Risk of Ruin
Audio documentaries about risk takers.
Risk of Ruin
Big. Chunky. Changes.
Citrini Research has taken Fintwit by storm with a series of prescient market calls covering themes such as AI, GLP-1, and the Trump election.
In this episode Citrini discusses how he arrived at thematic investing, the process he uses to find baskets of stocks, and also what it’s like to write up a thesis for tens of thousands of subscribers.
Find Citrini on Substack.
Find Citrini on Twitter.
Support the show by subscribing to the Premium Substack: https://riskofruinpod.substack.com/p/new-episode-big-chunky-changes?r=ckepv
Email the show: risk of ruin pod at gmail
Follow the show on Twitter: @halfkelly
Hey folks, before we start the show, I just wanted to mention that if you want to support the podcast, then the Risk of Ruin premium sub stack is the best way to do that. And if you become a premium member, you won't just be supporting the show. You're also going to get access to new episodes a few days before they hit Apple and Spotify. as well as the archive episodes. I've been taking the older shows out of the RSS feed and moving those to Substack. So subscribing is a nice way to let me know you'd like to see more episodes, and you'll also get some premium content benefits as well. Anyway, hope you enjoy this episode. So
SPEAKER_02:Part of the thing that I say to people is that I'm not betting on stories. I'm betting on the ways that humans react to them. Because that's a lot more predictable. It's a lot easier to predict how people will react to a certain story. I might be shooting myself in the foot here because we write research that's sometimes 100 pages long. But within all that research, what I always try to make sure of is that there's a part where you can just distill and sum up the investment thesis in a sentence or a couple sentences. Because if you can't do that, then the story doesn't have the staying power. If you can't take someone who knows nothing about AI and say, you know, hey, this is like, like, go to chat GPT and check it out. And I think that this is going to be a pretty big deal. And, you know, and they say, well, you know, how are we going to trade it? You know, you can't really do this without this company that makes these chips called GPUs. And they're kind of the only GPU. They kind of have the only GPUs that you can use. So they'll probably sell a lot more of them.
SPEAKER_01:You're listening to Risk of Ruin. I'm John Reeder. This is Episode 42. Big, chunky changes. I have to tell you a story about a little stock called SMCI. The name of the company is Super Micro Computer. Okay, I bought it originally on May 10th of 2023, and I paid$134 a share. Then, almost from the time I bought it, it just seemed to go up, like, every day. Actually, correction. It went up most days, until August. And then it got whacked. But it had already more than doubled. So I sold about half of my position for$317 a share. If you've ever been in a casino and seen some galaxy brain... take chips off the table, you know, coincidentally, the amount they're ahead and stick them in their pocket. It was about like that. Then the stock was kind of flat, kind of choppy for a few months. But then in January of 2024, it took off again. So on February 6th, I sold some more for$650 a share. It had doubled again. But this sale was a huge mistake because not long after that, it was up another 50%. All right, and then on February 16th, I finally got off this rocket ship. I sold what I had left for$883 a share. By this time, SMCI had become a pretty big story in the market. I mean, that's sort of required. For a stock to go up six times in less than a year, it needs more buyers than sellers. So it went from no one even knew what SMCI was to its price moves were being reported like it was Microsoft or something. I also have to say that when I owned it, I would just wake up and check the price and tell my wife how smart I was for owning SMCI, and she would listen intently and respond with words of encouragement like, uh-huh, and neat. Okay, but then when I sold, I would pray to God in heaven for no more headlines like SMCI up 10% in pre-market trading. I mean, it kind of went further than that. I was rooting against it in the way that you might have a quiet, burning animosity toward the Facebook page of someone you went to high school with. Keep in mind, SMCI had only done good things for me, and the people who then owned it? Well, I had to sell to someone, right? I mean, that's not very sporting to hope you sold someone a bag of dog shit. But my fragile ego could not handle the thought of someone else making money on this thing. Alright, then in August of 2024, SMCI dropped 20% in a single day. They ended up in a short report from Hindenburg Research. I actually felt bad for rooting against the stock at that point since you don't want it to be the case that you made money from fraud. So, yeah, my relationship with SMCI is complicated. It's just, there's a lot of feelings there. And just to be clear, this SMCI story is not actually about how smart I was to be on this thing at exactly the right time. I mean, if I'm being honest, I pretty much just got lucky. The other thing is that it wasn't even my idea in the first place. I kind of got onto SMCI because it showed up in a value and momentum screen that I run, but that's not usually enough for me to pull the trigger. There are hundreds of names that show up on this screen, and I don't do anything with most of them. But once I saw it on the screen, I also started to notice that there was an account on Twitter talking about SMCI, but not just talking about it. Okay, I went back and looked. During the time I owned it, this single account posted 100 times about SMCI. So that's a tweet pretty much every other trading day about a single stock. I mean, that's some pretty unhinged stuff. Which, to be clear, I mean that as a compliment. You know, we are fully on the side of borderline unhealthy special interests on this show. And if it sounds stupid to follow a random from Twitter into a stock, I would offer a rebuttal. Which is, now you have a thesis. And you can compare the data that comes in against the thesis. And not only can you compare the results of the company... versus the thesis, you can compare the reaction of the account you're following. If you pay attention, I promise you will be able to see when someone is confident because the results are what they expected. And you can also see really clearly when the opposite of that happens, i.e. when optimism turns to cope. So yeah, I wish that Twitter was actually full of accounts willing to post every day about the stocks they know a lot about. Anyway, The accountant question, the one that put me on to SMCI, was Citrini Research.
SPEAKER_02:It was like August 22. I think, you know, I forget how many splits there's been, but SMCI was at like$40. And I, you know, it was on my screen because it was just a super cheap stock and it was interesting and it was something where there was so much panic about semiconductors and, you know, they made the boxes that semiconductors go into. And, uh, so it was something that I was watching and, and had a, you know, small token position. And, and then when, um, when Chachapati came out, uh, and I started seeing, uh, this kind of like enhanced positivity towards semiconductors, uh, in like December, I, you know, I wrote about it again on Twitter and I said, if you're playing AI, you want to buy the picks and shovels and, uh, the the picks and the the picks and shovels are the gpus but the gpus need to go into something right you can't just put it between you on the ground and say like hey do ai you know you have to connect it into a computer uh into a server so that was something where you looked at the price and you said okay even and again the there were all there were tons of accusations of like They had put Chinese spyware in their servers. They were in accounting fraud. This was not unknown, unknown. The reason why the stock was so cheap was because of that. I looked at it and I said, if in my bull case for AI infrastructure a year out, this company will do more in earnings than it has in market cap. I'm going to buy that because the thing is, these server racks exist, right? You were able to like go out and it's like, oh, you know, this is a super microcomputer server rack. It's not like something where it's like a D-SPAC and it's like, do they even have a product, right? It's like, they do have a product. And in this kind of environment where like everyone is trying to buy every single GPU that they possibly can, it's something where it's like, okay, well, even if they are doing fraud, they're probably still going to sell more of these, you know? And like, you know, so... That's how I originally got into SMCI. And it was like a persuasive kind of thing. Because if you'll remember, by March, April of 2023, the biggest pushback... I remember the first podcast I ever went on, Forward Guidance. I went on to speak about, I think there's going to be a new market because of AI. And all the pushback that I got was mostly centered around how expensive NVIDIA was. And I remember saying, well... if you think NVIDIA is expensive, which, you know, maybe it is, maybe I'm wrong and NVIDIA is expensive. There's a ton of other semiconductor companies. And I laid out like, like all these semiconductor companies that were trading at like, you know, a fifth percentile valuation. Right. So the going into essence, you know, with the kind of eyes wide open approach of like there, there's something that has been wrong with this company. They're also a huge beneficiary of what's going on here. So I'm going to own it and I'm going to like not, forget that that's kind of a risk factor.
SPEAKER_01:Let me try to explain why I'm really interested to make this episode. Financial markets are primarily about the future because the companies that trade on the exchange are trying to make more money tomorrow than they made yesterday. And investors that trade in stocks are betting as to which companies will ultimately succeed at that game. So having a forward-looking perspective is really important. But it's also very hard. Human beings are kind of engineered for myopia. We're wired to believe that whatever the dominant narrative is right now will just keep rolling uninterrupted in perpetuity. But our guest, Citrini, has been on a tear because he's been able to overcome this problem. Actually, he's become a minor celebrity in the part of Twitter that talks socks all day, a.k.a. Fintwit, and his sub stack has taken off. He's gone from a few thousand subscribers to basically the top of the chart in terms of paid sub stacks. And it's all because of a string of very prescient market calls covering things like AI and the GLP-1 drugs like Ozempic and even a strategy for how to trade the Trump election. So if you want to try to think about getting better at seeing these kinds of trends ahead of time, which I do. So I kind of...
SPEAKER_02:I arrived at thematic investing through failure. The first kind of interaction that I had with trading was basically when I When I left med school and dropped out and ended up starting a business and then I ended up making some money, which was like a new thing for me. And then people were kind of like, you have to get a financial advisor. And that wasn't something that I was going to do. I don't have the kind of implicit trust that was required of that or like the faith. And also, I had no frame of reference by which to evaluate whether someone was a good financial advisor or not. So I just started reading about it and I started basically day trading just like off charts technical analysis really and very like Dunning-Kruger right like I thought that I knew everything because it was a relatively easy market to be trading in and I was making money and but that ended pretty quickly so like that 2018 taper tantrum was I had two trading accounts just get totally zeroed out. The second one actually never made a single cent of P&L. It was just negative.
SPEAKER_01:It's pretty much conventional wisdom that overconfidence is a bug in human behavior. You know, Citrini admitted that early on he was operating on pure Dunning-Kruger. But I always think that overconfidence mixed with tenacity is a completely different thing. It can produce value. So, I actually think overconfidence is more feature than bug. I mean, in some ways, the overconfidence slash tenacity dynamic is inherent to achievement. You need to set out to do something that's not easy. And the only reason you get involved is because you completely misunderstand what you're up against.
SPEAKER_02:You know, I think that you can pretty easily... determine things that you're interested in or things that you're passionate about if you meet with like real adversity and just real difficulty and your reaction is, oh, I want to do this more, right? So I was kind of hooked, you know, and it was the taper tantrum basically, like it was leading into that in 2018. I'd say like the turning point, you know, I built up an account I had like a string of really, again, just like trading momentum and charts. And I'd basically taken like a hundred grand. I turned it into like a little bit more than$3 million in 16 months. And then I took like a$700,000 drawdown in 19 days.
SPEAKER_00:And
SPEAKER_02:that kind of forced me, that was like a crazy amount of money for me and just forced me to kind of like reevaluate what I was doing and looking at my process. And I really, the driving factor behind that happening was macro. So I figured, okay, I should probably learn about macro. And luckily, I had a neighbor who was a macro trader and he kind of, you know, he He was like old school guy, uh, kind of took me, showed me the ropes and like taught me to kind of like, he said, you know, go back and like look at the periods in which you kind of made the most money and then let's, you know, and then you can kind of build a strategy out that. And, uh, when I did that, I realized that I was kind of like, um, I made more money when things, uh, not necessarily went bad, but kind of like, like very unexpected things happened. It was like kind of, uh, crisis alpha. I made more money in volatile markets, but not necessarily in volatile markets, but markets where there was a big change being perceived. And then when markets were boring, I didn't really do so hot. So I started asking myself about how I could... ensure that i'm benefiting from these large trends that i was really good at calling the uh like inflection point of but um not so good at like sitting around with like how could i capture the full value because i looked at these trades that i had and they were great trades if i if i kind of uh just stuck with them and and held them and and realized the end of my thesis and and uh didn't just like uh and they weren't trades they were investments so and then how also could i do that in a way that would mean that i was actually capturing what i was trying to invest in because what i was trying to invest in was like these big kind of chunky changes right whether it's macro or micro or you know uh whatever the case may be
SPEAKER_01:so trini says Big, chunky changes, which defines the magnitude of the thing we're looking for. So AI is a nothing burger pretty much for years. And then you get a catalyst where suddenly people realize that large language models can be useful. And now today you can walk through a Starbucks and have a decent chance of hearing strangers talking about chat GPT. I literally witnessed that just today. But these seismic shifts are also at odds with the idea of market efficiency. If markets are truly efficient, then they should be mostly pricing in the available information. So how did these paradigm shifts even happen?
SPEAKER_02:There's an interesting piece that I wrote before this was like an investment research. Back when this was just a newsletter, there's a piece that I wrote about this big mystery of why did bond markets not predict World War I? So besides the thematic equity stuff, I'm also a huge nerd about financial history. So that seems like a premise where you're like, well, markets don't predict stuff all the time. Why would you expect the bond market to? But the thing is, if you look back before World War I, consoles, basically, as they were called back, were extremely good at predicting geopolitical volatility. And there's a lot of explanations that have floated around out there in like the academic space about why markets were so sanguine about World War I didn't really predict them because they didn't really even, you know, Archduke Franz Ferdinand is assassinated and it takes like a full month before there's really any like volatility about this. Yeah. Some people blame the rise of retail investors or buy and hold investors. And there's all these, in my opinion, wrong explanations. And the explanation that I think is correct is if you look at when this kind of change in the ability of the markets to be good at forecasting geopolitics and forecasting the volatility associated with that stuff, it starts degrading when the telegraph is invented. And you think of that and it's kind of interesting, right? Because if you, let's say you're like a trader and it's like, you know, late 1800s, If you're getting news about what's going on in the European continent, the fact that you have heard that news, you have a filtering process for you. The fact that it made the voyage, the information, it's probably worth paying attention to. When the telegraph gets invented, it's just like you have a broker who's biased and has all this information. It's just coming at you, and you don't really have this filter. And I think that that has really... That's gone... That's gotten magnified a hundredfold, right? With the amount of kind of constant news, you really have to have a good filter for what makes an important piece of information and what doesn't. And I think that if you don't, you kind of end up like the frog in the slowly boiling water where you'll miss these big regime shifts because you're not seeing the forest for the trees. So I have a process of news review that... is basically information, the important information finds me and then I decide to kind of chase it down, if that makes sense. You know, like balancing kind of broad scanning of information with like deep dives and finding signals that are basically news is important to me when it's kind of orthogonal to consensus. And it's a lot less about consuming everything and a lot more about identifying what We have
SPEAKER_01:had episodes that touched on cognitive biases and the way those biases create opportunities in the market. Evan Tindall said that when he's looking for undervalued stocks, it's pretty much a requirement that he understands which cognitive bias is leading to the mispricing. Well, I think there are three biases that kind of add up. to what we're going to talk about next. Call it a cognitive bias cocktail. Basically, if you take anchoring bias and add a dusting of recency bias, and then there's this other thing called what you see is all there is, you know, this is all Daniel Kahneman type stuff. Okay, if you add all of those together, you get this really interesting feature of markets, which is just Complacency. For as much as our idea of efficient markets requires us to believe that the information is being accounted for, in reality, there are sometimes, not always, there are sometimes pockets of complacency. Citrini says that when Russia invaded Ukraine in 2022, the broader market just kind of didn't register the important evidence.
SPEAKER_02:The invasion of Ukraine, right? That was like, that was a really good... because it was like, if you exposed yourselves to all these like kind of consensus takes and people finding like a million different reasons for why this wasn't going to happen, you know, if you just took a step back and kind of looked at the trend of like, and there was a point where I think it was in January and there was like, you know, a few hundred thousand, maybe half a million Russian troops that were amassed on the Ukraine border. Like that was enough to say, I'm pretty sure they're going to invade Ukraine. And I don't need to go and seek out all these explanations for why it's a bluff or this, that, or the other thing. Because the process for me is once I think that that's going to happen, okay, let me go and really analyze if this is going to affect me.
SPEAKER_01:I have heard Citrini repeat something that George Soros says, which is, I'm not predicting, I'm observing. So that's like, pay attention to what's happening. Citrini took his observation about Ukraine and then focused his attention on two commodities, wheat and natural gas.
SPEAKER_02:And that's where, like, the really deep research aspect comes in. I remember getting, like, super, super intricately familiar with how wheat worked. And, like, to the point where... the difference between Chicago and Kansas City wheat. And then there was this ETF, the WEAT ETF, that mostly held Chicago soft red wheat. And it had caused this huge premium. But the thing was, most of the wheat that was taken off the market from the Ukraine invasion was the Kansas wheat. So that was like a long short trade that I had put on to kind of take advantage of that correcting itself, which it did. And also just saying, okay, well, I'm going to buy really wingy natural gas call options. And if it so happens that I'm totally wrong about this, there's like no, okay, I guess I lose like 40 basis points in my portfolio and who cares. And then if I'm not, I'm pretty much protected and I don't have to think about it again. So it's also a question There's always going to be big things and really potentially very impactful headlines. And my process for learning about whether I care about it is like, is this something that I'd be able to hedge against? Let me find out if I can accurately take this risk out.
SPEAKER_01:In order for humans to understand an idea, it's pretty helpful for it to have a narrative. But that creates a little bit of a problem because I could create a narrative for almost anything. I asked Satrani about this, and I asked him specifically to delineate how you would pick out which theme to invest in given two powerful narratives. I used the examples of space and AI.
SPEAKER_02:So let's look at like, you know, you mentioned space, for example, space travel. That's a theme that, you know, it's very easy for us to look at the future in a very sci-fi way and say, you know, eventually we'll have a moon colony and this, that, and the other thing and, you know, whatever, right? In that environment, you know, in 2019, it was something where these companies, you know, not many of them had revenue. You weren't really getting an opportunity there.
SPEAKER_01:Citrini is basically saying, space, sure, why not? But a story about the future just isn't enough. There has to be more.
SPEAKER_02:You could have been an investor in AI very far back, right? Like AI, for argument's sake, has kind of existed since computers were a thing. And definitely the focus on AI kind of accelerated with big data in the 2010s and says. But seeing Chad GPT and saying, you know, this is an interesting experience of, you know, normally when I want to talk to a machine, you'd have to use code, but I can use English and kind of speak with a machine. Now that seems like a big paradigm shift.
SPEAKER_01:So there was a trigger. And then if you read the Citrini research, you will come across this repeating idea, which is secular trend, cyclical price. I think of it as how great would it be if the next big thing also involves companies that are currently in the toilet? Citrini also mentioned the Gartner hype cycle, which you're probably familiar with, but if you're not, it's, you know, the roller coaster looking line that starts with hype expectations and then a trough of disillusionment and then finally a breakthrough with usable, accessible tech. So maybe there is something inherent about being able to pick these companies out of the dumpster precisely because they've been on this Gartner hype cycle roller coaster.
SPEAKER_02:the way that I envision themes is they exist on a spectrum. So if you imagine, uh, four quadrants, uh, and the X axis is macro to micro and the Y axis is, um, like either disruption or innovation. And then, uh, the, the opposite side is basically a continuation of something. So that's, that's where they kind of exist. Most of the technological themes, um, are going to be, you know, on the top right of that chart, they're going to be like, like about innovation or disruption to an existing way of doing something. And they're going to be very, uh, micro focused. So, Whenever I am analyzing whether or not a narrative or theme actually is something that you want to invest in, I always start from macro. Is the macro environment kind of conducive to this? But not necessarily just macroeconomics. There has to be a real-world kind of trigger. My favorite kind of theme is one that gives you the opportunity to buy a secular story at a cyclical price.
SPEAKER_01:NVIDIA fit this secular trend with a cyclical price because they had flown a little too close to the sun during the crypto boom.
SPEAKER_02:If you were trying to suss things out and you called a semiconductor analyst back then and you say, hey, you know, I'm thinking about NVIDIA. The only thing they were going to be talking about is how cyclically screwed they were, right? Well, there's a glut of GPUs from crypto mining. right? And, and all this stuff. And then if you, if you mentioned, okay, well, you know, did you see chat GPT? Um, and they would say, well, yeah, you know, maybe in like three years and, and, you know, we don't really know about AI or, or, or this, that, or it, we don't know if it's, uh, going to necessarily use this. That is something where it's super, super attractive, right? That's, that's an amazing, because the thing is like narratives that take hold are, When they can re-rate things that are cyclically priced or at cyclical troughs to massive, that's what we're talking about where in the beginning of any good theme, what's being discounted is reality and what's happening right now. What's being overvalued by the market is basically thoughts about the future. Then once you get to the top of that cycle, that flips, right? Where investors are very heavily overvaluing, you know, thoughts about the future and undervaluing what the reality is. And so with AI, the reason why it was so attractive was just because there was this tale, you know, I didn't necessarily think that you would use GPUs to run a large language model. And I was very strongly convinced that AI would become a transformative technology. I didn't know if it was going to happen in a month or a year or whatever, but it's very easy to get comfortable with allocating something like that. When you look at NVIDIA and you'd say that none of this tail is priced in at all. There is no, you know, and that was true for the majority of like big tech and semiconductors. And that's why, you know, AI has had, in my opinion, such staying power. Part of it was just, you were coming from a base of such negativity that It was something where investors had this kind of shared post-traumatic stress disorder from the kind of 22 blow up. And that was something where it gave all these names in kind of like the AI infrastructure supply chain, because that was the first place where the real money was being spent. So many opportunities to absolutely destroy market expectations. And that's something that it's right in your face and it becomes very hard to disagree with because you're seeing it happen. So the difference really is you have to start from, is the macro environment conducive to this theme?
SPEAKER_01:If you're right about the direction of a trend, it can still be maddening to figure out what the magnitude of the thing can be, especially when you have a stock sitting in your account that's up a bunch. It's like, You know why you got into it and the thesis is playing out, but you also know that if you're really right, then the thing that will happen is it will go sailing past your fair value. That's the ultimate result of finding something that's undervalued. Eventually, again, if everything goes well, it should become overvalued. And keep in mind, you also have to do this calculation, which is a mix of math and and psychology while other stuff is going on, while you have cross-currents in the market to deal with? I
SPEAKER_02:owned some AI stuff. I thought I owned Supermicro, for example, in 2022. I had Bond video, but I didn't really get super, super bullish on it until after Silicon Valley Bank collapsed. Because that was something where it kind of took away an element of macro risk that allowed me to say, okay, it's worth going full into because this theme exists, because the macro environment is getting more supportive. And I remember looking back at the biggest thing that I had to get over myself about was another kind of theme that I'd been involved in or tracking for a while was this hire for longer rates phenomenon, right? Yeah. I basically had to beat the concept out of my mind that it would be impossible for tech to outperform while rates were high. And that was a very, very strong narrative, right? And going back to the dot-com bubble and seeing that rates were held pretty much at 5% plus or minus 50 bps for the entirety of the bubble, that kind of got me a little more comfortable with it. And then went for it and we climbed a massive wall of worry. So yeah, the answer to your question is just you have to have a strong filter on macro. It has to be the right macro opportunity for a theme to kind of take hold.
SPEAKER_01:One of the trademarks of the Citrini research is they don't just leave it at AI will happen, here's NVIDIA. They try to game plan it out and think deeply about how it will happen. As of May, their AI basket was 20% of the Citrindex. And the Citrindex is just what would happen if you took all of the ideas and then multiplied those ideas by their confidence in each one. Also, the AI basket included 55 stocks, which I have to point out, not all of the names obviously benefit from AI. It requires some double takes before you can see the AI angle with some of this stuff. So you can't get from AI is a really big deal to these 55 seemingly randomly assembled stocks. without a lot of work.
SPEAKER_02:So with AI, for example, when we started out, we laid out something that I think that we were kind of like relatively early to was like viewing AI as a three phase kind of progression where the first phase was like picks and shovels for AI infrastructure because we don't know what AI is going to look like. And the second phase was like kind of the implementation of AI for actual use cases. And then the third phase is kind of like this revolution or democratization or whatever buzzword you want to use. We're not there yet. And it's going to be a while until we get there.
SPEAKER_01:When I'm figuring out if I want to pay attention to someone, one of the questions I ask is, have I ever seen this person express a bull and bear case sincerely? Not just, you know, devil's advocate type stuff, but do they really change their mind? Are they sometimes long and other times short the same stock? And on one hand, I think you could look at someone doing that and think, well, they must not have had very much conviction before. on the first trade, but actually I see it as a sign that someone is trying to stay calibrated to the facts. Well, Citrini was long NVIDIA, then exited, then they were short, and then they recently wrote up this massive piece about robots, and it included NVIDIA again. I
SPEAKER_02:do like active trading and stuff. NVIDIA was the name that was in my portfolio for the longest period of time without being sold. So, you know, I bought it in early 23. The first time I ever sold a single share of NVIDIA was in like November, December. And that was just because the way that I saw the... I started seeing risks, right? In the sense where for those like two years almost that I was... I would look at it and I would say, you know, here are risks and they're like... they won't happen. I can comfortably say that these risks that they exist, they won't happen. One of the big risks, obviously, because NVIDIA is still a semiconductor company, which is still cyclical, was a recession. And I could underwrite that. I was very convinced that we weren't going to have one. When it got to the point where there started to be risks in the sense of Chinese custom design ASICs or software competition. I just said it wasn't something where I said in the beginning when I first exited it, I think it was like$140 a share. And I said, I just think that there are better opportunities to play what I think is going to happen next in AI. And then that was around the time that I started like tracking DeepSeek. And that was a big, that kind of played into it where I said, if China can make this model and can like compete with OpenAI, in many ways that could be good for NVIDIA, but it can also signal like they don't have the ability to get all these chips in the same volume that the hyperscalers do. So they have this kind of constraint. And if they operate under that constraint and they throw as much talent as they have at it and they find a way to make this less computationally intensive, that's a risk. And so that's, you know, we went short in January and then it got like a lot cheaper and the things kind of changed and hyperscalers were very adamant that like they were still spending. And so I, you know, bought it back.
SPEAKER_01:I want to go back. to something we were talking about earlier, which is information processing. First, you have to notice that something is worth paying attention to, which isn't super easy. But then, where do you even go from there?
SPEAKER_02:Big part of that process is if I stay interested. When I was studying medicine, it was like, I would read the textbook before the class, before the first day of the semester. Just because I was super interested in it. And so... If something can catch my attention and then I'm researching it and continually researching and it's something where every kind of incremental minute adds another branch or road that I want to go down, that gives me confidence that the theme has staying power. The thing is, we are not deriving edge from... There are a lot of really talented investors that I personally know that could be right on the money about where earning for a name will come in next quarter, right? And they'll know exactly and some of them can even like very accurately estimate what the market reaction to that will be. And, you know, when you're doing something where the onus is just kind of like how significant is the opportunity from this theme the like the analytical kind of burden gets a little bit lower you still have to do the research doing it in a broader sense so and then once you go through that whole thing the cool thing is when you go through like with our robotics thing we created like a universe of just like we broke it down I always break themes down into like digestible bits so like with the election for example when we wrote the like Trump basket in March of 24 it was like here's the you know, here's like Trump idiosyncrasies, which is like Fannie Mae, Freddie Mac. Here's like European defense. Here's on-shoring, you know. So like breaking it down like we did with robotics into like sensing, perception, actuation. And once you go through all that and you have like a surface, ankle-depth understanding of all these companies, it becomes a lot easier to see which ones are really, you know, and that's the ones that we kind of like focus on a little bit more because it just becomes easier It's something where it becomes a lot more evident.
SPEAKER_01:I think if we spend all of the episode talking about a theme that has already materialized, it will make all of this seem easier than it is. Maybe you're sitting there thinking, of course, AI will be big. It already is. So move on. Okay, fair enough. Let's talk about something that still may be playing out or maybe not. It's to be determined. Citrini put out an 80-page report about humanoid robots. The thesis is basically that robots might be close to their chat GPT moment. You know, in China, they had a robot marathon, and the winning robot finished in under three hours. And there's also a company that's selling humanoid robots for less than$20,000. But also... If robots are close, humanoids make the most sense because the world is already built for the human form factor. That's my best attempt at distilling 80 pages into a few sentences. But the full report is stuff like which companies make which robot gears and which companies make the sensors and which companies make the chips that make it possible for robots to think. And then the report had Tesla in there, which I get because Tesla has a robot. But I asked Citrini how he thinks about Tesla's valuation versus, for instance, Nvidia, which remember, was beaten up because of the crypto crash.
SPEAKER_02:When I first do research on a theme, and this is pretty difficult to do, but I kind of like, I try to like abandon any bias that I might have, right? So I would agree with you, right? Tesla is kind of fully valued. It's priced in a bunch of stuff. But when I'm doing the research, my goal really is to uncover everything, right? So it would be, you make a basket and yes, Tesla is going to be in there because Tesla does have a robotics opportunity. It's when When it gets into the realm of like active management that you start saying, okay, you know, what risks do I want to take here? So with that, we started with something where, you know, this is a huge theme, just like how robotics is. There are hundreds of companies that could benefit or touch it or, you know, because I don't, I want to come in with like a beginner's mindset, right? I really want to come in where I'm not ruling stuff out prematurely. I just want to make sure that it's on my screen because seeing how these things progress, there's so many areas where you can be totally surprised right like something that was surprising for me I didn't realize that the kind of like power stuff with AI I didn't think that that would happen as quickly as it did That was something that I thought would take, but, but, you know, I had them on my, on my screen and seeing them kind of made it so that this is where active management comes into play. We are, we've already done the first kind of rebalance of our robotics basket as we're narrowing it down from these hundreds of names. And Tesla was one that we took out, right. And we replaced it with Amazon. It's just something where having, having it there and, and, and not like The valuation and basing off earnings, that's something that when you take this basket approach and everything's kind of like lower weighted, right? Like there's so much that each, you know, Tesla goes down 50%. The performance of the basket is not that significantly impacted. But at the same time, you know, let's say there's like the edge case where like Tesla tomorrow is like, oh, actually, this robot's already ready and it costs 15 grand. And, you know, who wants to buy it? I don't think that's going to happen. But having it on there and forcing myself to keep up with it and kind of do the work, that's where anybody can create a universe of stocks that have to do with something. It's in finding the kind of signs and the kind of consistency of this opportunity becoming more and more a key driver of a stock. And that happens over time.
SPEAKER_01:The dozen or so thematic baskets that Citrini has published seem like a lot to keep up with, considering that each basket might also have 50 stocks, but they also have baskets they haven't even published, because sometimes they make a basket just to monitor an idea, and then they might get weird results, like that group of stocks booming, but not because of their thesis.
SPEAKER_02:So we have thematic watchlists. In order to do that, I have an inkling about a trade or a theme. And I'll go and make the basket, right? And maybe it won't be the best basket ever because it doesn't have like deep, deep research about it. But it's kind of like I'll create like a tracking kind of just like watch list basket. A big part of it is going and seeing how even when they perform well, you have to go and like... you have to make sure that it's performing because of your theme if you want to add conviction on that, right? So, for example, back, like, two years ago, one of our, like, thematic watch lists was about superconductors. And we made, like, a whole basket. There were, like, 50 names in there. And, like, here's what would benefit if we had room temperature superconductors. And, you know, that basket might actually be one of our best performing baskets ever. But the thing is, that's not because of superconductors, right? It's because of, like, this This extreme bullishness over like the AI power, the power kind of supply and because of kind of like what's going on in industrials and especially like longer cycle industrials. So it's something where I always have a kind of watch list of themes that could be interesting that maybe I want to do more work on. And a big part of that is just seeing how the market kind of takes those because when you have a real, this is, we're still kind of, you know, there are also macro themes and themes One of the other trends
SPEAKER_01:that Citrini is known for is the GLP-1 theme. He was actually on an episode of the Bloomberg Oddlots podcast where he talked about Ozempic and the other weight loss drugs. Well, he also made a long short basket based on how Ozempic would affect other companies?
SPEAKER_02:You know, probably a year, like 13 months later after we had published the GLP-1 long short basket, which had a short side that was, you know, CPAP machines and some med tech stuff that had to do with like, you know, just lower rates of diabetes. And, you know, there were some fast food companies in there and stuff. Narratives are so strong that they can create opportunities on the opposite side. So it was something where We anticipated, okay, if penetration of these drugs is X, then we'll see a Y decrease in the necessary CPAP machines for sleep apnea because there'll be less obese people. By the time we got a year later, that worked great. And then come back with, again, that beginner's mindset of constantly analyzing where you could be wrong. And we realized there's this huge opportunity in some of these specific med tech names Because, you know, 150% of this effect has already been pressed in. So that, you know, that's like, that's always like a really fun. That's something that I enjoy is like, like when you're right about a trade, and then you're able to like, utilize that to inform the next one.
SPEAKER_01:I mentioned the Citrindex, which contains the portfolio weights for the various themes and the individual stocks. I think that's a pretty important part of this because if you put enough stocks in enough baskets, it would be impossible not to have some winners. But Citrini is saying, here is the scorecard you can use to judge our ideas.
SPEAKER_02:I was publishing these baskets and, uh, You know, I would I would like the question people would always ask, you know, like, are you putting this on or like, how are you viewing it? It's like the easiest way to to convey to someone. People ask you, like, what's your level of conviction in this? And it's like there's one way to answer that question. Right. One way is like, here's how much of my portfolio makes up. It's just like we could write an entire piece that's just like, hey, you know, here's how things have shifted or whatever. Or it could just be I'm taking this off. And like that accomplishes the same thing. So honestly, it was like something of convenience. That's why I started to publish the portfolio. And then like, and that kind of... It also forced transparency in a way that I found pretty constructive to my own investment process. There's a big difference between selling something and analyzing what selling something in front of thousands of people is going to result in. What kind of feedback you're going to get or whatever. It's a lot more difficult to just sell something and be like, well, yeah, not so much. You have to have good reasons for selling. So Um, yeah, that's like the Citrendex part is like my favorite part of doing this. Uh, it just, it's, uh, and so like I made like a, like a, now we have something it's like Citrendex.com and, and you can, and it's like real time.
SPEAKER_01:I think one way to look at the basket approach is that it's an attempt to strike a middle ground between on one extreme diversification that just includes everything and thus contains no opinion versus And at the other extreme, a super concentrated portfolio where you have to hit it dead solid perfect.
SPEAKER_02:It's diversified on a security level, but it's concentrated on a thematic level, right? So I don't view it necessarily as like, oh my God, you know, there's 250 stocks in here. I view it as like 15% of this portfolio is in the second phase of AI. You know, 15% of this portfolio is playing fiscal spending in Europe. 5% of this portfolio is playing what's going on with the airlines, you know? And that's what like, like if you see like the chart of how the Citrindex has done, it's like, you know, like this year it's like up modestly, you know, it's like, like I think like, I don't know, like 9% or something, which is good. Like it's definitely beating the S&P, but like, In 2024, it outperformed by 60% or something. That gives you such a big base. When you do catch these themes, those are the times where you really want to lean into them and be able to extract as much as you possibly can. If the thematic environment gets worse from here... for the long term, I can think, okay, well, we have all this P&L that was like stacked up on this opportunity. I'm okay with like taking some risk off and waiting for the next really good theme instead of just trying to jump in front of every single, you know, like narrative that might be a narrative.
SPEAKER_01:We've spent most of the episode talking about technology, but obviously the themes can extend beyond that. Citrini is looking for any major shift that might not be priced in by the market. In 2024, they put out a Trump election basket of stocks.
SPEAKER_02:I think probably if you, if, uh, I was going to say like, what has our best piece been so far? Cause it would have to like go back and look at performance. The, the election piece in March of 24 was like really good. Like I was really proud of that for two reasons. The, the, the first is like the basket did very well. Uh, and I messed up and like, I, like I, I took that basket off way. I basically took it off when Trump got shot because I figured like, okay, this is like the maximum probability of him getting elected. Um, and you know, but, uh, So we published this basket in March of 2024. And our thesis was Trump's odds being the candidate are higher than the market pricing. And we created a basket that was like 75 securities long, 75 securities short. And it tracked the odds so perfectly going forward, like to the point where that long short basket, which again is totally market neutral, like bottomed the day before the Polly market outbound.
SPEAKER_01:Citrini told me that he doesn't consider himself to be a particularly great fundamental analyst, but he is really dialed in to this basic mechanism of how markets and ideas interact. And that's the kind of thing you're really only going to get from paying attention to the intersection of humans and markets.
SPEAKER_02:There's, there's a big intuition aspect that I couldn't like quantify if I tried, there are times when it's wrong. And and that's like, that's something where, at least when I'm wrong, I know, like, I can pretty quickly ascertain why I'm wrong. And that obviously comes down to research, right? The edge of like, kind of anchoring to the second derivative and like, looking at the at the world and everyone's kind of like playing the first bounce of the tennis ball and like seeing kind of like, how it bounces from there. And then constantly looking for areas where you're wrong. And I think that there is a personality aspect where it's just like, I'm not afraid of being stupid. I think that investors have lost more money from like being afraid to like go out on a limb with something than they've probably ever made. So yeah, it's just like there is a personality aspect of just like I am open to being wrong and I'm like excited to find out why that might be the case.
SPEAKER_00:It's
SPEAKER_01:worth pointing out just the nature of being quote-unquote wrong. Okay, the easiest way to seem wrong is to just go outside the consensus. But this, you know, probably evolutionary impulse to stay with the herd, to avoid ending up on your own, is also a surefire way to have returns that aren't worth writing home about. That's inherent, right? I think we can actually break down Citrini's willingness to end up outside consensus into two stages stage one what is happening right now that everyone else is missing stage two the deep research because if you fill your head with information that others haven't bothered to collect then you will naturally end up in a different place
SPEAKER_02:a lot of my time is spent focusing on where I might be wrong. So, uh, I, I do like, uh, like if anything, I can lean more towards like, like I might abandon something, uh, because I, I think that I'm wrong more, more. Well, I, there's, uh, the ego side of like, uh, I'm when, when I feel that I'm right, it's like backed up by like a lot of like, um, like high conviction that, that I'm right when I'm, when I think I might be wrong, I get like, there was one that, um, you know, uh, we published like on a single stock, uh, this name, like American superconductor. And it was like the first single stock thesis that we'd published in a very long time. Basically, I think we've only done two or three. The first one was on this company, Celestica, and that worked really well. And then we published on this American Superconductor company. And the thesis was like good to me. And it was like the difference between when we had published Celestica and we had like 2000 readers versus when we published American Superconductor and we had like 60,000. It was like, it was something where, uh, the like constant barrage of, of like feedback got a little bit in my head.
SPEAKER_01:I don't know if you've ever read a quarterly letter from a fund manager and they have a position that's gone against them. So they originally wrote it up and, you know, had a big smart thesis for why the stock was great value. Then it went in the exact wrong direction. And now they're writing it up again, doubling down, uh, But the thing they're really doing is trying to thread a needle between protecting their current ego while addressing a mistake that's already been made. I see these things and I think I'm really glad I don't have to explain any of my investments to anyone because just the balancing act of trying to convince someone that you're developing ideas with conviction while also being able to get away from mistakes seems basically impossible. And I think that probably some managers would actually perform better if they just never mentioned their thesis to their clients. And then when something goes south, they could just quietly take it off without all of the throat clearing. Well, Citrini has written lots and lots of stuff. But actually, he says he's not worried about ending up with these ego-related sunk costs. He actually wants to find out if he's wrong.
SPEAKER_02:I genuinely enjoyed the feedback that we got, right? It's like, of course, because all these different areas that I might not have otherwise looked at, I don't identify with like the, you publish it and now you have to stay in it, right? It's like what I'm looking for are good trades. I'm not like writing stuff to like educate people on these like themes. I'm writing stuff because I think it's going to be a good trade.
SPEAKER_01:I said I wanted to do this episode because I personally... want to understand what Citrini is doing and how I might be able to incorporate that into what I'm doing. So let me try to see if I understand it all. Citrini is looking for big shifts in the market. He's doing that by paying attention to things that are actually happening. He's observing, not predicting. He's paying special attention to news that is uncorrelated to the dominant market narrative. Then he goes to work to try to find every company that might touch that trend. His favorite kinds of companies? will benefit from a secular trend, but are currently in a cyclical depression. And then he is constructing baskets of stocks so that he can monitor the idea broadly and also have the best chance of sticking with the trend. Lastly, no matter how in-depth he's gone on a theme, he wants to be able to tell the story in a couple of sentences. I think that's roughly it. I said that Citrini has been on a tear, but, you know, that's the cruel thing about the market. He could have two years of great calls, But if he doesn't just keep doing that, no one will care.
SPEAKER_02:The cool thing about this process so far is that, although sometimes it strikes me as not that cool and super stressful, but that you're only as good as your next trade type thing. I really do internalize that. So and I just try like the other cool thing is like as the newsletter has kind of grown into a research firm, it's like there's more resources and I can continually like make it better.
SPEAKER_01:risk of ruin is written and produced by me special thanks to citrini for doing this episode i'll put a link in the show notes so you can find him on twitter and on substack if you want to get in touch with the show you can email us risk of ruin pod at gmail.com if you want to support the show you can find a link to the premium substack in the show notes you can also follow us on twitter at half kelly
SPEAKER_00:so