Using AI to Find Investing Stories with Perscient Co-Founder Ben Hunt

32:35Again, it's not complex, but it is at scale. And what we're doing with that is we're reading the world's news to understand the world's stories and narratives. And you're right. It's much bigger than or it's much more focused. We have much higher resolution than just saying, oh, I'm bullish on financials. I mean, that's a narrative, sure. But there are 20 different variations of that. You're bullish on financials. Why? And we can track all those. We can track all those stories and how they wax and wane over time. So let me roll back to, I want you to explain what Persean is, who are the clients, I don't mean names, but what type of clients do you have, and what do they do with Persean's output?

33:31So we started Persean in 2018. My partner from, we were at an asset manager, spun out there to take the technology that I've been working on for years and really been writing about with Epsilon Theory. So we started that in 2018 to do the basic research into processing enormous amounts of financial news data and to track the stories and how they rise and fall and wax and wane over time. That was the story. I love that description because there are a lot of trades going on where the storyline changes on a regular basis. Probably the Mac daddy of that is crypto. First, it's, hey, you know, there's fiat currency.

34:27This is outside of the system. It's DeFi. Then it's a hedge for deflation. Then it's a hedge for inflation. Now it's scarcity. And digital gold. Digital gold. There are no fundamentals, right, with crypto. People will say there are, but there aren't, right? It's driven by the waxing and waning of stories. And do they find purchase? Or do they kind of – people get tired of them? Well, they got tired of the DeFi story, and then once J.P. Morgan and BlackRock started creating, like the Ibit is the fastest ETF to$100 billion. And so the old story of DeFi is gone, and the new story is, oh, no, this is an asset class that Wall Street's embracing.

35:15That's why you have to own it. That story, that story was in very – that was a very similar story, by the way, or a transitioning story from physical gold to GLD. When that ETF came out, which was a very similar pattern, because once it became a table at the Wall Street casino, then it takes on a different meaning, specifically around gold. Gold changed from being, okay, something that you bury in your backyard or you're having in your vault, along with ammo and seeds for when the hard times come. Bottled water. Right. Meals ready to eat. Gold and lead. It becomes a security. And its meaning changes from that, you know, apocalyptic bottled water.

36:08Right. And the meaning of gold today is as an insurance policy, a security against central bank error or government error. That's the meaning of gold today. and is that the dominant narrative that you're identifying as gold rallied over through 4 000 absolutely so and we've really been able to track that one in specifically so here's the really big here's the million dollar or trillion dollar question how do you identify a narrative and say oh gold is going to double from here based on this narrative or or are we not there yet No, you identify the narratives because the stories don't ever change.

36:50So the story, leave gold aside for a while. Think about, we're talking about you're bullish on company XYZ because. There are about, I don't know, depending on how, again, how finally you want to resolve that, there are only about a dozen stories for why you're bullish on something. They can be management change, top line growth opportunity, consolidation in the industry. Upcoming catalyst. Catalyst. So every catalyst story. New product. New product. FDA is going to approve the new drug, new molecule. So that story, you just change the name. That's the same story that's repeated over and over again about any pharma or biotech company.

37:39The stories, we think that they are amorphous and variable. The fact is that the core of the story, what we call the semantic signature, the meaning of a story, they're amazingly constant over time. So what we're looking for is for – it could be dormant for a long time. But what you want to know is when that story starts picking up again, when someone starts playing that story, when it appears on Kramer and starts happening in the financial press. That's the stuff we can pick up with real precision. So it's both the stories that are starting to fade. But I think the really interesting stories are the stories that have been dormant for a long time and they start picking up again.

38:33You are reminding me of Campbell's Hero's Journey, that there's only so many. My hero, right? There are only so many stories in the world. And that's a, now, I like to talk about in Hollywood, famously, there are only like five scripts. That's right. Right? Tolstoy is supposedly Tolstoy. He said there are only two stories, that a man goes on a journey or a stranger comes to town. Those are the only two stories in the world. Not quite, but he's pretty close. You're on the right track. Right, right. The point is there's a finite number of stories. You can drill down, so you can get a couple of dozen about any sector you want to talk about.

39:18But it's a finite number. And so what we do, and what I think is really interesting, is to track that finite number of stories. And you're right, it's not just around markets. We track several thousand of these stories today. So let's delve into that. So how do you have this massive database? You're sucking in every news feed, magazine, newspaper, anything that you could quantify and run into a linguistics model. What's the process for analyzing this? How do you use artificial intelligence to go through this? And how do you make sense out of that heap of how do you find signal amidst all that noise?

40:08The crucial thing is you can't just ask AI an open-ended question and say, what are the narratives for this company or for this sector? Don't do that. And this is a mistake that people make all the time. They ask open-ended questions of ChatGPT or whoever. The problem is ChatGPT will give you an answer. Just not a good one. Not a good one. It'll hallucinate a lot, right? It'll go out. It'll find its own data. The secret to using AI successfully is to take this magic genie, because it's a magic genie, and you stuff it into that bottle, right? You do not let it out. You constrain it dramatically.

40:52You don't let it go out and find data. You give it the data. crucial thing you don't allow it to think you tell it how to think so the the most important step that we do is we don't ask ai what are the narratives that you look at we tell it this is human direction you have to have human control i'm hearing data set is uh controlled by you as well as the thinking, the prompt. Yes. And it's more than prompt, right? So it includes prompt, but the phrase that's used in this world is called not prompt engineering, but context engineering. Okay. That makes sense. So you want to control everything around the AI because you You want to limit it to being that linguistic calculator.

41:50You want it to be your operating system. That's really the way. And if you do that, then it will give you the same answer twice for the same inputs and the same question. That's the crucial thing. So it's consistent. For it to be consistent and for it to be real signal. So this is a human-directed process. You can't ask AI an open-ended question. You have to control all the inputs. You have to control the output, meaning you judge it. You run it back through a different AI system to say, how'd they do? Did they go off the rails here? But the most important thing is you have to give it the scaffolding.

42:28You have to give it the skeleton. You have to tell it, these are the thoughts you are allowed to think about. And those are the signatures. I've kind of learned I have to avoid asking questions that have a human emotional subtext. Like, tell me what was most surprising about this. It doesn't know what a surprise is. Tell me what was most interesting about this. It does. It's not able to do that. You really have to treat it like it's a dumb machine. Well, that's right. This is why I mean you treat it. You need to treat it as an operating system. you need to constrain every bit about it, particularly in how you allow it to think, because it wants to please you so badly.

43:17It does. So if you ask it what's interesting, it will look back at its history of communication with you and it'll think, what will Barry find interesting? And it will give that answer to you. And if it can't find it easily, it'll make it up. It'll make up an answer that you will find interesting. I find when I try and prompt, hey, tell me about Ben Hunt's background and give me the timeline of his career. That it's good at. Hey, what was Ben Hunt really good at? It's got no idea. So what you're able to do if you're able to, again, put the genie in the bottle, tell it how to think about a problem, is you're able to identify, and this does go back to the work from 35 years ago, the type of stories we tell, that we humans tell about stocks or politics, we tell two types of stories.

44:15We tell descriptive stories. Oh, the Fed cut rates by 25 basis points today, a descriptive story. And we can also tell the description of the, you know. It's the because clause. Because we're seeing slowing, increasing layoffs and slowing consumer. And that's high resolution and very descriptive, right? The Powell was surprisingly hawkish today. And he was. That's a description. There's another type of story we tell, Barry, and that's prescriptive. Meaning? Meaning the Fed should be hawkish. The Fed should cut by 25 basis points. Those are the stories that are indicative of an effort being made to move public opinion in a certain direction.

45:17Mm-hmm.

45:47identify the stories that are trying to tell the reader how they should think or how policy should go. That we find has a lot of predictive capability to it. So you're in the business of analyzing the world's narratives every day. How is that even possible? It seems like that is an impossible task. Crazy, right? And it used to be. It used to be crazy. I mean, Barry, I really do remember in the academic days I would literally hire grad students and give them a cup of dimes so they go down to the microfiche machine. I remember those machines in the library. You remember those machines? Yes, yes.

46:34Remember those machines? I would hand code the or hand record the coded data. I would type it into remote access for a digital equipment mini frame, and the next day, you know, something would churn out for me. Today, it's, they say we're processing hundreds of billions of tokens. We get millions of documents overnight like that. And we have infinite computing resources available to us. Is there anything you can't access that you wish you had access to? any data source so one of the things that's happened on wall street is that the banks and the sell side have become very jealous of their publications because they tend to think and they're wrong that they're good at it right they're that they're good at analysis i personally don't think they are well let's just say some are better than others some are better than others, but none of them are really, if there was, I'll call it, kind of significant alpha there, they wouldn't be a bank.

47:49They'd be a hedge fund. Yeah, they wouldn't be on the sell side. Now, so I'm interested in reading the sell side research, not because I think there's some nugget of truth in there, but because I want to see what they're all talking about. Right. It's reflective of, if not a consensus, certainly a popular set of ideas. And this is a crucial thing to talk about what we do. I don't know what the truth is. Right. I have no idea. Does it matter? And I don't think it matters. I want to provide this information to people who do have a view on the truth. Let's say you're a value investor, right? You're running a fund.

48:28You've got your views. You've done your homework. You've done your research. You've got a good back test. Yeah, you've got – so I'd say I've got – here are the companies where I think I've identified something special, something that's valuable that the market does not recognize. And so I want to buy it, and then I'm just going to wait. I got to wait until one day the market realizes, the market comes to their senses and says, oh, wow, that should trade at a higher multiple or a higher price because that special thing that you saw, that source of value, the rest of the world comes to see that.

49:08Well, what I can think I can show you is when the rest of the world starts to wake up to whatever it is you're looking for. So you're catching the early lift off the bottom. That's the goal. When a value investment only works when the market recognizes it. And we are tracking when something like that gets discovered by the market. So before NVIDIA is$5 trillion, when it starts ramping up to$500 billion, hey, something's going on here. what i found in my experience as an investor is that you make the most money on a trade during what i call is the discovery phase of a trade when the rest of the world wakes up to something that you had noticed and identified before that's when the money is made once it gets out there.

50:04Then the market is eventually efficient. It's a different risk and reward profile. Let me put it that way. You get a lot more ups and downs post that discovery phase than you do when you're enjoying the discovery phase. It's a lot harder. Now, we may say, oh, but there's going to be this other catalyst. And then you say, well, But that discovery phase is, I'm going to quote Doug Cass, that's the, there's a phrase he uses, it's like a contrarian perspective, a variant perspective where you have some insight that is not widely held. Right. And the market being mostly kind of sort of eventually efficient, if there's a truth or at least a good story in your variant perspective, the market gets it.

51:00All it takes is a story. Right. All it takes is a story. It's got to be a good story, though. Yeah, it's got to be a compelling story. Right. Right. And that's why you're looking for a CEO who can tell that compelling story. You're looking for the ability to tell a compelling story because that is what gives you a multiple. Multiple is story. So let me ask you a few more questions on Persean before we go into a few other areas. So first, you've been doing this for seven years. What are you doing today that was unimaginable four, five, six, seven years ago? I'll tell you something that was unimaginable really two years ago.

51:43That's amazing. And that is the AI. So what we were doing in our early days, we were basically doing small language models. We were making the language models essentially by hand. And, you know, we did our first models and our first versions of this, and we licensed it to some big banks. And here's the problem, Barry. We had constructed a net, and it was a pretty good net. I mean, when we would dip it into a data stream, we'd catch a fish, meaning a signal. And the signal, oh, that signal works. You know, good hit on the signal that we put up. The problem was we were missing too many damn fish.

52:24Our net was too small. And we'd say, oh, we got this one. And then we'd look back at whatever it was we were designing the model. So we'd say, well, how did we miss all these other fish? And the answer was the small language model we were constructing, it's incredibly complex if you're building these models without the probabilistic approach that modern LLMs allow you to take. So this is the whole notion of embedding so that there are a million ways to say I'm bullish about the management change at company XYZ. There are a million ways you can say that. And if you're in the business, as we were, of kind of handcrafting, let's write down all the ways you can say that.

53:12Oh, really? You miss a lot. You've made a small net, and what you're trying to smack. And a small data set. And it's a pretty small data set. So we rebuilt all of our software, again, using AI as an operating system. context engineering, control how you dole out the text data, how you test it, how you allow it to think. And we thought, all right, you know what? We're building a bigger net. I bet we see five, maybe even six X improvement in our signal. And what did you end up with? Over a hundred X. That's unbelievable. Over a hundred X. It's over. And now it's, we've had a multiple of that again.

53:58It is hard to describe. And the expansion of the net is in so many different directions. Everything we were doing before was just English language. So now you're global, and how many languages are you pulling into the novel? Anything that an AI has been trained on, we read it. Is that 40 languages? How many languages? Well, there are about a dozen languages that are useful in markets. So it'll be Japanese, Chinese, a variety of European languages. So we can tell... Indian. Here's the story about Chinese domestic markets. There's a Western story, Western narratives. And there's a domestic Chinese story.

54:44And the domestic Chinese story. Which is very different. Often it's very different. So we were able, for example, picking up... We were able to pick up way before it got picked up in Western press. The demand for luxury goods in China went off a cliff in last November. Listen, you can't go from a double-digit GDP to, like, low to mid single digits and not have it affect, especially in a country like that. But it's interesting, right, because they've had ups and downs on business cycle before. But this is the first time where you saw consumer behavior that really was kind of similar to what you might find in a Western consumer behavior.

55:31Point being, you didn't hear that from LVMH or the Macau gaming guys until February. And we were picking that up in November from the domestic Chinese media. So that raises a really interesting question. You know, at the end of the day, clients want to be able to make money on your research. How are people putting this to work? How does what you're building help your clients generate alpha? What we think we've discovered is an entirely new source of data. And what we're confident is that we've built the systems that are very different from what you see with this sort of analytics that are out there from anyone else.

56:17Right. So this is not sentiment. Right. We're not tracking or using mean words or nice. By the way, that was a big thing. I was that 10 years ago. We're we're scanning Twitter to identify investor sentiment. Oh, my God. I mean, but I was just looking today and not to pick on Bloomberg, but, you know, it was their live coverage of the Fed statement. Right. And they were meeting. They were analyzing the sentences in the Fed statement for hawkish and dovish sentiment. How it changed from the last meeting. So you're saying there's not a whole lot of signal there. I want to be careful with what I say, which is that there are a number of, I'll call them high-frequency stat-arb guys, where if it's important for you to note the difference in word choice on a millisecond level, I think you can get something out of that.

57:15I do. And so there are firms that can do that very well. That's not our game. Right. This is not the what we're trying to identify is not just sentiment, not just word choice. This is not Google Trends. And how many times did they mention AI in the earnings report? There's we're able to track the actual stories that drive behavior. So that that's the next question. And I'm going to give Dave Nott a credit for asking this. does every narrative turn into a decision? How do you know when something is merely noisy versus where there's a significant tradable signal? I don't, right? So this is, my goal is not to, oh, you know, do the trades.

58:07My goal is, Mr. Hedge Fund guy, Mr. Asset Allocator, You know China, right? You know your companies. You know your commodities. Here is data that... I think you'll find useful. The efficacy of it, though, is up to you. I don't know what you want to do with it. I want to sell the picks and shovels, honestly, for this vast new data set that we all know is important, but we haven't been able to measure it in a very predictive way before. So let's talk about a few things related to that. Can I say one more thing? Sure. And it's not just about using this for investment. So we have a product for financial advisors, which is your client's coming in.

58:59You've got a portfolio. You need to be able to say, what is my client? What are they worried about? What are they nervous about? What are they hopeful for? What are the stories they're reading? And you're pulling this out of the flow of media. And we can tell you exactly for the portfolio you've got for that client. Here's what they're going to be asking about, worried about. And here are the answers you can give them to show this is when it's happened before, when this story has come up. Stay the course. It's going to be fine. This is the sort of stuff we can do for financial advisors. Huh, that's really interesting.

59:32I would love to see some of that. It's not just in markets, Barry. I got to tell you, the policymakers, corporate executives, bankers. So we did before in the run-up to the Russian invasion of Ukraine. And we published this on Epsilon Theory. And this is my old academic work, the book Getting to War. Before a country starts a war, they mobilize public opinion. Right. And so we were looking at domestic Russian media, looking and saying, friends, this isn't going to be a limited thing. This is happening. This is happening. I mean, it's going to be a full-scale invasion because that was the messaging in domestic Russian media to the Russian people.

1:00:16Wow. So brands, right? So we're working with some, it sounds great, but pro sports teams. You want to tell a story to build a stadium. Sell some tickets. Sell some tickets. Build a stadium. Give me some examples of people who have told good stories. And how did that work for them? How can we do the same thing? So let's just start looking at storytelling. If Jeff Bezos was a subscriber to this back when he was trying to build a tax-funded HQ on the Hudson, had he had your data and you were crunching all the New York City news stories about this, might you have been able to give him advice that he suffered such a backlash?

1:01:06Because you know what we have today? we have polls, right? Which are terrible, which are terrible, which are mostly terrible. Now, you know, and I love the poly market stuff and other things where you try to get as many people as possible to put money on something. Right. So, you know, when you ask a person a question, you're asking them, hey, what do you think you think? And what do you think you're going to do in the future? When we know people are terrible at both of those, terrible at both of those. But if you put a little money on it, all right, maybe they might be a little more circumspect.

1:01:35That helps a ton. So in places where you can make a bet, I think that improves the kind of information you can get. It still lends itself to a lot of manipulation and a lot of issues around it. Listen, polymarkets and Calci and all those things, it's not the bond market. It's not a hundred and something trillion dollars. It's a couple of bucks on each of these. And sometimes a million dollars moves them or half a million dollars can move a market. But for a lot of things, let's say you're polling for a political candidate, right? You can't ask them to put money down on something, right? Right.

1:02:09But you really want to know, these policies that my candidate is thinking about taking on, is that popular? Does it resonate? Is it a compelling story? We can absolutely see if that is true by looking at local media, local social media, all of that. So it's pretty wild there. I mean, once you start looking at how important stories are, and once you've got a tool where you can actually measure them and visualize them, it's like, I feel like it was like when they invented whoever was Lovenhook or whoever invented the microscope. When you're actually to see something that we all know is there. Well, now we do back then.

1:02:52Remember, germ theory took, what, a century to catch on? It took a century. And it takes seeing it, right? It takes an instrument to actually measure it before you actually believe in it. And so I feel like that's kind of where we are right now. It's these early days, but to actually see and measure the storytelling at this level of resolution, this magnification, is pretty freaking cool. So we're having this conversation with market at all-time highs, and you've written about the ravine. Yes. Tell us a little bit about what is the ravine? How does your data identify that? Tell us what this means.

1:03:31Well, there's clearly been a change in policy regime out of Washington. New administration and a radically different approach. Even from the first Trump presidency. Even from the first presidency. And so what we're able to measure, really measure, is how does that, I'll say, play, but also what comes – narratives never happen in a vacuum. There's always a counter story. For every bull story, there's a bear story. And you can almost kind of see the battlefield of ideas, the battlefield of stories and how it emerges. So my strong sense, Barry, is that we are going towards politically in this country towards trench warfare, greater and greater, I'll call it narrative violence.

1:04:34And if you look historically how that plays out in countries, it doesn't play out well. Civil war, domestic political violence, things like that. Sadly, yes. Exactly like that. Exactly like that. So there's that element, and so that's a sad one or a very troubling one in trying to say, well, how do we navigate that? But even when in markets, and I alluded to this earlier, how capital markets have become a political utility. And the role of markets in our society, you can really see how that changes in the stories we tell ourselves about the role of markets, what it's there for. End of day options and speculation.

1:05:25What else is it supposed to be there for? But that's kind of what I'm getting at, Barry. You can see an enormous change in the meaning of markets, and it connects with – yes, they'll call it the speculation layer, but it also connects with financial nihilism, YOLO. So it it it leads to a very, I think, less attractive future for how we think about money and the role of markets and the role of capitalism. Coming up, we continue our conversation with Ben Hunt, president and co-founder of Persiant, discussing how money managers use their output to generate alpha. I'm Barry Ritholtz. You're listening to Masters in Business on Bloomberg Radio.

1:06:41I'm Barry Ritholtz. You're listening to Masters in Business on Bloomberg Radio. My extra special guest today is Ben Hunt. He is president and co-founder of Persean, a data analytics, narrative storytelling, large language model using artificial intelligence to find some signal amongst the noise firm. Uh, their clients range everything from large, uh, money managers to hedge funds to academics and corporate America. So you write Epsilon theory and some of it is for subscribers. Some of it is public. A lot of it's public. Yeah. One of the things you wrote is absolutely my favorite item from the first few months of the Trump presidency, because I like to talk in probabilities because I don't know what's going to happen.

1:07:36But here's the best case scenario. Here's the worst case. Here's all the middle scenarios. You wrote a piece, the end of Pax Americana, that I thought was the most cogent, imaginative, well thought out. Hey, here's the worst case scenario. And we are becoming dangerously flirting with this possible outcome. And I use that as, all right, so here's what I think is high probability. Here's the best case. But if you really want to think about how this can go off the rails, check out what Ben wrote. And so tell us a little bit about how did Percian inform the end of Pax Americana, because I get the sense that domestically, that wasn't really how a lot of people were thinking, but I got the sense from overseas that was a much more common thought.

1:08:33Tell us a little bit about both the piece and how the data informed it. I'll start with the, I'll call it the international dimension. I want to come back to the domestic to mention, because I think we've got some really interesting data recently to share. It goes back to this, again, I cringe when I talk about it, but it is game theory, right? And all game theory is, is strategic interaction. It's that the United States, yes, is the most powerful player on the world stage, but every country has some degrees of freedom and some autonomy in the policies that they implement. There is not a dominant strategy, meaning an outcome, an equilibrium outcome.

1:09:23Again, I hate using these words, but I'll use them anyway. All an equilibrium means is it's a balancing point where both parties, let's call them the United States and a European country, where do they end up where they all say, okay, I can live with this, or I can live with this. And the America First set of policies, and it's all these set of policies, what you end up with is less potential economic growth, trade, all of these things. All those things that we enjoyed post-World War II, that realignment that imbued to our benefit so greatly. Enormously to our advantage. Our advantage, yeah. Enormously to our advantage.

1:10:09It goes by the name of soft power and the dollar system, the Bretton Woods system, all of this. Reserve currency on and on. Reserve currency to finance our deficit. All of this is enormously to our advantage. And yes, there are free riders on that system. Yes, there are costs to that, particularly on defense and some other areas. Tariffs are another area where there's absolutely free riders on that. So there's improvement that can be made in that system for sure. But this is a different system. This is a different set of rules of the road, and it leads to a different strategic interaction between countries.

1:10:49So that's what the note was about. And I'm not trying to predict. I'm just trying to observe, which is a great line actually by George Soros, which is, I'm not predicting, I'm observing, which I love that as a line. I'm not and our technology is not there to try to predict the future is trying to tell you what is true in the present today. I'm amazed how many people think they can forecast the future when they have no idea what's happening. I want to I want to now cast is what I want to do. Not not forecast. I want to now cast. And I guess what I wanted to talk about kind of domestic. So we started tracking the different narratives around immigration policy early last year.

1:11:41And what you saw in some of the kind of – This was during the run-up to the election. Yes, exactly. And what you saw in really going back – because we take this stuff back for a decade or more. And what you absolutely saw up to last October, there was an event from last October, not the election, but an event from last October. You see a steady increase in regardless of your political affiliation. You know what? Immigration isn't working for us as Americans. So a steady increase. starting with the they're eating the the cats they're eating the cats they're eating their dogs they're eating the dogs when you saw the Columbus moment when you that that moment one of the most that's the real moments in debate history and we see very clearly in our media data and this is this is not mainstream media this is everything that we're pulling up we've seen a really significant decline in the volume and density of oh my god immigration is a problem we need mass deportations on the contrary we've seen an enormous increase again regardless of political affiliation including republicans uh-huh uh no immigration is a good thing for this country this is the stories of america that are pro-immigrant and immigration.

1:13:13Now, you would not believe that if you were looking at the policies that the White House has implemented during these first, you know, 10 months of the administration. So wait, when did you first notice that? Regardless of policy, we think immigration is a net benefit to America. When did that first start showing up? It started changing right in October because it was like a bridge too far. It was like, this is just stupid. This is silly and stupid. And it's been a steady increase. You know, and you wouldn't you wouldn't believe that if you were immersed in Twitter. But but so this country is actually quite pro-immigrant.

1:13:57I know that sounds correct. We are a nation of immigrants. Here's another data point that's kind of mind blowing. We were talking about a different data point in the entirety of the U.S. history. 2025 looks like the first year where the U.S. population will decrease. So it's a decrease in legal immigrants, not not only illegal immigrants, but legal immigrants add that with the deportations and just people staying away. Well, this gets back to the economic picture and from other – so what you've had is a clear – and again, we see this in our data from other countries. There's a clear effort to repatriate assets and funds away from the U.S.

1:14:43There's been a clear effort outside the U.S. You see this with central banks but also with – Hence gold, the big lump in gold. Correct. Treasuries used to be a safe haven asset. Right. No more. That's that's it's you understand how significant that charge is that you're making. You're basically accusing the president of submarining one of the single greatest assets America holds. Well, that it's a it is built up to such an extent. Right. The way I think this is an iceberg that is melting, but it is melting. What we do not see, we do not see capital flight. Right. We saw it for like a week in April and then it quickly reversed.

1:15:29That was it. So there's no capital flight. There are no U.S. investors leaving the country. There were fears, and we can track it. If that starts to happen, we'll see it immediately. That's not happening. Repatriation continues to happen. Slowly, measured, balanced. Melting iceberg, right? Because there are limits to if you're a Bermuda reinsurer. I mean, you're stuck with treasuries. Right, right. You could buy some gold to offset it, but you can't sell$100 billion worth of treasuries. You can't. You can't. So it's a melting iceberg, but we can clearly see that. The dog that's not barking yet, and maybe never will, is capital flight.

1:16:11What we see politically, domestically, is that actually, and this was validated by a Gallup poll they've been doing 20 years, which also showed what we had started seeing a lot earlier.

1:16:26where immigration as a thing is actually pretty darn popular in the United States. Yeah, but so is restricting assault rifles, and we can't get any change on that. That's 75%, 80 % or the extra large... I got to tell you, that depends very much on how that question is asked. Well, that's true for all polling. Well, this is the benefit of what I'm talking about. So when we're doing these, we call it the semantic signatures. It has nothing to do with how you're phrasing the question. We're not doing polling. We're seeing what people are actually the meaning of what they're talking about in media.

1:17:05So how does that play out if people are legitimately saying, no, immigration is a good thing for America? Is there an impact on population and the economy? Is there an impact on markets? Is there an impact on policy and politics? I think there's an impact on the election, the midterms next year. So we're talking literally 12 months from now. Yeah, because that's how this stuff gets – on the political front, narratives and opinions, they get cashed out in elections. Markets are different. You cash stuff out every day. The feedback loop is so rapid with markets. Exactly. politics is a different story so it gets cashed out in the election if you were to ask me before this conversation what's the most significant impact on the midterm elections there was just a gallop poll yesterday um gop questions on the economy they were plus 14 two years ago they're minus four percent this year i saw that and and that's an amazing swing and i'm i'm saying to myself, you know, listen, the out of power party usually picks up, you know, 10, 15 seats.

1:18:23You have the redistricting question, which may blunt that. But if if the most important question during the election was on the economy and that's an 18 percent swing, this is looking like a pretty substantial shift. What I'm hearing from you is, hey, this isn't just about the economy. It's not. It's not. So it's the economy, it's immigration. What else? What else are you seeing that's interesting? That's pretty. Now, there are other aspects, though, where the stories, this is people talking about immigration as a thing. Now, whether that gets translated into a political party position, because I've got to tell you, the Democratic Party is enormous.

1:19:06There are no narratives that are being put forward by the Democrats that are powerful or popular. Other than the mayoral contest in New York. Yeah, right. That is one. What I'm saying is that many of the policies that are being presented by this administration are unpopular, not just with the Democrats, but with Republicans, and increasingly so. Immigration being one of them. I think economy being another one. Tariffs aren't really popular amongst – people perceive that as a tax increase or pressure on small business. government shutdown, right? So both in polls and also in this, it's that, because the question, oh, well, the Democrats will get blamed.

1:20:00Well, that's not really true. That's not what's happening. Now, how that all pays out - And by the way, listeners should know - It's a long way off. You are not a Democrat. Oh my God, no. Like, I know you and your politics and the stuff you're saying, I could hear people shrugging and saying, oh, that Ben Hunt is just a liberal Democrat. I'm like, no, no, no. Well, that's not who Ben is. No. No, it's not. You're just talking about here's what I see in the data. I'm saying I'm observing. I'm observing. And I think that a lot of times if we are, like me, very online and on Twitter too much, you don't see the broader picture of what is happening on blog posts and local newspapers and everything else.

1:20:49So I think having the ability to read everything and track these stories that, you know, they don't – like I say, they wax and wane, but they don't ever go away. It's something we're very excited about to track the stories of America. I'll give you another one. And this one actually, I don't know what to do with this. One of the stories of America is that America has raised the world's wealth, right? Global standard of living. Global standard of living, that America has been this powerful force to, and capitalism and America have been this powerful force to raise people out of poverty. That's a story that in other periods of time has been prominently talked about.

1:21:53Very resonant. That story is nonexistent today. Well, you cut things like a few million dollars to inoculate all of Africa from HIV. That's going to have repercussions. Well, it's not just that that has those repercussions that you're describing. What I'm saying is that the story that would support that, it's gone. Is it gone forever? No, no. These things are never gone. These stories, stories have, they never die, right? They're waiting for a new force to give them life. The next version of it. If U.S. is not seen as a force for raising people out of poverty around the world, do people see China as that force?

1:22:45Haven't looked at that. We've been looking at the U.S. story. But my personal sense is that we've essentially seeded the field, CEDED. With a C, not an S. Yeah, with a C, not an S, particularly in South Asia and Africa to China. I mean, that seems pretty. You know, I'm glad I asked the question about the end of Pax Americana because I just found that piece so insightful and so useful. And I could keep you here talking about this stuff for hours. But out of deference and respect for our listeners' time, I'm going to jump to our favorite questions that I asked all of our guests. But before I do, there's a question that I have to pose to you, which is so you get to see things before they sort of bubble up into the mainstream, before they become a well understood narrative.

1:23:47What's a narrative that you're just starting to sniff out that most of the world or most of the country or most of Wall Street and finance hasn't seen yet? It's a story that people are talking about since Tricolor and First Brands went out. So it's a story on credit, story of credit. I mean, First Brands is legitimate. It sounds like felonies were happening there. So we'll have to see how that plays out. Tricolor as well. But my point is that every day you have a new bank CEO come out and say, no, no, everything's fine. So yesterday it was David Solomon at Goldman Sachs saying, no, no, it's fine.

1:24:37Today I was hearing the Brookfield guy saying, no, no, it's fine. For every story, every interview, the guy says it's fine. You're getting two articles in the FT or the journal saying ain't fine. So you think thou doth protest this? Well, I am. All I'm saying is I am observing. I'm not predicting. I am observing the level of volume for alternative asset managers, their exposure to private credit. This is problematic. Oh, I'm wondering what's going to happen when the music stops. Oh, I think that this blows back into the commercial banking system. Those stories, those narratives are higher by an order of magnitude than they've been at any point in the last 10 years.

1:25:29At any point, including during COVID, during when you had similar concerns over, oh, my God, private credits. COVID was easy to rationalize everything. Hey, listen, we're all frozen. Just ignore it. But this is a story that has legs. It is growing in a way that I rarely see. And once a story like this grows, it doesn't just go away. And it's not fixed by Bernanke saying, oh, subprime is contained. Right. It's immediately what I thought of as soon as you mentioned that. Absolutely. It does not get fixed by people saying, don't worry, there's no problem. Jim Grant said he was right. It was contained to earth.

1:26:10The rest of the solar system was fine, which turned out to be a very witty and clever observation. This story, this worry, this concern absolutely has legs and we're seeing no signs of it easing off in financial media and press. So the big question is, is it systemic or is it the specific? I don't know the truth. I don't know reality. All I can tell you is— We're seeing more of this. Well, we're seeing this is the story. Really, really interesting. Let's jump to our favorite questions, starting with, who are your mentors who helped shape your career? Well, I mentioned that the people both in undergrad and graduate school who turned me on to the science part of political science.

1:27:04And in particular, in graduate school, Gary King, who runs the whole social science research center up there at Harvard, and has for a long time now, he wrote the original, really the book on inference. You hear about inference all the time now. Well, Cialdini, of course. So the notion of inference, taking large data sets and pulling out. Oh, inference. I thought you said influence. No, no, no, no. Inference. Inference. So the science of inference, I learned that from Gary 30 years ago. And it's so interesting to see that come full circle because that's at the core of – Jensen Wong is always talking about it and the whole notion of AI, that inference spin they talk about.

1:27:49I was there at the beginning for how you – what inference is and why it is so powerful. That was a huge influence on me. Let's talk about books. What are you reading now? What are some of your favorites? I'm a science fiction guy. Okay. I really am. You're talking to another sci-fi guy, so let's give us something new and something classic. Well, something classic would be Liu Shixin and the Three-Body Problem. Right. By the way, the Netflix show on that was surprisingly watchable. I thought it was excellent. Before that, and I named a company after him, was the Foundation Trilogy. Asimov. With Asimov.

1:28:30Those books hold up less well, honestly, and I thought the series had its moments. I didn't love the series. It was pretty so-so to me. Current science fiction, Rebecca, she goes by R.F. Kuang, and she wrote a book called Babel. And she's got a new one out. Two B's or one B? B-A-B-E-L. I like Tower of Babel. Right. B-A-B-E-L. And she's got a new one out, Catabasis, I think it is. But science fiction, dealing with language and linguistics, I love that stuff. That sounds like that's right in your sweet spot. Let's talk about streaming. What are you watching or listening today? A podcast or Netflix or whatever?

1:29:13So honestly, I don't listen to any podcasts. Isn't that terrible to admit? Well, if you host, I will say this. If you host a podcast, you're either preparing for a podcast, doing a podcast or ordering a podcast. So it's like, all right, that's three or four. That's all it is. Every now and then I'll catch something because I want to either listen to this specific guest. Like I will not listen to any podcast that you're on before we do our podcast because I don't want to steal anybody else's narrative or questions or line of thinking. So that's purposeful. But every now and then. So I listen to the things I listen to are like John Pizzarelli's Radio Deluxe.

1:29:56Yeah. which is sort of like a podcast slash music series. Something a little different. Exactly right. What about streaming? What do you watch? Well, you and I are both big Godfather fans, so we love all the mobsters. Have you seen Mobland yet? No. All right. Worth your time. Okay. Pierce Brosnan eats up every scene. Tom Hardy is awfully good. I just love him in anything. I don't know if you ever saw Peaky Blinders, right? I started that and I kind of couldn't get into it. So my issue is I have to watch something that my wife tolerates. I have friends who he goes into this room she goes into that room they don't watch anything together.

1:30:40I don't know how that works. She will watch some stuff without me I will watch some stuff without her. But 80 % of what we watch so I have fallen down the British upstairs, downstairs, Gilded Age. So we watched The Crown. We watched The Gilded Age. We finally... Let me give you one. We went back to Downton Abbey, which I missed when it first rolled out. Check out... So my guilty pleasure, and I do watch this with my wife, is Diplomat. The new season just dropped. New season, yeah. And that's teed up for this weekend. I'm looking forward to that. Excellent. The first... Was this season two or season three?

1:31:19The first season was great. Yeah. And I'm going to give you, if you like that, so my wife finds these really interesting. She got me into Killing Eve, which was a little more spy fare than Diplomacy. And if you get a chance to watch Slow Horses. Oh, that's the one I've got to see. Yeah. It's really. So the first season is great. And people have told me the most recent season. It some people will say, I don't like the second or third. I know people who loved it all the way through. But it's it's really it's really an interesting, well told, beautiful cast. Fantastic. Yeah. You'll love that. Our final two questions.

1:32:03Yeah. What sort of advice would you give a recent college grad interested in either investing or working with large language models, working with narrative analysis and artificial intelligence? Hmm. Don't. I say that tongue in cheek. You had a lot of fits and starts going back seven, eight years. For sure. And I tell you what I think is important.

1:32:42You have to, what I think you can accomplish in academia, either in grad school or whatever it is, is you build your intellectual capital. And so I think a lot of times when you get out of college, you say, okay, I'm just ready to kind of live my life and start something. And you haven't built that intellectual capital yet. The issue is that once you enter particularly the investment world, where if you're responsible for managing other people's money, brother, that's it. Right? I mean, that's got to be your total focus. You are spending your intellectual capital. You're not gaining intellectual capital once you take on a role like that.

1:33:35So my first advice is find a path that's often in academia where you're building intellectual capital. because once you leave that environment... Then you're spending it down. You're spending it down. Makes sense. You're spending it down. Our final question, what do you know about the world of fill-in-the-blank? Investing, data analytics, narrative storytelling today would have been helpful 25, 30 years ago. In investing, I wish I had understood the role of... Let me step back a second. I think that whether you're talking about data or whether you're talking about investing, and I'll speak for myself, but I think there are a lot of people like me, we think there's an answer with a capital A right there in the numbers.

1:34:36And if you just look hard enough and if you work hard enough, you'll find that answer right you'll find the secret the secret formula and you have learned since then ain't no such thing right now there is magic and there are patterns but it's it's it's not in the structured data it's not in the numbers it's actually in the error it's in the probabilities it's in we're calling you know the stochastic element right the the role of chance and understanding that there are patterns and there's real magic in understanding that. I didn't get that when I was either starting with data analysis or with investing.

1:35:22I was looking for the answer with a capital A as opposed to a process with a capital P. I love that answer because – so my exercise in confirmation bias is I love the fact that you looking back, We all have the benefit of hindsight. And unfortunately, that can be a bias to certain people. But when you're looking back at it, you know how it happens. At the moment when you don't know what the outcome is, thinking about it probabilistically is a much healthier approach than saying, here's a binary up or down, yes or no, and I'm either right or wrong. And I see people struggle with that constantly.

1:36:06Constantly. When I first started in the investing world, again, late, I got two pieces of good advice, right? One was never go all in. Right. Which is really interesting because it's this business of investing, it's a wonderful life. You're resolving puzzles. We meet interesting people. We get to have these sort of conversations. It's a career for a lifetime.

1:36:33And the two pieces of advice are really, never go all in. And also, reputation is absolutely the most important thing. And again, especially when you're young, you think, oh, well, that's not the most important thing. You know, being right is the most important thing. It ain't. It's playing the long game here, which you want to. It's never go all in. And the reputation is - By never go all in, you mean never put everything at risk so that if it doesn't work out, you're out of the game. You're done. Stay in the game. Was it Gerald Loeb's book, The Battle for Investment Survival? Stay in the game.

1:37:15Avoid that risk of ruin. Or don't ever risk ruin. That makes sense. Because once you don't think in those terms, but that's really all in means risk of ruin. It's the risk of ruin. And it also connects with reputation because once you. Once you blow up, it's tough to go back. And you start saying, oh, I can fix it by taking this shortcut or doing this other thing. And once you do that, it never ends well. All right. And you tell yourself, oh, just this one time. It's never this one time. Ben, this has been absolutely delightful. I'm so glad we finally got around to doing this. I'm just entranced by your thought process.

1:38:05Some of the things like I love reading stuff of yours that I totally disagree with. And then because it forces me to say, well, he's not just making this up. I know how your brain works. And it's like, all right, if Ben is saying this, then I'm going to make this my worst case. Because it's easy to dismiss it. Everybody goes out seeking confirming information. And rather than being dismissive of it, all right, you claim to think probabilistically. Where does this fit into the range of probabilities? And once you start thinking in those terms, it's like, oh, so the worst case scenario is worse than my worst case scenario.

1:38:44I got to move the bottom of my range down further because if this goes off the rails, this is really bad. That's what your Pax Americana piece did with me. Well, thank you. And it really helped me figure out how to think about, especially in that week between April 2nd and 9th where everyone was losing their mind. It's like, oh, no. So they're losing their mind because, hey, this could happen, but maybe something good comes out of it. Trying to avoid tunnel vision, I think, is so important in our business or any business, but especially our business. In our business in particular. Well, thank you for being so generous with your time.

1:39:21We have been speaking with Ben Hunt. He is the co-founder and president of Prescience, and you can find his writing at Epsilon Theory. If you enjoy this conversation, well, check out any of the 593 we've done over the past 11 and a half years. You can find those at Bloomberg, iTunes, Spotify, YouTube, wherever you find your favorite podcasts. Be sure and check out my new book, How Not to Invest the Ideas, Numbers and Behaviors that Destroy Wealth and How to Avoid Them at your favorite bookseller. I would be remiss if I didn't thank the crack staff that helps put these conversations together each week.

1:40:04Alexis Noriega is my video producer. Sean Russo is my researcher. Anna Luke is my podcast producer. Sage Bauman is the head of podcasts here at Bloomberg. I'm Barry Ritholtz. You've been listening to Masters in Business on Bloomberg Radio.

1:40:29participates. Thank you.