bianco discusses the recent election, noting that while the directional outcome (blue wave) was predicted, the magnitude was missed. He argues that pollsters and models are smart but often share the same inputs and methodologies, leading to collective errors.
- 18:53 – 25:00
The AI Trade: Overbuilt vs. Underbuilt
Jim Bianco analyzes the AI sector, noting a split between the tech industry (arguing AI is underbuilt/needs more compute) and Wall Street (arguing it is overbuilt/needs ROI). He suggests the market is nearing a bottom due to extreme bearishness.
- 25:00 – 34:00
Inflation Myths & Fed Assumptions
The discussion shifts to inflation, with Bianco arguing that post-COVID structural changes (remote work, deglobalization) make inflation sticky. He critiques the Fed's reliance on static economic models that fail to account for these regime changes.
- 34:00 – 34:18
Outro
The host wraps up the episode, thanking the guests and listeners.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| Crude oil | commodity | neutral | I think you'll find that if you look at, as an example, large speculators, net positions in crude oil right now, the data does kind of follow, price follows flows, or flows, I'm sorry, |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
More than 60% of analysts herd towards the consensus when forecasting earnings.
Alex Miletus ·2:51
How to verify: Search for academic studies on analyst herding behavior in earnings forecasts, such as those published in the Journal of Finance or Accounting Review. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
In the 1906 livestock show, the middle guess for the ox's weight was around 1,207, and the actual weight was 1,198, less than 1% off.
Alex Miletus ·2:13
How to verify: Verify via historical records of the 2026 West of England Central Livestock Exhibition or Francis Galton's 2026 paper 'Vox Populi' in Nature. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Bloomberg survey economists consistently expected inflation to return to the Fed's 2% target in about a year from every surge.
Alex Miletus ·3:36
How to verify: Access the Bloomberg Economics Survey historical data archive and compare forecasted CPI trends against actual CPI releases from 2026-2026. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
In December 2022, the median Bloomberg Survey estimate put the odds of recession at 65%.
Alex Miletus ·5:56
How to verify: Check the December 2022 Bloomberg Economics Survey results, which are archived on Bloomberg.com or in financial news reports from that period. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
The S&P 500 returned 26% in 2023.
Alex Miletus ·6:05
How to verify: Verify the total return of the S&P 500 index for the calendar year 2026 using data from S&P Dow Jones Indices or Yahoo Finance. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
In early October, only 16% of newsletter writers were bullish on bonds, rising to 25% later.
Unidentified speaker (speaker_2) ·9:46
How to verify: Access Consensus Inc.'s historical sentiment data for bonds/treasuries for October 2026. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceThe AI stocks in the UBS AI Winner Stock Index corrected more than 20% (down 23%) as of Thursday, November 5th's close.
Unidentified speaker (speaker_3) ·19:51
How to verify: Check the performance of the UBS AI Winner Stock Index on November 5, 2026, via Bloomberg Terminal or UBS research archives. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceAI stocks are up 457% since November 2022, while the S&P 500 is up 84% through November 5, 2026.
Unidentified speaker (speaker_3) ·19:42
How to verify: Calculate the cumulative return of the UBS AI Winner Stock Index and S&P 500 from November 2026 to November 5, 2026, using market data providers. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
The AI stocks had corrected more than 20% in August 2024 and April 2025.
Unidentified speaker (speaker_3) ·20:03
How to verify: Review historical price data for the UBS AI Winner Stock Index for August 2024 and April 2025 to identify drawdowns exceeding 20%. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Kevin Warsh stated at the last Fed meeting that there was a 'big family fight' and that Fed members have different views on the state of inflation.
Unidentified speaker (speaker_3) ·33:09
How to verify: Review the transcript of the Federal Reserve press conference or meeting minutes following the most recent FOMC meeting prior to the podcast release. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Do your own research
- Verify CFTC Commitments of Traders (COT) data for current large speculator net positions in bonds and equities.
- Review Consensus Inc. sentiment data for current bullish/bearish ratios in financial newsletters.
- Check recent Bloomberg Survey economist forecasts for CPI and recession probabilities against actual economic data.
- Analyze The Economist cover stories from the past 12 months to identify potential contrarian signals.
- Assess whether current AI sector positioning reflects 'overbuilt' or 'underbuilt' narratives based on capex data.
- Evaluate Fed economic models for assumptions regarding post-COVID structural changes like remote work and deglobalization.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): I used to say at the time this is not 2019.
- 0:03
- Unidentified speaker (speaker_0): This is now a post-COVID environment. Then when I started looking at the supply chain problems, the shifts in attitudes about the globalization, the shifts in work about going remote and not going back to the office,
- 0:19
- Unidentified speaker (speaker_0): that's when I came to the conclusion that this inflation is going to be sticky. And it still is to this day.
- 0:45
- Unidentified speaker (speaker_0): For a more dynamic approach to core fixed income, visit wisdomtree.com slash WTBN to learn more.
- 1:04
- Unidentified speaker (speaker_0): investing involves risk principal loss is possible before investing carefully consider a fund's objective risks charges and expenses along with other information contained in the perspective, available at wisdomtree.com slash US. Read it carefully.
- 1:21
- Alex Miletus: I'm your host, Alex Miletus, joined by my colleagues, Greg Blaha and Jim Bianco, founder and president of Bianco Research.
- 1:30
- Alex Miletus: Every week, we try to counter consensus thinking while we discuss the world of markets and finance. And what a week it's been.
- 1:37
- Alex Miletus: Our topic today, what is a contrarian opinion actually worth? Being different isn't just the point, being different and right is.
- 1:46
- Alex Miletus: And before we can figure out when to go against the crowd, we need to understand how the crowd gets there in the first place. So I want to start by giving the consensus some credit. It's usually the average view of a lot of smart and informed people. And most of the time, that's a pretty good place to start.
- 2:04
- Alex Miletus: I want to use an old story to explain what I mean.
- 2:07
- Alex Miletus: In 1906, there was a livestock show where a bunch of people were paid to guess the weight of an ox.
- 2:13
- Alex Miletus: The middle guess was around 1,207.
- 2:17
- Alex Miletus: The ox came in at about 1,198, less than 1% off.
- 2:22
- Alex Miletus: Most of the guessers were nowhere close on their own. Together, though, they nailed it.
- 2:28
- Alex Miletus: Here's the thing.
- 2:30
- Alex Miletus: It only worked because everyone created their guess on their own. Nobody was looking over their shoulder or over their neighbor's shoulder.
- 2:38
- Alex Miletus: Markets don't work like that. Everyone in this business reads the same data releases, the same Fed statements, the same And research, they don't just usually read on the the data, same they screens. read each And other. they don't just read the data, they read each other.
- 2:51
- Alex Miletus: One study found that more than 60% of analysts heard towards the consensus when they're forecasting earnings.
- 2:58
- Alex Miletus: Some of that's just safety. If you land near everybody else and you're wrong, so is everybody else.
- 3:05
- Alex Miletus: If you're out on your own, though,
- 3:08
- Alex Miletus: and you're wrong, it's your name on the call and you're the one that's wrong.
- 3:12
- Alex Miletus: So part of what we call consensus is really real. an echo, and that echo works fine until the facts change.
- 3:20
- Alex Miletus: I want to use this chart to show what I mean. This chart shows CPI. The black line is CPI. Every colored line in the Bloomberg survey is a Bloomberg survey forecast taken in a given month. They're color-coded by year.
- 3:32
- Alex Miletus: Each time In the meantime, economists were surveyed, this is a Bloomberg survey,
- 3:36
- Alex Miletus: they expected inflation to return back to the Fed's 2% target in about a year.
- 3:42
- Alex Miletus: So every single time from the big inflation surge to now,
- 3:47
- Alex Miletus: they didn't just
- 3:49
- Alex Miletus: miss together, they miss in the same direction, the same shape, every single time. And you don't hear anyone calling for the heads of all those economists surveyed. That's the comfort of being wrong together. When everybody misses, nobody really looks all that bad.
- 4:05
- Alex Miletus: So that is an example of oneness. This next chart I want to use is the pattern.
- 4:11
- Alex Miletus: This is the Gartner hype cycle chart, and we've used it to describe a couple other things as well. It was built to track new technologies, but I think it works just as well for beliefs Thank you or so much understanding for joining consensus. us.
- 4:23
- Alex Miletus: The five phases, you get an innovation trigger, peak expectations.
- 4:45
- Alex Miletus: Everybody piles into the same story at the same time, and the consensus is it's most confident at that point and least questioned.
- 4:54
- Alex Miletus: Then comes the trough.
- 4:56
- Alex Miletus: When things fail to deliver and the crowd overshoots all the way on the way down. My favorite critique of this chart is that the Garmin's own track record is mostly a lottery.
- 5:05
- Alex Miletus: I think that actually proves the point more than anything else. Because I don't think the curve is necessarily tracking technology. it's tracking the rise and fall of the belief of that technology in other words consensus and consensus is somewhat decoupled from reality
- 5:21
- Alex Miletus: so put it all together you get everyone's using the same shared inputs people like being wrong together and expectations run along a similar curve or pattern and that's how consensus And because gets built markets market are participants forward-looking, that naturally opinion gets crowd priced in long into before the a outcomes collective actually show opinion up. and because
- 5:44
- Alex Miletus: And when the crowd is wrong, it gets expensive.
- 5:48
- Alex Miletus: An example of that one is heading into 2023, the Median Economist Bloomberg Survey estimate put the odds of recession at 65%.
- 5:56
- Alex Miletus: It stayed there from December 2022 all the way through the first half of that year.
- 6:03
- Alex Miletus: The recession never came.
- 6:05
- Alex Miletus: The S&P returned 26% that year.
- 6:09
- Alex Miletus: The consensus was reasonable, it was widely shared, and it was wrong. And anyone positioned with that consensus missed one of the best years of the decade.
- 6:19
- Alex Miletus: All of this is easier So to I think see Greg's in going hindsight, to start by walking obviously, us through some but of the places actually that people identifying actually do look where for those a consensus consensus studies. is isn't really
- 6:37
- Unidentified speaker (speaker_2): Right. So I think you did a great job of kind of conceptually showing what this herd mentality is and how consensus forms. But if you want to get a bit more specific, there's a lot of different ways you can kind of try to pinpoint what a consensus is. and maybe in the most abstract form or least numerical way of looking at it you can look at some anecdotal
- 6:57
- Unidentified speaker (speaker_2): evidence and the best example I think of that type of thing is the economist cover story something people have covered for many many years decades even and just the general idea that when the economist puts out Thank a you cover so much story for joining us. that's kind of
- 7:33
- Unidentified speaker (speaker_2): We've seen
- 7:36
- Unidentified speaker (speaker_2): a
- 7:37
- Unidentified speaker (speaker_2): lot of people on Twitter point out that that cover story came out almost the exact day that yields peaked and started heading lower. So there's stories like that. Even earlier in the year, they had a story about crude oil prices saying that even though prices are high, they have a lot higher to go. Again, that was a situation where they printed Thank you so that
- 7:57
- Unidentified speaker (speaker_2): much cover for story joining us. pretty much right during one of the highs in crude oil prices this year.
- 8:01
- Unidentified speaker (speaker_2): So...
- 8:21
- Unidentified speaker (speaker_2): AI ARCHITECTS
- 8:45
- Unidentified speaker (speaker_2): These are kind of anecdotal things. I think if you wanted to take a step further and start kind of quantifying some of these sentiment ideas, you can look at some surveys. There's a lot of different ones out there. There's the American Association of Individual Investors. There's Institutional Investors.
- 9:02
- Unidentified speaker (speaker_2): I'm going to look specifically at Consensus, Inc.'s data.
- 9:06
- Unidentified speaker (speaker_2): They look at over 30 different markets from stocks to bonds and all kinds of commodities.
- 9:12
- Unidentified speaker (speaker_2): and they look at newsletter writers across the internet. and kind of look at okay this newsletter writer put out a story about say bonds and is it bullish or bearish and they go crawling the web finding all these financial newsletter writers and just trying to determine are these people bullish or bearish the advantage to this approach
- 9:31
- Unidentified speaker (speaker_2): is that you can and then start saying quantifying this and charting it and one example i'll just show here of consensus inc data is newsletter writers are really bearish right now on bonds or tenure notes to be more specific
- 9:46
- Unidentified speaker (speaker_2): if you look at early october only 16 percent of newsletter writers were bullish on bonds now that's up to 25 percent but that's still a pretty low percentage so that's a situation where you know you might think okay everyone's pretty pretty bearish here Maybe. it's time to go bullish. And Jim, I know you were, you know, one of the first people to
- 10:06
- Unidentified speaker (speaker_2): get out there and say, okay, let's go bullish bonds here. And this crowd has just overextended itself. And maybe, you know, this was one of the reasons that you look at this type of sentiment.
- 10:16
- Unidentified speaker (speaker_2): so yeah maybe this type of sentiment is it is just kind of opinion based there's not really any hard data behind it in terms of okay people are actually making bets with their own money. So if you want to go even one step further, you can look at sentiment from the angle of, okay, what are people doing with their money? You can look at
- 10:36
- Unidentified speaker (speaker_2): flows into or out of ETFs.
- 10:38
- Unidentified speaker (speaker_2): And one of the better examples that we like to look at from a report perspective is the CFTC does a report And so every the advantage week called their here is, Commitments well, one, you of know, Traders these people Report, are actually COT. placing bets and with they money. just show different groups.
- 11:09
- Unidentified speaker (speaker_2): with money So this isn't just a newsletter writer, you know, spouting an opinion with nothing really backing it up. but if you're actually making these bets you can look at the net positions of say large speculators and the cot report generally you know you want to kind of fade the large speculators the idea being that speculators get kind of follow
- 11:28
- Unidentified speaker (speaker_2): price movements and they get more and more bullish as prices go up and that also means that then And when prices do inevitably turn lower they're at their most extreme you know bullish position
- 11:39
- Unidentified speaker (speaker_2): I think you'll find that if you look at, as an example, large speculators, net positions in crude oil right now, the data does kind of follow, price follows flows, or flows, I'm sorry,
- 11:51
- Unidentified speaker (speaker_2): are kind of an indicator of sentiment. And so
- 11:55
- Unidentified speaker (speaker_2): if you look at this data, you might ask, okay OK, well, if it's just
- 12:00
- Unidentified speaker (speaker_2): sentiment follows prices, you know what you use see. is this
- 12:04
- Unidentified speaker (speaker_2): and i think if you want to look at this data it's not you're not going to be able to make a trading system off any one of these examples that i've shown we like to say in the office that any one factor model is going to be a terrible model you're never going to be able to make money off of that but i wouldn't say that that makes any of these things useless
- 12:23
- Unidentified speaker (speaker_2): you're gauging sentiment opinion here and if you're coming at this from a perspective of okay i've looked at a market from a lot of different perspectives and i'm kind of thinking i want to get bullish and then you look at sentiment data and you see for example on the treasury market everyone's really bearish then that's where this data becomes maybe kind of
- 12:43
- Unidentified speaker (speaker_2): useful that you could say okay i've got a lot of reasons to be bullish the crowd is really bearish so let me jump you know against them form this contrarian opinion and then if, as I think, prices start moving higher, all those bears are going to start moving out of their position, and that's going to provide a great tailwind for, you know, my thesis of being bullish.
- 13:02
- Unidentified speaker (speaker_2): So, these are just a few examples of different types of sentiment data and how you can either measure them or just see where the consensus Thank you is. so much And for joining Jim, I guess us. I would throw this to you as someone who has kind of made a career being a contrarian.
- 13:33
- Unidentified speaker (speaker_0): My
- 13:35
- Unidentified speaker (speaker_0): bugaboo is when I look on financial Twitter or when I'm talking to people,
- 13:39
- Unidentified speaker (speaker_3): they
- 13:39
- Unidentified speaker (speaker_0): tend to kind
- 13:40
- Unidentified speaker (UNKNOWN): of
- 13:40
- Unidentified speaker (speaker_0): shorthand consensus is, yes, when everybody's got an opinion, go against it because they're all
- 13:45
- Unidentified speaker (speaker_3): dumb.
- 13:46
- Unidentified speaker (UNKNOWN): Well, it's not
- 13:47
- Unidentified speaker (speaker_0): that they're dumb. It's more to the point
- 13:50
- Unidentified speaker (speaker_3): of
- 13:50
- Unidentified speaker (speaker_0): if everybody's got an opinion
- 13:52
- Unidentified speaker (speaker_3): and
- 13:52
- Unidentified speaker (UNKNOWN): they all get
- 13:52
- Unidentified speaker (speaker_0): on
- 13:52
- Unidentified speaker (UNKNOWN): the
- 13:53
- Unidentified speaker (speaker_0): same
- 13:53
- Unidentified speaker (UNKNOWN): side of the
- 13:53
- Unidentified speaker (speaker_0): boat, who's left to
- 13:54
- Unidentified speaker (speaker_3): buy? And
- 13:55
- Unidentified speaker (UNKNOWN): that's
- 13:56
- Unidentified speaker (speaker_3): why if
- 13:56
- Unidentified speaker (speaker_0): they all
- 13:57
- Unidentified speaker (UNKNOWN): get
- 13:57
- Unidentified speaker (speaker_0): bullish and everybody's bullish, there's no more marginal buyers or not enough
- 14:01
- Unidentified speaker (UNKNOWN): marginal buyers might be a
- 14:02
- Unidentified speaker (speaker_0): better way. to
- 14:03
- Unidentified speaker (speaker_3): say it. Or if they're all bearish, there's not enough marginal sellers. And that's why a trend tends to exhaust itself and turn to the other direction.
- 14:13
- Unidentified speaker (speaker_3): And you're right. It's that they're smart. And this is what makes consensus opinion so hard to go against because they're These the are smartest, best people that you know, and they're all on one side of the trade.
- 14:27
- Unidentified speaker (speaker_3): And how can I, me mere mortal, say that they're wrong to be on that side of the trade? But if you think about it in terms of the marginal buyer and the marginal seller, if everybody I'd look at is long, if everybody I'd look at is short, who's the next person to come into this market to buy? Who's the next person Thank to you for convince joining us. to sell, whichever way we
- 14:47
- Unidentified speaker (speaker_3): happen to be? And you kind of run out of that.
- 14:50
- Unidentified speaker (speaker_3): No.
- 15:09
- Unidentified speaker (speaker_3): internally we always like to say that there's no such thing as a one-factor model that seems to work well one factor model is oh let's look at the cot and measure it against the tenure yield and when the cot gets to this level buy it when it gets to that level sell it that's a one-factor model if you have a 10-factor model and one of them is cot
- 15:29
- Unidentified speaker (speaker_3): that additive might be beneficial to your 10-factor model but one-factor models tend to fail time and again and the reason i say that is if you traffic on social media or something like that, somebody will say, this indicator did this. So therefore, buy stocks or sell stocks. That's a one-factor model.
- 15:46
- Unidentified speaker (speaker_3): And one-factor models, like I said,
- 15:49
- Unidentified speaker (speaker_3): don't have a good track record, nor should they in
- 15:53
- Unidentified speaker (speaker_3): a random walk type of market. So that's kind of why we wanted to go over consensus opinion at the beginning. is to try and make people understand how this works. And the final thing I'll give you about how hard it is to measure consensus opinion, I'll go with the big consensus of this last
- 16:13
- Unidentified speaker (speaker_3): week, the election.
- 16:15
- Unidentified speaker (speaker_3): We have spent, we being the collective of everybody,
- 16:20
- Unidentified speaker (speaker_3): hundreds of millions of dollars, if not more,
- 16:24
- Unidentified speaker (speaker_3): with prediction markets and... polls and models and frequency of all of those updates to try and predict what was going to be the outcome of the election.
- 16:38
- Unidentified speaker (speaker_3): Now, you could argue, well, we got it right. We said that there would be a blue wave, And there was a blue The wave. Correct. But we got the magnitude wrong with that. The blue wave. was way bigger than anybody thought it was going to be.
- 16:55
- Unidentified speaker (speaker_3): I could have saved you hundreds of millions of dollars and told you
- 16:58
- Unidentified speaker (speaker_3): midterm election and a second term president always goes against the second term president. They had a two seat majority in the House. The Republicans did. They're going to lose the House. There, I saved you hundreds of millions of dollars. And that was correct. But was that correct by that they won three seats or they won two seats? 40-plus seats? That's really the question
- 17:18
- Unidentified speaker (speaker_3): that we kind of missed. So we got the magnitude wrong. And the reason is not because everybody's dumb,
- 17:24
- Unidentified speaker (speaker_3): and I'll push back on what you're thinking, not because everybody's got a political bias. Pollsters are trying to get it right.
- 17:32
- Unidentified speaker (speaker_3): They have reputations and money on the line, and they cannot just be using political bias because they'll be. former pollsters, because nobody will engage with them if they're never right. And because of that,
- 17:45
- Unidentified speaker (speaker_3): it, you know, it's very difficult to measure the consensus opinion.
- 17:51
- Unidentified speaker (speaker_3): So, yes, I look at all of these things all at once. To your point, Greg, the magazine covers are very good.
- 17:59
- Unidentified speaker (speaker_3): Why? Because most magazine covers like The Economist. or Time, they don't need to put a financially related theme on the cover. They could put a cultural theme on it, a sports theme on it, a political theme on the cover. But when a financially related theme so cuts through
- 18:19
- Unidentified speaker (speaker_3): everything else and becomes the dominant You story, know, you know,
- 18:23
- Unidentified speaker (speaker_3): AI is bullish or AI is bearish.
- 18:27
- Unidentified speaker (speaker_3): That tells you everybody's bought or everybody's sold. And that's why they can be useful. But again, it's a one-factor model, just like the Time Magazine Person of the Year didn't work quite as well as everybody thought. So with that said, maybe as a quick understanding how AI, or excuse me, how
- 18:46
- Unidentified speaker (speaker_3): consensus opinion works, Let's talk about the most important market. in the world right now, and that is AI.
- 18:53
- Unidentified speaker (speaker_3): And we'll start with this chart on the drawdown. And I'll quip that
- 18:58
- Unidentified speaker (speaker_3): my friend Ed Yardeni
- 19:00
- Unidentified speaker (speaker_3): has a weekly podcast, and in it, he's been saying the last couple of weeks, I continually try to say to myself, this week I don't want to talk about AI, and then we wind up talking about AI, because it is the story right now. It's the story. of the market. Not only the stories of the markets now, it might be the most dominant story we've seen in markets, maybe in our
- 19:20
- Unidentified speaker (speaker_3): careers. So what this chart shows you in blue is the UBS
- 19:26
- Unidentified speaker (speaker_3): AI winner stock index. And in black, it shows you the S&P 500. And
- 19:33
- Unidentified speaker (speaker_3): it starts in November of 2022 when ChatGPT released its free research service, the Free ChatGPT.
- 19:39
- Unidentified speaker (speaker_3): WHICH KIND OF LAUNCHED THIS WHOLE AI PERIOD.
- 19:42
- Unidentified speaker (speaker_3): EVEN WITH THE SELL-OFF, THE AI STOCKS ARE 457% SINCE NOVEMBER 22, WHERE THE S&P IS UP 84%.
- 19:51
- Unidentified speaker (speaker_3): THIS IS THROUGH THURSDAY, NOVEMBER 5TH'S CLOSE. THE BOTTOM PANEL SHOWS YOU THE DRAWDOWN. THE AI STOCKS ARE IN A MEDIA-DEFINED BEAR MARKET. THEY'VE NOW CORRECTED MORE THAN 20%. They're down 23%.
- 20:03
- Unidentified speaker (speaker_3): THIS IS THE THIRD TIME IN THE LAST TWO YEARS THEY'VE DONE THIS. THEY DID IT IN AUGUST OF 2024 WHEN WE HAD THE YEN CARRIE TRADE UNWINDED. THEY DID IT DURING LIBERATION DAY ON A SELL-OFF IN THE MARKET IN APRIL OF 2025.
- 20:18
- Unidentified speaker (speaker_3): so this is kind of the focus is as to where we should be is where is everybody on ai now i'll look at sentiment in terms of what i see on twitter acts i'll look at cover stories i'll look at measures of what's the flows are all of that and use some judgment
- 20:39
- Unidentified speaker (speaker_3): notice i didn't say i've got a model and i'll stick it all in the model because it's really hard like i said we've spent hundreds of millions of dollars putting put together models to try and figure out who was how where everybody was going to vote.
- 20:52
- Unidentified speaker (speaker_3): And we got it magnitudinally wrong, although directionally we got it right.
- 20:56
- Unidentified speaker (speaker_3): And so I would argue right now that when it comes to AI, you've got a lot of bearishness, almost too much bearishness in AI.
- 21:06
- Unidentified speaker (speaker_3): That's why we're down 23%. That's why it's sold off so much, is that the belief there is a real interesting.
- 21:15
- Unidentified speaker (speaker_3): um you know split in the world right now
- 21:19
- Unidentified speaker (speaker_3): the tech community going into the election maybe they're going to back off of this a little bit on the election was arguing there's not enough compute in the world that the anthropic s1 said it's the biggest risk i know what the headline was AI you know ai is going to destroy humanity but the biggest risk that was clearly delineated is
- 21:39
- Unidentified speaker (speaker_3): we're not going to have enough compute. We're not going to have the ability that you're going to open a cloud account and when you go ask it a question, it's not going to give you an answer because they don't have enough compute.
- 21:51
- Unidentified speaker (speaker_3): So they're arguing we have underbuilt AI.
- 21:55
- Unidentified speaker (speaker_3): Wall Street is selling off because they're wondering, where's the payback on everything you've built? Where are you going to get the money to pay back the debt
- 22:04
- Unidentified speaker (speaker_3): to satisfy shareholder expectations? They're arguing you've overbuilt. So you've got the industry arguing you've underbuilt. You've got Wall Street arguing you've overbuilt both at the same time. I don't think I've ever seen it, you know, be that different opinion. So the consensus you're seeing Thank you is for the joining Wall us. Street consensus
- 22:24
- Unidentified speaker (speaker_3): that we're overbuilt. And that's why the stocks
- 22:45
- Unidentified speaker (speaker_3): And we're going to be left with people that are no longer in AI or those that are still in AI related investments have, you know, strong hands or as the crypto calls them, diamond hands. If they didn't sell now, they're just not going to sell. They're not part of the universe of potential sellers in the market. And then you
- 23:05
- Unidentified speaker (speaker_3): start to find some kind of a bottom. So that's how you got to think about contrarian opinion or what is the consensus. and have you exhausted the potential pool of people that could buy or sell. You can't measure precisely to the day. I know the Economist magazine did it, but that was more happenstance
- 23:25
- Unidentified speaker (speaker_3): than anything else that it got it on the exact day. But I think that if you were to measure AI trade right now, we're very close to that bottom just because there's so much negativity because everybody's. arguing, have we overbuilt AI? And they're coming to the conclusion, yes, where the
- 23:45
- Unidentified speaker (speaker_3): industry is trying to tell you, no, it's underbuilt, and we don't have enough for expected demand in the future.
- 23:52
- Alex Miletus: so i think you hit on a couple really interesting points here i want to
- 23:57
- Alex Miletus: bring to focus one of the things you said which is what you both said which is you know one factor model doesn't work. But something that you actually kind of breezed by, Jim, there was
- 24:08
- Alex Miletus: models themselves are often shared. We spend millions of dollars, whether you are within an institution like an investment bank or you are a
- 24:19
- Alex Miletus: economic forecaster or a pollster,
- 24:22
- Alex Miletus: there's millions and millions of dollars spent on models.
- 24:27
- Alex Miletus: and most often someone builds a really complex model and then it's distributed, it's shared. Everyone then uses that model as the gold standard, right? A great example would be Nate Silver and polling, right? So
- 24:41
- Alex Miletus: what you see is
- 24:43
- Alex Miletus: kind of back to what I mentioned, where you don't only have the same inputs, but you have the same types of analysis, the same methodologies method. used to parse through the inputs themselves.
- 24:55
- Alex Miletus: So I think that's something important to highlight in how all of that consensus kind of forms. But the thing I wanted to ask you was, you
- 25:04
- Alex Miletus: covered AI really well in terms of what's currently happening. I want to get to something that
- 25:10
- Alex Miletus: was more, it was recent, but, you know, has since passed past. and is sort of coming back into light.
- 25:17
- Alex Miletus: You made the contrarian call a couple years ago that inflation was not transitory. The consensus call was inflation is transitory. This was a blip. And it is supply chain disruptions. It is post COVID.
- 25:35
- Alex Miletus: That was proven significantly. That was proven to be wrong.
- 25:40
- Alex Miletus: What did you look at in those moments? And what do you look at now to compare that kind of environment where everyone sits on this one side of inflation is obviously transitory, the Fed should be doing XYZ based on that premise. But you then said, the premise itself is
- 26:00
- Alex Miletus: wrong. And then everything else follows from that. as to what that means. The Fed should do X, Y, and Z is different if the transitory argument was wrong and you were one of the few to push back and say, you know,
- 26:13
- Alex Miletus: this is persistent inflation. And it turns out, I mean, we are 67 months above the Fed's 2% target. We are
- 26:21
- Alex Miletus: not going in the right direction with inflation either. So what are your thoughts on that? What did you look at? and what are you looking at now in terms of inflation and what people think about
- 26:32
- Alex Miletus: the consensus in how inflation is measured right now?
- 26:36
- Unidentified speaker (speaker_3): A couple of things, just like I said with contrarian opinion,
- 26:41
- Unidentified speaker (speaker_3): that it works. It's a good idea. But the biggest problem The with it is measuring what the consensus is. So, yes, contrarian opinion is. an excellent way to go about investing, but trying to figure out what the consensus is, is extraordinarily hard.
- 26:58
- Unidentified speaker (speaker_3): In inflation, it's something similar. Milton Friedman's dictum, too much money chasing too few goods,
- 27:04
- Unidentified speaker (speaker_3): is probably the best argument you could use for measuring if we're going to have excess inflation or not. the problem is now you could can spend the rest of your life trying to figure out what money is. Is money M2? Does it include unrealized stock gains? Does it include appreciation of your house? How much? How do you measure it?
- 27:24
- Unidentified speaker (speaker_3): So if you're going to say, no, just use M2, you're going to fail because we've tested it. It doesn't work. But the concept of when there's too much money in the system, you're going to get inflation. When there's too little money in the system, you're not going to get inflation. Probably does work. And like I said, try and figure out what too much money in the system means that is a very difficult subject now i'll throw
- 27:44
- Unidentified speaker (speaker_3): out one other concept for you too on this and that other concept is that when you have models on the economy the fed has a model called us us one and that This is basically their big model for how the economy works the conference board has a model for the index of leading
- 28:04
- Unidentified speaker (speaker_3): economic indicators. It's 10 economic indicators that tells you how the economy works. We've often said
- 28:11
- Unidentified speaker (speaker_3): that when you have a financial crisis or recession,
- 28:14
- Unidentified speaker (speaker_3): everything changes, that you shock this system, and then when you come out of that recession, it comes out of the recession differently. And I've We've talked about here and in other places for years, you know, remote work and de-global. deglobalization
- 28:29
- Unidentified speaker (speaker_3): and the like were some of the things that came out of the post-COVID period. And so if you do have an economic model like the Fed does or like the conference board does, you should almost put into your thinking the minute Thank that you so much have for joining a recession, us.
- 28:48
- Unidentified speaker (speaker_3): rework your model. You have to start reworking your model.
- 29:09
- Unidentified speaker (speaker_3): And the second thing is, I started with, we are coming out of a big financial crisis and a recession unlike any other. And things are changing. And because I started
- 29:20
- Unidentified speaker (speaker_3): with the premise that whatever you thought, I used to say at the time, this is not 2019.
- 29:25
- Unidentified speaker (speaker_3): This is now a post-COVID environment. Then when I started looking at the supply chain problems, the shifts in attitudes about the globalization, that the shifts in work about about going remote and not. going back to the office
- 29:41
- Unidentified speaker (speaker_3): that's when i came to the conclusion that this inflation is going to be sticky and it still is to this day so that's the other thing you got to keep in mind is a lot of this stuff is very difficult to measure and to what we've been talking about how could you say 80 economists are wrong. Do you have any idea what schools they went to?
- 30:01
- Unidentified speaker (speaker_3): What backgrounds they have? how learned they are to say that they're all wrong?
- 30:07
- Unidentified speaker (speaker_3): I don't say that because they're dumb. I say they're smarter than me. I say it because are they using the right set of assumptions? And I was arguing at the time they weren't, that the assumptions needed to change. And that was the basis for the argument. So you laid And that out a good was representation the basis for of the the past contrarian call and you talked call.
- 30:25
- Alex Miletus: about so a lot you right laid now. out
- 30:32
- Alex Miletus: You talked about AI right now. What are your thoughts on the,
- 30:37
- Alex Miletus: is there a consensus? Is there a contrarian call in inflation right now? Do you see, do
- 30:43
- Alex Miletus: you see that as something that is, Are you know are people expecting like they were then that that inflation will just return to the 2% target? I know the Fed consistently pushes that return to 2% target two years out. I think now it's at 2029.
- 31:02
- Alex Miletus: Earlier this year, it was 2028.
- 31:04
- Alex Miletus: So they're essentially just kicking the ball down the line. So do you have an understanding now of what inflation? what inflation
- 31:15
- Alex Miletus: is in terms of a consensus or a contrarian in the markets?
- 31:19
- Unidentified speaker (speaker_3): Yeah, I think that the consensus on it right now, I mean, this kind of gets more towards Fed policy a little bit,
- 31:27
- Unidentified speaker (speaker_3): is that inflation is not a problem, that it's ultimately going to return to 2%,
- 31:32
- Unidentified speaker (speaker_3): which is why you continually see this pushback about the Fed having to hike rates. that it's not necessary. Inflation is on its, it's been on this glide path lower, really hasn't for the last year and a half, but it's been since 2022 on a glide path lower to 25 and then it started back up. And that the
- 31:52
- Unidentified speaker (speaker_3): consensus there is probably that the Fed doesn't need to raise rates, that inflation is going to become under control. But unlike the AI consensus, where I think. it's much stronger it's overbuilt they can't pay this back these stocks have to be revalued down these bonds have to be revalued down that's a much stronger consensus the
- 32:12
- Unidentified speaker (speaker_3): inflation consensus that it's on its way back down towards three percent it's much weaker i don't think you know if you use the the framework Now,
- 32:21
- Unidentified speaker (speaker_3): have we run out of buyers or in the case of ai if we run out of sellers everybody's selling Well, maybe not at the moment we're recording, but we're very close to it is where I think we are. And it still might go lower to finally get all of those people out.
- 32:35
- Unidentified speaker (speaker_3): But when it comes to the, you know, the consensus on where inflation is and what the Fed should do, there's still a lot more. if the consensus is leaning towards Thank you for the joining idea us. that inflation is under control, there's a fair
- 33:09
- Unidentified speaker (speaker_3): That if you read the transcripts of what
- 33:12
- Unidentified speaker (speaker_3): Kevin Warsh said about the last Fed meeting, he said, we had a big family fight and we had sobering discussions because I don't even think within the 12 Fed members, if you asked them, what's the state of inflation, you're going to get 12 different answers. And that's fine. Nobody knows what the state of inflation is. Actually, what it used to be the wrong answer was, well, the model
- 33:32
- Unidentified speaker (speaker_3): says this, and we all agree with the model. That's what leads you to astray. it's fine to have a bunch of different opinions and try and hash it out internally. But I don't think there's any, there's a, you know, I would argue the majority of people think that inflation's headed back to two, but that might be a 55, 45, or even a 51,
- 33:50
- Unidentified speaker (speaker_3): 49 opinion. It's not like an 80, 20 opinion that inflation is heading lower right now.
- 33:56
- Unidentified speaker (speaker_3): And so it's, it's still very much up in the air.
- 34:00
- Alex Miletus: You heard it here first.
- 34:02
- Alex Miletus: I think that'll wrap us up for this week. Thank you all for joining us. Make sure you like, subscribe, and feel free to leave a comment if you want us to talk about something specific or have any ideas or
- 34:13
- Alex Miletus: questions for us. Thank you for joining, and we'll see you guys next week.