uncoveredalpha

In Cowork, Claude can now build live artifacts: dashboards and trackers connected to your apps and files. Open one any time and it refreshes with current data.

In the 1995-2010 era, it took a lot of bodies to run a scaled Tiger-Cub strategy. 100+ headcount was common across research, back office & trading (starting in '08, I was one of a large research team) Largely due to technology, that headcount requirement declined dramatically: what once had to be built in-house was increasingly offered in a more effective and cost efficient matter on an outsourced basis with technology. Traders, in particularly, deeply understand the impact of technology from 1995-2015 on their role. You no longer needed a full time person sourcing short borrow or a full time person doing forensic accounting. This happened both on front office & back office roles: the idea that you need seven investors on a consumer team to cover that space seems anachronistic even before LLMs (I was one of seven). You see similar approaches now operated with ~15 people across front & back office. This efficiency benefit fed back into alpha compression...as headcount barriers to entry dissipated, traditional alpha in Tiger style investment approaches compressed as it was easier to practice that particular approach to investing. Today, the multi-strategy approach to long/short investing is ascendant. And it has been a very headcount intensive approach, with the large firms employing 3,000-5,000+ bodies. Is the technology impact to Tiger style investing a relevant prior for the next 5 years? With AI, can the multi-manager firm of the future operate with 500 bodies instead of 5,000? And what does that mean for alpha pools & the "peak pod" debate, talent demand and the future of fundamental investing?

I’ve had a few requests for prompts I use and have found success with, so here’s one I like for stress‑testing a single assumption in a 3‑statement model.
 The basic idea: upload your own model (I like Perplexity Computer for this task and I pre-select the GPT-5.4 model -seems the best for excel based tasks), then tell the assistant which assumption you want to attack — for example, Chipotle same‑store sales growth — and have it (1) benchmark that assumption versus history and category data, and (2) role‑play a skeptical PM to surface second‑order questions. You either come out with conviction that’s been stress‑tested, or a list of the 3–4 things you actually need to diligence. Example prompt: stress‑testing a same‑store sales assumption I’m going to upload my own 3‑statement model for Chipotle (CMG) that includes an explicit same‑store sales (SSS) assumption by year.
 Your job is to attack one assumption in that model: my base‑case U.S. same‑store sales growth for the next 5 years.
 Step 1 – Read and restate my assumption
 Ingest the uploaded model and identify the line item(s) that represent same‑store sales or comparable restaurant sales growth by year.
 Restate back to me, in plain English, what SSS path I am underwriting (e.g., “you are assuming ~7% SSS in year 1, then 5–6% annually through year 5”).
 Flag any inconsistencies between the SSS assumption and other model drivers (traffic vs. price mix, margin progression, unit growth).
 Step 2 – Historical and industry context check
 Using public data from SEC EDGAR or company-specific press releases, summarize Chipotle’s historical SSS and traffic/price mix over the last 10–15 years, highlighting periods of high growth, normalization, and drawdowns
 Compare my forward SSS assumptions to: Chipotle’s own historical SSS distribution and volatility.
 Recent SSS trends for major QSR/fast‑casual peers (McDonald’s, Taco Bell, Starbucks, Cava, etc.), noting whether the category is in broad acceleration, normalization, or deceleration.
 Tell me explicitly: relative to that backdrop, is my SSS path aggressive, conservative, or in line with history and peers?
 Step 3 – Structured adversarial role‑play I want you to run adversarial dialogues where you role‑play different skeptical PMs who each attack my SSS assumption from a distinct angle.
 For each persona:
 Run a short dialogue (5–10 exchanges) between that persona and me (as the analyst) focused only on whether my Chipotle SSS assumption is reasonable.
 Each persona should ask pointed, practical questions that a real operator or PM would ask, and surface second‑order issues rather than generic “what if growth slows?” prompts.
 Step 4 – Synthesis and decision
 From all personas, extract the 5–10 most important challenges to my SSS assumption, grouped into themes (e.g., “category traffic headwinds,” “unit growth cannibalization,” “LTO fatigue,” “labor/throughput ceiling”).
 For each theme, tell me: How big a swing factor it is to my SSS (e.g., could it move SSS by ±200 bps?). What specific evidence I should diligence with management or third‑party data (e.g., recent weekly traffic trends, saturation analysis, new‑unit AUV cohorts).
 Conclude with a one‑page style summary: “If you keep your current SSS assumption, here is the conviction statement you should be comfortable making.” “If you are not comfortable with that, here are the 3–4 targeted diligence items that would most update the SSS view.” Important constraints
 Do not rewrite or “fix” my entire model; stay focused on SSS and variables directly tied to SSS (traffic, pricing, mix, throughput, unit growth).
 Be explicit when you are inferring or generalizing from industry data vs. citing Chipotle‑specific history

Every time I post about this company on X, the replies come fast and they come ugly. “Boomer brand.” “Going to zero.” “Who even uses these products anymore?” I have been called an idiot more times on this one name than any other idea I have shared publicly in the last two years. I love it. There is a principle that has guided value investors since Benjamin Graham first wrote it down: the market is not always right, and the times it is most wrong are precisely the times when everyone agrees on the narrative. When a stock has nearly 14% of its float sold short, when the comment section fills with contempt, when the sell-side has a consensus “Hold” and the price is near a multi-year low, that is not evidence the thesis is wrong. That is evidence that expectations have been reset to a level where almost any improvement becomes a positive surprise. Here is what I will tell you in the free section, and it is already more than most people following this story have bothered to figure out. This company is being valued by almost everyone, including the sell-side analysts who cover it daily, using a completely wrong enterprise value. Not slightly off. Not a rounding error. The number most investors are looking at is roughly five times higher than the real number. The reason is a balance sheet technicality that takes about twenty minutes to work through, and almost nobody has done it. When you do the work and strip out the non-recourse financing subsidiary that has nothing to do with the core operating business, you find a company trading at an enterprise value of roughly one billion dollars on a business that has historically generated hundreds of millions in annual EBITDA and billions in revenue. You also find something else. A new CEO who just bought stock in the open market with his own money. A board member who did the same. Three hundred and forty-seven million dollars in share buybacks completed in 2025 alone, retiring eleven percent of shares outstanding in a single year. An investor day coming in May that could be the most important single catalyst this company has seen in a decade, specifically because management has signaled the return of the one product that dealers loved, that entry-level riders needed, and that the previous CEO killed for reasons that had more to do with his own preferences than with any rational business logic. The previous CEO is gone. He was pushed out after a proxy fight. The new man came from a consumer brand background, flew to Milwaukee, sat with dealers, and listened. His first moves were to stop doing the things that were destroying the business and start doing the things that the people actually closest to the product had been begging for. Meanwhile, there is a pension fund sitting on the balance sheet that is overfunded by nearly half a billion dollars. There are four owned manufacturing facilities plus a corporate headquarters building that would take hundreds of millions of dollars to replace. There is over a billion dollars in net cash at the parent company, completely separate from the financing subsidiary, that most investors do not realize is there because they are reading the consolidated statements without doing the work of separating the pieces. Nearly 14% of the float is sold short. That is not a warning sign. That is a coiled spring. I am a member of the Ben Graham school. I do not buy on hope. I buy on math. The math here, when you do it correctly, is among the most compelling I have seen in years on a company of this size, this brand recognition, and this operating history. The full analysis is below for paid subscribers. It covers the balance sheet deconsolidation in detail, the real enterprise value, the valuation case, every catalyst I can see between now and year-end, and what I think this is worth on a conservative recovery scenario.

Waller is willing to "look through" energy price inflation provided that trade flows through Hormuz resume.

I'm actually impressed with the ChatGPT plug-in so far. The below are the results to "make me a DCF valuation based on this model". AI is going to be a real problem for junior finance jobs.