Brett Caughran

LAUNCHING ANALYST ACADEMY, AUG 10th COHORT. Hello! We are very excited to be launching our next cohort of Analyst Academy. What is Analyst Academy? A six month guided cohort of the 60+ hours of institutional training curriculum. The basic tools needed to survive & thrive on the buy-side: - How to build your investment process - How to conduct a deep dive research process - How to figure out what will move the stock - How to generate investment ideas - How to prepare for a management meeting - How to manage the deluge of information - How to develop & communicate a thesis - How to be a good buy-side analyst - How to leverage AI into your investment process (NEW, including 10+ hours of AI curriculum) - And much more In the Analyst Academy, we attempt to teach the "hedge fund equity research process" and endeavor to try to present what we think are best practices for the buy-side equity research process. We have hosted over 1,500 students from all walks of the buy-side - analysts from large multi-managers, tiger style funds, long only firms, and family offices. Increasingly, Fundamental Edge is building internal training programs in partnership with Top 100 Asset Managers. Roughly 2/3 of students are already on the buy-side, with the remainder working to break into the buy-side. This is a rigorous, intensive curriculum - not an investing for beginners program. If you would like to learn more, I am embedding a ~30-minute

In today's Invest with AI episode, Khe and I sit down with Joe O'Donnell, who ran the short book at Tiger Global for almost a decade before leaving to build Canary, an AI intelligence platform now used by some of the largest hedge funds and mutual funds in the world. Joe's take: because so many investors are misusing AI, there's more alpha available today than there has been in a long time. "The amount of alpha available to the world is perhaps greater than it's been in a very long time, primarily around the misuse of AI." - Joe O'Donnell We get into: •Why there is more alpha available now, not less, and how the misuse of AI creates it •Why AI is only as good as the investors who build and train it, and the role of domain expertise •Canary's "layer cake": proprietary data sets, an intelligence layer, then tools and agents on top •An AI forensic accountant that unpacks accounting manipulations across roughly 10,000 companies •Super Analyst, an autonomous junior analyst that covers close to 4,000 companies and refreshes daily •AI idea generation agents that deliver 20+ page investment reports backed by primary research and expert calls •Why summarizing an earnings call with AI is "lossy," and how it can inflate conviction you have not earned •Where off-the-shelf AI shines (information gathering, a far better vertical search) versus where it fails (judgment) •What fine-tuning a model really takes: thousands of hours of manual analyst work, not a plug-in •Why "AI for financial services" is a red flag, and whether a fund can rebuild Canary itself •How the investment firm of the future organizes around judgment, and why more small funds may sprout up Highlights: 00:00 Intro 03:28 "AI, for this use case, is only as good as the people who build the AI and train the models." 04:57 "My strong suspicion is that Buffett and Druckenmiller would win. It's all around judgment." 06:25 "It's sort of a layer cake where all of these things are necessary to generate results that are actually useful for investors." 17:16 "The output is a fully baked 20 to 25-page investment report backed by primary research, including expert calls." 19:22 "AI sounds super smart, and it is increasingly smart, but it's still lossy." 29:28 "It does this autonomously across close to 4,000 companies and refreshes every day." 36:07 "There's no such thing as AI for financial services." 54:52 "The amount of alpha available to the world is perhaps greater than it's been in a very long time, primarily around the misuse of AI." youtube.com/watch?v=fl3R7a…

LAUNCHING ANALYST ACADEMY, AUG 10th COHORT. Hello! We are very excited to be launching our next cohort of Analyst Academy. What is Analyst Academy? A six month guided cohort of the 60+ hours of institutional training curriculum. The basic tools needed to survive & thrive on the buy-side: - How to build your investment process - How to conduct a deep dive research process - How to figure out what will move the stock - How to generate investment ideas - How to prepare for a management meeting - How to manage the deluge of information - How to develop & communicate a thesis - How to be a good buy-side analyst - How to leverage AI into your investment process (NEW, including 10+ hours of AI curriculum) - And much more In the Analyst Academy, we attempt to teach the "hedge fund equity research process" and endeavor to try to present what we think are best practices for the buy-side equity research process. We have hosted over 1,500 students from all walks of the buy-side - analysts from large multi-managers, tiger style funds, long only firms, and family offices. Increasingly, Fundamental Edge is building internal training programs in partnership with Top 100 Asset Managers. Roughly 2/3 of students are already on the buy-side, with the remainder working to break into the buy-side. This is a rigorous, intensive curriculum - not an investing for beginners program. If you would like to learn more, I am embedding a ~30-minute

@FOMOMO263039 Yes fully agree and that’s 100% the part of the exercise

The most valuable output may not be the batting average. It may be the error map. If every miss is tagged by cause — wrong KPI, stale consensus, guidance parsing, sector-specific context — 100 failures become a training set for judgment. Without that taxonomy, a 55/45 result could be skill, luck, or one regime doing all the work. Measure the misses so the system knows what to learn.

I had some interesting conversations this week that inspired an experiment. The Bridgwater / Thinking Machines paper has been a key topic of conversation in AI Investing this week. The idea of delivering superior outcomes via codifying investor judgment blended with LLM intelligence is incredibly, incredibly interesting. However, digging a bit deeper, this paper was more about expert “filtering” than ex-ante judgment, i.e. “is this document dovish or hawkish”, etc. That’s not a profound discovery, imo. Rather than continue to have philosophical conversations about the value of blending humans with machines, I was thinking through ways to apply a more empirical lens to this hypothesis, so I’ve been designing an experiment. Here it goes: > Take S&P 100 for Q2 2026 earnings season, which begins next week > Run all names through my earnings Skills that encode process & judgment > Create an earnings preview for every print in Q2 ‘26 earnings with an estimate on 1) direction P&L (does stock trade up or down on print), 2) forecast quality (revenue & EPS accuracy), 3) guidance calls (guidance revised up/down/held). I’ve considered something like this in the past but I was 95% sure it would fail (or if it succeeded it would just be variance/luck). Principally, LLMs/agents have continued to be very poor at identifying KPI materiality & market expectations, and excel ingress/egress is still a bottleneck, so it was kind of a pointless exercise in my opinion. Fable, however, has been noticeable more incisive in this dimension. I’m only 85% confident it will fail, now. But I figure i will give it a shot, with the base expectation that it will be 100 data points that I can use to study and improve the system for Q3 prints and beyond. So, I’m going to stomach the token bill, run 100 earnings previews with these forecast dimensions, and send a “week ahead” report to Ai Accelerator students (to keep me honest). Do I expect a 60% print batting average? I mean, no. Do I expect it will create good education content and that we can work through ideas to hill climb this together, to dial in all of our custom agentic earnings previews? 100%. Specifically, the concept of durable accuracy with Skills instructions alone and a few MCP connectors *probably* isn’t going to work, but it will create a starting base with which to refine & improve for coming earnings, including generate ideas of what the necessary MCP stack should look like across various GICS industries. It’s hard to overstate how transformational an agent that can call prints 60/40 or even 55/45 could be, particularly with good slugging, so in my mind its an experiment worth starting. It doesn’t seem impossible to me that we can get there. As always, If you working on something similar this earnings season and want to share ideas, DM me!

Happy 250, America!

Interesting paper from Bridgewater & Thinking Machines on judgment in financial tasks: natively, LLMs are quite poor here, but when trained, results can improve materially. Captures an issue inherent to machine learning & stocks: natively, machines are not good at identifying signal vs. noise & materiality. One of the first questions a good trader will ask is “what is the trigger that will move the stock?” Machines struggle intensely with this task, as the answer is often subtle or boring, and the question stocks ask changes dynamically with price & market regime. LLM’s have shown no real improvement vs. prior ML here: large language models aren’t sourcing materiality, they are sourcing salience, and treat lexical intensity as a proxy for significance, which is a really poor way to capture materiality. This is why out of the box earnings previews run through off the shelf LLMs are pretty bad / not useful at all for decision making. They require a lot of oversight, but in my prototypes, a few critical steps systematically breaking down key drivers that can either be earnings levers, narrative influences, or risk factors, set against a multi-dimensional expectations engine that include sell-side but also market inferred expectations improve results materially. Once you do this, you can tangibly see the vision elucidated in this paper on a company by company basis. But it’s obvious that an out of the box LLM trained on a web-scrape won’t (likely ever) have these capabilities. It is a scary idea to, as an investor, consider that perhaps with this approach I could put my portfolio on “auto-pilot” during earnings and the machine could make the right trades add/trim/enter/exit/flip with say a 60% batting average. Even a low-mid 50% earnings print move batting average that clips the tails of blow ups & short squeezes has the potential to so dramatically change the alpha generation approach at multi-managers. This is a problem I’ve been working on for almost 10 years in concept. It feels like it’s now becoming possible to productionize and scale. I unfortunately won’t share much of the specifics here (there are some aspects to this I consider proprietary), but this is an idea that we are working on co-designing with a few clients. Please reach out if it is interesting. t.co/g7KLEl71fI

@BennyDubsMTG She’s not pretty, but she’s beautiful to me

@FundamentEdge Love it

Summer office

It’s an interesting question. Obviously AI cannot automate in person primary research. I assume as both buy-side and sell-side automate more of the desktop investment process with AI, the natural impulse will be to go deeper into primary research (more travel). Competitive drive to find new alpha signal will demand it. As a research business it seems hard to scale as which conferences/tours do you select? How do you get a critical mass of clients caring enough to justify each trip. I think the Cleveland Research model (deep and consistent verticalized channel checks) gets more valuable in an agentic world.

Many investors (people in general, really) don't like to travel. Is there a market for outsourced primary research (conferences, factory tours, etc.)?

@P_Remarks Perhaps. If true, amazing time to be a long term fundamental stock picker

@FundamentEdge I would argue the market now does the opposite. Rerates or derates via extrapolation of current earnings trend rather than what it will be in 2-5 years

It is easy as a stock picker to blame the external environment. "My portfolio sucks because the market is all about AI and retail flows". I had an experience with this early in my career that stuck with me. The investment team was bemoaning the macro driven market (this was '08/'09), citing high correlations as a challenge to L/S spread generation. The head of the fund responded with a table of realized stock price dispersion, "our job is to, ex-ante, identify the winners and the losers...and I don't know about you, but I see a lot of spread potential in this analysis". The message was clear: winners don't complain, they figure it out. This was a clear pattern I saw in the best PMs I worked with over my career. They didn't complain, they just figured it out. The job of a L/S PM is to find spread between longs & shorts, wherever it is. They adapted to the market environment when necessary. I saw this in real time when I was a PM at a large multi-manager. Sure, great PMs would have drawdowns as market conditions would change, but they were flexible when needed, flowing with the market regime to make money in new ways. Sort of the antithesis of the calcified value investor "I have one way of making money, buying cheap assets, and I'll go to my grave doing so". And even in this AI & retail driven tape, there is plenty of evidence of dispersion. In my healthcare coverage of 152 names, 75 of them have of them have hit threshold returns this year (longs up over 25% and shorts down over 15%), only halfway through the year. A few of these are AI-influenced, but the vast majority are idiosyncratic and business/industry driven moves that could have been identified with the right research (of course, it's always clear in hindsight). There is plenty of spread potential in this tape, you just have to find it. (Caveat: it's insanely hard. But it should be, as the rewards to consistently generating 5%+ long/short spread in equities are immense. And I fully grasp that I am shouting this message from the sidelines, not "in the arena", so evaluate this message as such)

@NotaburnerJack Lol

@FundamentEdge so what’s your number?

I know I have the right Twitter friend group when all my true buddies here (mostly successful former Wall Street people looking for a more authentic life) start roasting this take. But I will point out one glaringly obvious error with his framework: Scott Galloway is going to die. We all are. But he is 61. Per Claude, the average 61 year old American male has about 21 years left to live, and, on average, about 15 of those years are expected to be "active" / largely unimpaired. And for how many men does this jet-setting, Aspen & St Barths lifestyle stay compelling past ~70? For only the truly mentally ill, I would argue. The 4%/25x FIRE rule doesn't apply to him, and he doesn't even know it (worse, he's attempting to poison the rest of us with this mimetic crap of feeling like we don't have enough to feel secure). And the fact that we are going to die is great news. Life is precious precisely because it ends. A few years back, I did a deep dive on near death experiences (NDE). What was most remarkable was the consistency of the aftereffects: reordered priorities, markedly reduced fear of death, increased sense of meaning & connection to source, and durable improvements to mental health. Buddhist "die before you die" meditations and the dissociative reframing activated in psychedelic journeys activate the same fundamental experience. What is learned through these experiences is that you don't need $125m to feel secure...you are secure precisely because your journey is deeply supported by a benevolent universe. Though this study, I developed a reflection question that has been one of my most reliable mirrors, for myself and when people ask advice: how would you live your life if you knew you were going to die in five years? As a professor teaching (mostly) young men and by being present on Twitter over the years, I have had many occasions where high-performing but burnt out men have reached out to me for advice. It sort of feels uncomfortable, as life advice is a really hard, and also really dangerous thing, to give - because what worked for me likely won't work for you. "Buy rental real estate" was great advice in 2012 and probably leads to bankruptcy in 2026. "Quit your Wall Street job making $650k" may be great advice for some people, and awful advice leading to financial distress & divorce for others. Each of our skillsets and situations are so deeply unique that advice is hard. But I do know one thing: we are all going to die. Don't live like you aren't. Life isn't a game of making more and more money to chase this elusive feeling of security (which is almost certainly a moving goal post). True security is an internal feeling, a feeling of "I can handle anything thrown my way because the benevolent universe has my back". When I was in my mimetic, competitive, "I have to get back in the game" mode after being let go by Citadel in 2018, my coach Justin Doyle looked at me and said "Brett, you are so stressed out, I'm more worried about you having a heart attack in front of your Bloomberg screen than finding your next PM job". It was kind of a throwaway comment, but with an amazing wife and three beautiful young sons at home, it was the right comment at the right time, and it hit me hard and led to a sharp course correct. I literally booked a 10 day trip to India the next day leaving in 2 weeks, on which I was flooded with insight on what to do next (i.e. leave NYC and build a new life in Arizona). The hedge fund headhunters who saw fresh meat and an easy placement were apoplectic. "Build a vocation that doesn't feel like work, that you could do for the next 20+ years" was his next advice, and he helped me do precisely that. These days, when people ask for my advice, I offer them the "5 year test". What would you do if you knew you were going to die in 5 years? Figure that out, then do it. For me, it's simple. My sons are 12, 10 and 8. I want to give them the things my father couldn't give me: most critically, deep & joyful presence & unconditional acceptance. I want to spend a lot of time with them, having crazy adventures & building a deep level of love & trust. But it's also complicated. I could never be JUST a "retired, stay at home dad", as my intellectual muscle would atrophy (which has happened at moments in my life, and is a terrifying and motivating thing for me). I love working hard to solve complicated problems with motivated, intelligent people. I deeply love my job, and this moment of time where AI is fundamentally transforming knowledge work is just incredibly exciting and invigorating to me. The process of creating things is deeply spiritual for me and it's such a delightful experience to be working on a problem and get that magical "ping" of insight that unlocks it. Reading "When Breath Becomes Air" which is the beautiful account of a brain surgeon facing death, but in his dying days loving the craft of conducting surgery was deeply resonant to me around the power of vocation & purpose. It's fine to love your work, in fact it's ideal for a life fully lived, which is anathema to the FIRE (financial independence, retire early) crowd that all seem to view their jobs as prison time. And it's also complicated in the sense that I have a deep duty to provide & protect, and a man or woman who holds the critical mantle of head of household has a sacred responsibility to be economically viable in a world where a nice life is expensive. I've had some well-timed luck in this regard, but I always want to be anti-fragile to all scenarios, and AI certainly expands the fan of outcomes in this regard. In sum, it's a complicated balance. But I have learned one other thing. The courage to embrace living by this "5 year test" has a way of unlocking the support of the universe. "Jump and the net will appear" is super woo woo, but it's a real thing. I can't explain it, but it's real. It's the Neville Goddard Law of Assumption as depicted in the Indiana Jones Leap of Faith, and it's shown up in my life in so many ways (I could write for hours about spirit guides & pre-life planning & karmic balancing via reincarnation...but that's a little woo woo for the moment). TLDR: You're going to die. You don't need $125m to feel secure. The universe has your back.

@jamesthomason 🫡

@CSPopoff 🙏🏻🙏🏻

Beautiful, Brett. “I can handle anything thrown my way; the benevolent universe has my back”. Let the universe hold you in its arms, realize you are indeed one with it. What could possibly be more secure for this consciousness? Agree that the net, or the wind, shows up after you leap. I’m working on accepting that the failures were only ever relative to my expectations, and that the Universe still indeed had my back.

@justnblock Love it

I think I can simplify it for you a bit…I live my life 50% like im going to die tomorrow, 50% like I have many many year left. What does that mean? I call my grandma and tell her I love her, I do something I love everyday (tennis or driving my cars). I also invest in myself and my businesses, hoping to bear the fruit in the future. Ive found this the best way to live.

@vrexec Your writings have been and continue to be highly influential to me. A critical counterweight to the pervasive, mimetic "never enough" mentality around career, money & lifestyle. Please keep writing.

I was going to write this long reflection on freedom, money, self-employment, starting a business, etc…. I have it drafted… but does anyone really care? Does anything I muse about or share improve anyone’s life in any way?