S&P 500 (SPY), the thing to beat: -1.41% on $100,000 since Aug 24, 2026
Most mispricing is not a disagreement about numbers, it is a CATEGORY ERROR. The market prices a company by the label it wears — this is a retailer, this is a biotech, this is a chip company — and the label determines the multiple long before anyone examines the cash flows. The opportunity is a business whose economics have quietly stopped matching its category. The mismatch is different every time — what a company actually sells, who actually pays, what the customer cannot easily stop doing — and naming it correctly for THIS business is the work. Reach for the comparison the facts in front of you suggest, not the one that worked last week; an analogy you could apply to any name is not an insight, it is a habit. Ask what would have to be TRUE for today's price to be correct, then ask whether anyone actually believes that. Reason across industries — the useful analogy is almost never from the same sector, because the same sector is where everyone is already looking.
6 to 9 names. Highest-conviction position may reach 20%; nothing below 6% (a position too small to matter is a fee, not a view). At most 3 changes per rebalance.
Holds a portfolio of common stocks included in the S&P 500 index, with each stock weighted to match its weight in the index.
Cash the rules would not let this book spend sits in the index rather than earning nothing. Not a pick.
Sells annual home warranty plans covering repair or replacement of home systems and appliances under the American Home Shield and other brands in the US.
This is a prepaid-contract business carrying a home-services label. Customers pay annual warranty premiums up front and claims are paid later. The figures show TTM FCF of $386M on a $5.8B market cap (6.7% yield), ROIC of 22%, a 4.5% share-count reduction and net debt of about 1.0x EBITDA ($1,176M debt less $627M cash, against $525M EBITDA). Conviction is lowered from my entry for two reasons: quarterly revenue growth was only +4.7% YoY, and the float mechanics are not yet verified by a deferred-revenue figure in front of me.
Develops and sells drugs for blood cancers, graft-versus-host disease and other cancers, with JAKAFI (myelofibrosis, polycythemia vera) listed first among its marketed products; it also has skin-disease and tumor drugs in trials.
The figures describe a franchise: 92.6% gross margin, revenue +26.9% TTM and +37.7% in the latest quarter, an 8.2% FCF yield, and $3.98B cash against $38M debt. The stock is priced at 14.4x trailing earnings, as a patent cliff. On the analyst's bear case, the 2028 Jakafi patent expiries are what the 14x multiple already discounts. The ~$1,270M IPR&D charge is a one-off that I will not treat as an exit signal. Drift from 16.3% is trimmed back to target.
Provides grocery delivery and fulfillment technology to retailers and advertising solutions to consumer brands through the Instacart marketplace and enterprise platform.
The economics are retail media, not grocery delivery. Ads were 28.5% of Q2 revenue, TTM gross margin is 72.6%, FCF is $1,176M on a $14.45B market cap (8.1% yield), and the company holds $757M cash against $34M debt. The label has partly corrected: P/E is now 33.3 against 27.3 at entry and the stock sits at 100% of its 52-week high. The 11-05 print decides whether the ad share holds.
Sells vehicle history reports (CARFAX) to consumers and dealers, and automotive data and forecasting to carmakers and suppliers. Notable: renamed from S&P Global Mobility Holding Company in February 2026.
CARFAX is a toll on used-car transactions, priced as an orphaned spin-off of an auto vendor, with a 7.0% FCF yield. Using the 10-Q figures, net debt is $1,795M ($1,981M − $186M), about 2.9x the $618M TTM EBITDA, against the confirmed 3.50x covenant. On the analyst's bear case, standalone costs are the real risk. The leverage line is the test, and today it has about 0.6x of headroom.
Sells a software platform exclusively to property and casualty insurers covering policy administration, claims, billing, underwriting, pricing, and analytics; a single-vertical software provider.
Guidewire is the system of record for P&C carriers. ARR was $1,242M at 7/31/26, +19% in constant currency. FCF is $359M and GAAP operating margin is 10.2%, with no carrier loss disclosed. At 112.9x trailing EPS and 8.1% ROIC, the price already asks for most of what the thesis offers. I hold it at its drifted weight and do not top it back up to 13%.
Develops CAR-T cell therapies for cancer; lead product cilta-cel targets multiple myeloma and is developed and commercialized under a collaboration with Janssen Biotech.
This is a New Jersey-headquartered CAR-T franchise with Janssen: revenue was +51.9% YoY in the latest quarter, and five-year CARVYKTI data show half of patients progression-free. It is priced as China-risk biotech at 56% of its 52-week high. It stays at the floor weight because the geo leg is unchecked and TTM FCF is −$122M, with cash down from $902M to $673M.
Returns are decided on the supply side, not the demand side. Almost everyone forecasts demand, which is hard and widely attempted; far fewer watch capacity, which is observable and mostly ignored. When capital floods an industry — capex rising, entrants arriving, everyone expanding into an obvious opportunity — future returns are being competed away no matter how good the demand story is. When capital flees — capex cut, capacity closing, consolidation, an industry nobody wants to fund — the survivors inherit pricing power. So the question is never 'is this a good business today', it is 'which way is capital moving, and who is left standing at the other end'. This is a slow lens. It is usually early and it is usually uncomfortable, because the moment capital flees is the moment the news is worst.
8 to 12 names, nothing above 14%. At most 4 changes per rebalance — this lens is meaningless at high turnover, and a change counts a sell as well as a buy, so four is two swaps and not four new ideas.
Sells combustible cigarettes under brands including Camel, Lucky Strike, and Newport, plus non-combustibles including Vuse vapes, glo heated tobacco, and Velo nicotine pouches.
This is the cleanest capital-exit name in the book. Invested capital is down 30.1% while revenue is flat (-1.0%), and ROIC has risen from 6.8% to 11.3%. FCF is £7.5B against £0.94B of capex, and nothing in the pitch book shows new capital entering combustibles. Thesis unchanged; nothing new has arrived since entry.
Makes consumer products including torches, propane cylinders, and helium kits, plus pressurized building products such as refrigerant cylinders and water tanks, sold under multiple brands.
Capacity in its markets is down 44.1%, revenue is up 18.4% TTM and ROIC has doubled to 12.0%. Survivors are collecting both volume and price. The CEO confirmed 'price actions where we thought we needed to', but no cylinder price percentage was disclosed, so pricing power is inferred, not measured. The gross-margin tripwire is how I police that.
Provides retail media and performance advertising technology connecting brands, retailers, and media owners; operates commerce-focused supply-side and demand-side ad platforms.
Criteo is an ad-tech consolidation survivor at 7.4x earnings with $177M of FCF on a $709M market cap, and its share count is down 6.5%. The position is -16.7% since entry and quarterly revenue is -11.3%. I hold and do not add until the 2026-11-04 print tells me whether the client-specific issues are structural.
Cultivates, processes, and retails cannabis products through dispensaries in eight US states, with its largest dispensary concentration in Florida.
Sector capacity is down 46%, Trulieve's revenue fell only 3.9%, and ROIC rose from 1.7% to 8.8% on $186M of FCF. Florida licensure stays closed after SB 1242 died in committee. The open question is whether rival raises (Curaleaf $500M notes, GTI +$50M, Verano $195M) fund new capacity or only refinance. That is unresolved, so conviction stays at medium.
Founded in 1807, publishes scientific and academic journals and educational books and digital courseware through Research and Learning segments, serving libraries, universities, and professionals worldwide.
ROIC has risen from 2.9% to 10.1% on a shrinking base, journal retention is above 99%, and Research revenue grew 12% last quarter. Leverage has risen from the ~2.7x I bought on to 3.3x on the pipeline's GAAP reading. That leaves only 0.2x of headroom to my 3.5x line while HY spreads widen, so this is the name I am watching hardest.
Sells connected cutting machines, accessories, and materials for consumer crafting; also offers Cricut Design Space software and Cricut Access subscription service.
Invested capital is down 48.9%, debt/OCF is 0.06, and $148M of FCF sits against a $1.33B market cap, with $267M of cash. Latest-quarter revenue of -9.2% keeps category permanence as the weak link. Gross margin is the tell for whether a funded competitor is taking price.
Makes radiofrequency-based minimally invasive and non-invasive aesthetic medical devices for body contouring, skin tightening, and women's health, sold in the US and internationally through distributors.
InMode trades at 0.58x EV/invested capital against an 18.5% 4-year mean ROIC. It has $396M of cash, about half the market cap, and negligible debt. Invested capital is down 13.2% in two years. FY guidance of $365–375M was held, and my thesis does not depend on the 10-30 offer, so I expect and accept a drawdown if the offer lapses.
Sells private mortgage insurance in the US that protects lenders against borrower default on individual home loans; also offers contract underwriting and reinsurance services to mortgage originators.
MGIC trades at 9.0x earnings with $2.7B of PMIERs excess, and its share count is down 10.9%. The analyst found no new private mortgage insurer, and PMIERs 2024-01 became fully effective on 2026-09-30, which raises the capital bar for any entrant. Weight stays below 10% until the Q3 loss ratio is seen.
Holds a portfolio of common stocks included in the S&P 500 index, with each stock weighted to match its weight in the index.
Cash the rules would not let this book spend sits in the index rather than earning nothing. Not a pick.
Start at ten equal weights and make the book earn every change from there. The job is not to find the best stock, it is to own a set of positions whose risks are genuinely different from one another. Before anything else, ask what this book is actually exposed to: not what sector a name is labelled, but what has to go wrong for several positions to fall together. Two buy-now-pay-later lenders are one bet on the consumer wearing two tickers. Six biotechs are one bet on the funding cycle. And a position with no clear reason it should stop being cheap is not an investment, it is an inventory item — if there is no catalyst and no sell-off worth exploiting, the honest answer is to not own it.
Exactly 10 names, 10% each at inception. Afterwards: at most 2 sells and 2 buys per rebalance, weight changes capped at 3pp per name, nothing above 15%.
Sells software, consulting, and server and storage hardware, plus financing for clients' purchases, across four segments; software covers hybrid cloud and AI platforms, and consulting covers technology services and managed services.
It is the only holding whose catalyst has already printed and passed. Software grew +8% against the bear's ~5%, guidance was not cut, and FCF was guided up by more than $1B. At 21.6x with $14.0B of TTM FCF it is the least momentum-dependent name in the book, so it stays at the top weight.
Develops and sells drugs for blood cancers, graft-versus-host disease and other cancers, with JAKAFI (myelofibrosis, polycythemia vera) listed first among its marketed products; it also has skin-disease and tumor drugs in trials.
The Jakafi franchise trades at 14.4x trailing earnings with a 93% gross margin, $3.98B of cash against $38M of debt, and TTM revenue up 26.9%. The analyst's bear case is the 2028 patent cliff plus a ~$1.27B Q3 IPR&D charge. The charge is a one-off that I will not treat as an exit. The cliff is real and is why I am not adding. Oct 29 decides the Jakafi leg.
Sells vehicle history reports (CARFAX) to consumers and dealers, and automotive data and forecasting to carmakers and suppliers. Notable: renamed from S&P Global Mobility Holding Company in February 2026.
CARFAX-anchored spinoff at about a 7% FCF yield. Net debt per the 10-Q is about $1.8B, roughly 2.9x GAAP EBITDA, against a 3.5x covenant. The leverage line is checkable only on the Q3 balance sheet, so I hold to that print.
Develops and distributes energy drinks globally under the Monster Energy brand through authorized bottlers; also holds a portfolio of alcohol beverage brands.
My GM<56% falsifier has fired (computed 55.80%), and I already cut 3pp on that same reading. I am not cutting twice on one TTM number five sessions before the quarter that tests it. Revenue is still growing ~20%, FCF is $2.08B, and both price legs are intact. If the Q3 quarterly GM on 11-05 is also below 56%, I exit.
Sells chips and systems for data center AI computing and networking, plus GeForce and RTX graphics chips for gaming and workstations; also automotive platforms. Data center products are sold to cloud providers, OEMs and distributors.
This is the book's one deliberate AI-capex position: +106% quarterly revenue, a 65% TTM operating margin and $127B of FCF. Both momentum legs are intact (90.3% of high, 12-1 +21.1% against the +8.9% median). I hold and do not add, because it has the thinnest margin over SPY of any momentum name and sits at geo watch 2 of 3.
Sells annual home warranty plans covering repair or replacement of home systems and appliances under the American Home Shield and other brands in the US.
Carry position and the book's only housing-services exposure: $386M of FCF, a 4.5% lower share count, 22% ROIC. I would not buy it today. It is held to the Nov 5 print as committed, and quarterly YoY growth is already only +4.7%.
Develops CAR-T cell therapies for cancer; lead product cilta-cel targets multiple myeloma and is developed and commercialized under a collaboration with Janssen Biotech.
Commercial CAR-T with revenue up ~52% YoY and cash exceeding debt by $0.39B, at 56% of its high. FCF of -$122M and the unchecked geo leg are why this is a low-conviction holding.
Sells a software platform exclusively to property and casualty insurers covering policy administration, claims, billing, underwriting, pricing, and analytics; a single-vertical software provider.
The business is undamaged: ARR is $1,242M (+19% cc), TTM revenue is up 22.7%, and GAAP OM is ~10.1% against my 8% line. But the September sell-off I bought has reversed, and there is no announced catalyst. By my own rule this is inventory. I keep it this run only because every alternative destination for the weight reports within days.
Sells consumer cybersecurity, identity protection, and privacy products under multiple brands—Norton, Avast, LifeLock, AVG, and CCleaner—targeting individuals, families, and small businesses.
Consumer-security subscriptions at 14.3x P/E, a 10.6% FCF yield and a 43% OM, sold off 28.5% on a reported ~$12B GoDaddy approach. The bear case is a debt-funded deal on top of ~$7.6B of existing net debt. It is unanswered until tomorrow's print, which is exactly why the position stays at 3%.
Manufactures aircraft avionics, engines, auxiliary power units, braking systems, and other aerospace components for commercial, military, and business aviation; formerly a Honeywell International subsidiary.
The falsifier fired at ~4.8x net debt/EBITDA against my 4.375x line. My own arithmetic says one quarter of ~$0.6B FCF only gets to ~4.6x, so the exit is essentially decided. The stub holds one more session to the Oct 28 print, as committed, and is sold next run unless the ratio is inside 4.375x.
Value a stock the way a buyer values a whole company: what the business would fetch in a sale, conservatively underwritten, and whether the market is offering it at a discount to that. Recurring revenue, retention and gross margin decide the multiple. The balance sheet and the capex it takes to stand still come before any story. A value investor's discipline and a buyer's eye.
5 to 8 names, nothing above 25%. At most 4 changes per rebalance. Holding the index rather than owning something merely acceptable is a valid answer.
Holds a portfolio of common stocks included in the S&P 500 index, with each stock weighted to match its weight in the index.
Cash the rules would not let this book spend sits in the index rather than earning nothing. Not a pick.
Develops and sells drugs for blood cancers, graft-versus-host disease and other cancers, with JAKAFI (myelofibrosis, polycythemia vera) listed first among its marketed products; it also has skin-disease and tumor drugs in trials.
93% gross margin, $1.89B TTM FCF, and EV of about $19.5B (~9.4x EBITDA). It is still cheap, but I cannot defend it as a 15% position. The analyst's bear case is right that Jakafi patents expire June/December 2028 with confidential settled entry dates. That makes much of the stream a two-to-three-year runoff, and I cannot split Jakafi from the rest of revenue to value what is left. I trim to 12% and hold the rest into the 10-29 print, where revenue growing 37% says the non-Jakafi business is real but unmeasured.
Sells vehicle history reports (CARFAX) to consumers and dealers, and automotive data and forecasting to carmakers and suppliers. Notable: renamed from S&P Global Mobility Holding Company in February 2026.
CARFAX/Polk is the kind of data franchise a strategic or sponsor buyer underwrites at mid-teens EBITDA. The market offers it at ~12.4x EV/EBITDA with $413M FCF on $28M capex, which is about a 25% discount to my value. The analyst's bear case (standalone costs, 2.9x leverage against a 3.50x covenant) is the live risk. Leverage has room before the covenant, but the covenant's EBITDA definition is still unlocated, so I hold at 8% and do not add.
Sells chips and systems for data center AI computing and networking, plus GeForce and RTX graphics chips for gaming and workstations; also automotive platforms. Data center products are sold to cloud providers, OEMs and distributors.
This is a franchise holding that my own discipline does not support: a 2.4% FCF yield at 27x trailing EPS offers no discount. I keep it only because 72% ROIC, a 71-75% gross margin and EPS growth of 125% are still confirming. Both momentum tripwire legs are intact (12-1 +21% vs +8.9% median; 90% of high). I will not add, and either leg firing is an exit.
Sells annual home warranty plans covering repair or replacement of home systems and appliances under the American Home Shield and other brands in the US.
I hold this for the carry, not a discount. EV is ~$6.3B (12.1x EBITDA), with a 6.7% FCF yield, a 4.5% share shrink, and annual contracts paid upfront. Some of that $627M cash is likely needed to pay claims rather than available to owners, so I do not count all of it. Quarterly growth slowed to 4.7%, and the 11-05 print is where the renewal book has to show up.
Sells creative software, marketing and analytics platforms, and publishing tools to enterprise customers and individuals worldwide, primarily via subscription and license.
Subscription software with an 89% gross margin, 41% ROIC and 0.8% capex, growing revenue ~13%. At ~$98B market cap it is ~9.7x EV/EBITDA and a 10.5% pre-SBC FCF yield, which is well below what a process would clear for this profile. I hold rather than add because SBC is still unfound, and I am aware I am drawn here largely by cheapness.
Encode the mechanisms the evidence supports and let them run. Every name in this book is here because it passed arithmetic — a forced-selling window, a price dislocation that has not closed, an R&D backlog in the sector's top decile, a monopoly still making new highs. No model is consulted before a name is chosen. The only thing a language model does in this pipeline is write the explanation afterwards.
Mechanism weights, 10% max per name, 30% per mechanism, no minimum hold: a name that leaves the proposal is sold, unless it left only on an entry gate (rank, entry window, off-low cap), when it is held until its price target, research horizon or a forced exit.
Sizing guardrails the code enforces on the control arm: no single name above 10% of the book, no mechanism above 30%, a 10% cash floor, and a three-year minimum hold that only a veto, the incremental return-on-capital tripwire or a monopoly trim can override.
Holds a portfolio of common stocks included in the S&P 500 index, with each stock weighted to match its weight in the index.
Cash the rules would not let this book spend sits in the index rather than earning nothing. Not a pick.
Sells the proprietary FICO credit score to lenders and directly to consumers, and offers analytics and decisioning software; the credit score is the core named revenue segment.
The bet is that FICO's 32.6% fall in five days around 2026-09-30 overshot whatever actually changed in the business, and that the price recovers as forced and frightened selling fades. The stock has already regained 19.5% from its low. No headline provided explains the drop, so nothing shown suggests the business itself is impaired. The figures describe a business that still earns well: an 82% gross margin, a 47% operating margin and a 128% return on new capital over three years. FICO's Scores segment sells credit scores that clients build into their transaction and decision processes, and its Software segment sells decision tools. A fall this sharp in a company like that looks more like selling pressure than lost earning power. We are positioned for a move toward $879.62, about 24% higher, over 6 to 18 months, and we re-examine on 2026-11-04. Evidence that earnings or margins are actually slipping would prove the damage real and end the case.
Offers buy-now-pay-later installment payment options at checkout for US and Canadian consumers, with subscription tiers that extend usage to any merchant online or in-store.
The bet is that Sezzle's 28.1% fall in five days around 2026-08-12 overshot any real damage, and that the price recovers as frightened selling fades. It has already recovered 19.2% from the low. No headline provided explains the drop, so nothing shown suggests the business is impaired, and the figures look sound: a 70% gross margin, a 36% operating margin and a 71% return on new capital over three years. Sezzle lets shoppers split purchases into installments such as Pay-in-Four, so the risk to watch is whether customers repay, but no source shows that deteriorating. We are positioned for a recovery toward $174.40, about 37% higher, over 6 to 18 months, with a re-evaluation on 2026-11-04. Evidence of rising customer defaults or shrinking margins would show the damage is real and end the case.
Sells consumer cybersecurity, identity protection, and privacy products under multiple brands—Norton, Avast, LifeLock, AVG, and CCleaner—targeting individuals, families, and small businesses.
Gen Digital, the maker of Norton, Avast, LifeLock and MoneyLion, fell about 28% in five days around September 28. The headlines tie the selling to news that it wants to buy GoDaddy: one asks why its own stock crashed 12% on September 25 even as it wanted to buy the company. That reads as a fear about the price and risk of a deal rather than damage to earnings. On September 28 the CEO pointed to a recently raised fiscal 2027 outlook (revenue growth of 9% to 11%, non-GAAP earnings per share growth in the mid-to-high teens) and said the quarter and full year are tracking toward the higher end, and a September 29 article noted analysts revising earnings estimates higher. With a 43% operating margin, we are positioned for the price to recover toward $29.15 as the deal worry fades. The evidence that would prove the damage real is a signed acquisition on poor terms, or guidance being cut.
Makes high-speed connectivity chips, cables, and transceivers for Ethernet and PCIe applications, and licenses SerDes IP into data networking markets.
Credo fell 31.7% in five days around Sep 3, and no provided headline explains why, so no source shows the business itself is impaired. The numbers we can see point the other way: a 68% gross margin, a 33% operating margin, and returns on new investment of 26.7% over three years. The bet is that a fall this sharp is mostly forced and frightened selling, and the stock is already 15.1% above its low. We are positioned for a recovery toward $240.24, about 39% higher, over 6 to 18 months, and we will reassess on Nov 30. Evidence that margins or sales are actually deteriorating would show the damage is real and end the bet.
Designs mixed-signal audio chips, camera controllers, and power ICs for smartphones, PCs, tablets, wearables, automotive entertainment, and professional audio applications.
Cirrus Logic's stock is down 13.9% over a year, and on Oct 15 KeyBanc's analyst John Vinh downgraded it to Sector Weight, saying about 80% of revenue is tied to Apple, which has cut wafer starts equivalent to 15 to 20 million iPhones, so growth looks limited for 6 to 12 months. The R&D bet looks past that: Cirrus has spent about 26% of its market value on research over four years, and that spending is expensed immediately, so none of it shows on the balance sheet. What it built, per the filed summary, is a range of chips for audio, camera control, haptics and sensing, and battery power in smartphones, plus products for PCs, wearables, headsets, automotive and industrial uses, with amplifiers, codecs and signal processors at the base. We expect that breadth to turn into revenue beyond its main phone customer, though KeyBanc flags a PC diversification opportunity as likely pushed out by high memory prices. The company has more cash than debt, so it can wait. The Nov 4 results will show whether non-phone products are growing; if they do not, the unbooked asset is not turning into earnings.
Makes and licenses toys, games, and trading cards including Wizards of the Coast brands; licenses IP for entertainment, apparel, and third-party digital game development.
Hasbro makes and licenses toys, trading cards and games, and its position rests on owned brands and characters, including Wizards of the Coast games, that it can sell as products, license to others and turn into digital game experiences. The figures show a business confirming itself: revenue up 17%, gross margin of 65%, and a share price at 92% of its 52-week high after a 37.8% rise over the year. The recent headlines add to this: on October 13 Hasbro opened a Wizards of the Coast studio in Montréal focused on Dungeons & Dragons content, and another headline asks whether expansion of its brands into experiences in China and Brazil strengthens the bull case. We expect continued results to keep pulling the price higher, with the earnings report previewed on October 6 as the next test. If growth and margins stop confirming the story, the reason to hold it is gone.
Runs a demand-side platform letting ad agencies and advertisers buy and optimize digital ad campaigns across CTV, video, display, audio, and other channels.
The Trade Desk's stock fell 29.9% in five days around August 11, 2026 and has not recovered from that low; no provided headline explains what triggered the move, and that silence is itself part of the case — no source points to a product failure, a lost customer, or a competitive shift that would explain permanent earnings damage. A platform running 79% gross margins and 20% operating margins, used by advertising agencies and advertisers to manage campaigns across connected TV, video, display, audio, and mobile, does not lose its economics in five days without a traceable cause. The bet is that a fall of this size and speed reflects forced or panicked selling that overshot the real situation, and that as that pressure fades the price recovers toward the $19.34 target — 56.7% above entry. What would prove the damage real is a quarter showing revenue contraction or margin compression with a specific named cause; until that evidence appears, the accounts describe a business that is intact.
Sells vehicle history reports (CARFAX) to consumers and dealers, and automotive data and forecasting to carmakers and suppliers. Notable: renamed from S&P Global Mobility Holding Company in February 2026.
Mobility Global began NYSE trading on June 26, 2026, spun out of S&P Global — confirmed by the July 4 Simply Wall St. article and by the filed business summary, which states the company was formerly known as S&P Global Mobility Holding Company. The SEC Form 10-12B filed May 7, 2026 by registrant 'Mobility Global Inc.' is the formal separation record. The standalone business operates CARFAX, automotiveMastermind, Polk Automotive Solutions, and Market Scan, providing data, analytics, insights, and technology across the full vehicle lifecycle to automakers, suppliers, dealer groups, media, financial institutions, and consumers. S&P Global shareholders who received MBGL shares owned a diversified financial data company, not an automotive data specialist, and many are selling without reference to what Mobility Global earns or is worth on its own terms; the July 4 article flags that shares are highly illiquid, meaning bid-ask spreads are wide and the selling pressure can overshoot further before it exhausts itself. The position buys that indiscriminate supply, now 59 days post-distribution, and profits when a buyer base that values a standalone vehicle-data platform sets the price.
Develops and sells drugs for blood cancers, graft-versus-host disease and other cancers, with JAKAFI (myelofibrosis, polycythemia vera) listed first among its marketed products; it also has skin-disease and tumor drugs in trials.
Incyte's franchise is Jakafi, which the filed summary lists for myelofibrosis, polycythemia vera and steroid-refractory graft-versus-host disease, with Iclusig, Monjuvi and Niktimvo alongside it and further candidates aimed at myeloproliferative neoplasms. The margins show a defended position: a 93% gross margin, revenue up 37%, and a return on new capital of about 119%. What the market priced is a company on 14.0 times trailing earnings, against 29.0 for Vertex, even though a Trefis comparison dated 2026-10-09 shows Incyte's trailing revenue up 26.9% against 10.2% at Vertex, with a 31.6% operating margin. The bet is that this gap closes upward if growth keeps showing up in the third-quarter report. The warning sign is that Barchart expects a steep fall in the bottom line versus the year-ago quarter. We hold while results keep confirming the franchise, and exit if the growth or margins stop doing so.
Develops CAR-T cell therapies for cancer; lead product cilta-cel targets multiple myeloma and is developed and commercialized under a collaboration with Janssen Biotech.
Legend Biotech's shares are down about 42% over the past year, and an article dated 4 October put its last close at $18.57, while the research behind its lead product is already paid for and expensed. That lead product is cilta-cel, a CAR-T cell therapy for multiple myeloma developed under a collaboration with Janssen, backed by earlier-stage CAR-T candidates for several other cancers. A 25 September headline reports that half of patients with relapsed or refractory myeloma remained progression-free and alive five years after a single infusion of CARVYKTI, and a new CEO, Ingrid Zhang, was appointed on 15 September, which a Stocktwits headline said prompted buyout speculation. The risks are real: UBS flagged pressure from rivals ahead of a key FDA decision, and an article cites a loss of $91.5 million. We expect that five-year durability, and the pipeline behind it, to be reflected in sales and profit that the market has not priced, and we would be wrong if CARVYKTI loses share to rivals.
Sells two approved drugs—NUPLAZID for Parkinson's disease psychosis and DAYBUE for Rett syndrome—with several neurological candidates in phase 1–2 trials.
The market has marked Acadia down hard, with the shares off about 28% over 30 days after the RADIANT Phase 2 trial of remlifanserin in Alzheimer's disease psychosis missed and analysts cut price targets, yet the company is pressing on and the data are reported to guide the drug into Phase 3. Years of research expensed through the income statement built two marketed medicines, NUPLAZID for hallucinations and delusions in Parkinson's disease psychosis and DAYBUE for Rett syndrome, plus a pipeline that includes remlifanserin, ACP-211 for major depression, ACP-711 for essential tremor and ACP-271, and none of this sits on the balance sheet as an asset. A new patent allowance for weight-banded dosing of trofinetide extends U.S. protection to March 2041, so that spending can keep paying out for longer. We bet the market is pricing one failed trial as if it erased the whole pipeline, and that the two products and remaining programs will earn more than a $3B valuation implies. We would be wrong if remlifanserin stumbles again in late-stage trials or if DAYBUE and NUPLAZID fall short of the roughly US$1.7b of 2028 sales that management is targeting.
Makes motorcycles, automobiles, power products, and HondaJet aircraft, and provides retail and dealer financing; operates across Japan, North America, Europe, and Asia.
Honda makes motorcycles, cars, power products such as generators and lawn mowers, and HondaJet aircraft, and it has financing arms for customers and dealers. It has spent heavily on research, equal to 52.9% of its market value over four years, the highest ratio among 133 industrial companies; accounting expenses that spending as it is incurred, so the value of what it built is not carried as an asset. Recent headlines show projects it is involved in, though the sources do not say how much of the research went to any of them: a partnership with Applied Intuition on software-defined vehicles (October 5), an exploration of space robotics with Redwire (October 5), and a Triton-based pickup truck for Southeast Asia (October 12), which the source describes as a collaboration with its Alliance partners. We expect the spending to appear as products and profits the market has not priced, over a 12-24 month hold. Honda holds more cash than debt, so it can fund this work while we wait; if the spending keeps producing no new products or profit, the bet is wrong.
Sells medicines for rare diseases including Duchenne muscular dystrophy, SMA, and AADC deficiency; Evrysdi (SMA) is developed and distributed in collaboration with Roche.
PTC Therapeutics has spent heavily on medicines for rare diseases, with four years of research costs equal to about 43% of its market value, and the accounts record none of that as an asset. The filed summary lists what the spending has produced or is building toward: Sepiapterin for phenylketonuria, a splicing medicine (PTC518) for Huntington's disease, and vatiquinone for Friedreich's ataxia, alongside marketed Duchenne muscular dystrophy treatments. Recent headlines show the pipeline advancing: on 24 August PTC released new Sephience (sepiapterin) data showing sustained metabolic control and a consistent safety profile, and on 21 September it completed the acquisition of ST-920, a Fabry disease gene therapy. The shares are down about 5% over the year, and one 20 September headline argues they could be 36% undervalued after pipeline progress. We expect these products to turn expensed research into reported revenue and profit the market has not yet priced, and the bet is wrong if Sephience sales or the later-stage programs fail to deliver.
Sells chips and systems for data center AI computing and networking, plus GeForce and RTX graphics chips for gaming and workstations; also automotive platforms. Data center products are sold to cloud providers, OEMs and distributors.
The franchise is NVIDIA's data center computing and networking business, which the filed summary describes as AI infrastructure sold to cloud providers, system makers and others; a 71% gross margin and 132% revenue growth show customers still pay up for it. The bet is that a franchise that keeps confirming itself in results keeps getting valued higher, and the market is not paying a premium yet. The 13 October coverage shows it: the shares trade at about 25.7 times forward earnings, roughly 39% below their five-year average of 42.2 times, with earnings expected to grow about 95% this fiscal year. The same article cites a Goldman Sachs estimate of $4 trillion to $8 trillion of AI-related investment over five years, and former Wedbush analyst Dan Ives calls NVIDIA the foundational compute platform for that spending. We expect the stock to keep being re-rated upward while results keep arriving at this pace. What would end it: a 247wallst.com piece notes Marvell, a custom AI chip designer, is up 214% this year against NVIDIA's 22%, so a visible shift of customer orders from NVIDIA's general-purpose chips to custom designs would be the signal to sell.
Pharmaceutical company selling prescription drugs globally across diabetes, obesity, oncology, and immunology; Mounjaro (type 2 diabetes) and Zepbound (obesity) are its highest-profile current products.
Eli Lilly's franchise is its obesity and diabetes medicines, including Mounjaro and Zepbound, alongside insulins and a growing immunology and cancer range. The margins show how well it is defended: 83% gross margin, revenue up 77%, and a fool.com headline saying profit more than doubled to $20.6B. The recent headlines keep confirming the story: a 13 October article reports exploratory Phase 3b data showing Taltz combined with Zepbound moved more immune and metabolic markers than Taltz alone, and another recent result combined eloralintide with tirzepatide for obesity and diabetes in a Phase 2b trial. One article's narrative projects $118.9 billion revenue by 2029, and the stock sits at 92% of its 52-week high. We hold while results keep confirming this; the sources name pricing pressure on GLP-1 drugs and litigation exposure as what would end it.
Agriculture company selling crop protection products against weeds, insects and disease, plus seeds and nitrogen stabilizers; formed in 2018, with operations in the US, Canada, Europe, Latin America, Asia, Middle East and Africa.
Corteva is a pure crop protection company: it sells herbicides, insecticides and other products that protect crops from weeds, insects and disease, and it has just completed the spin-off of its seed business into a separate company, Vylor. The screen shows it was removed from a major index on 2026-10-06, and the stated reason is 'Market capitalization changes', meaning a size rule, not a judgment on how the business is run. Index funds must sell a removed stock whatever it is worth, and that mechanical selling tells us nothing about farmers' demand for Corteva's products; the company produced $2.6 billion of operating cash over the past year and carries little debt relative to its earnings (net debt is 0.4 times EBITDA). We are positioned for the price to recover once the forced sellers have finished, and we expect that over the 3-5 year hold. What would prove us wrong is evidence that the crop protection business itself is weakening, not just the share price.
Last scan Nov 3, 04:21 UTC. The four managers choose from the same pitched names; their books hold 31 distinct names between all five books.
A business with wide margins whose stock fell at least 18% inside five sessions at some point in the last 90 days. The screen does not know why it fell, only that the drop was violent and the margins underneath were worth defending, so the name goes on the list with the event window attached and the price before the drop as the recovery target.
Held now: FICO, CRDO, GWRE, SEZL
Four years of R&D spending added up and divided by market cap, ranked inside the sector, top decile only, with a balance sheet that can survive. Expensed R&D never shows up as an asset, so the market is often pricing the losses and not what the losses bought. Bought as an equal-weight basket, because the best name on this screen is where the payoff lives and nobody has the nerve to size it alone.
Held now: HMC, PTCT, ACAD, LEGN, CRUS, INCY
Large cap, high margin, growing fast and within 15% of its 52-week high, ranked on gross profit over assets. Stay on the businesses the market has already decided are monopolies for as long as the tape agrees, with a written rule for when it stops agreeing: below 85% of the high, or behind the universe on twelve-month momentum, and the position gets trimmed and reviewed.
Held now: HAS, LLY, MNST, GEN
Stocks thrown out of the S&P 500 for a reason other than a takeover or a bankruptcy, bought within 30 days of the removal as long as the business still generates cash. Index selling is forced selling. Equal weight, small positions, held three to five years. The anti-trap veto deliberately does not run here, because cheap and hated is the point. Silent most weeks by design.
Held now: CTVA
Companies spun out of a parent, bought between day 20 and day 120 after they start trading, which is roughly when the parent’s shareholders who never asked for the shares have finished selling them. Half-size positions, the veto runs in full, held one to three years.
Priced so a control buyer could pay a premium and still make it work. The public market prices equity; a control buyer prices the whole enterprise and underwrites how much debt it can carry. A business cheap on enterprise value and carrying almost no net debt is one a sponsor can pay up for and still make the returns, a real discount to private-market value rather than a low multiple on a levered stub.
Held now: CRTO, CRCT, TTD, CRUS
Capital withdrawn, returns on what is left rising. A firm removing capital from its industry while its returns on the remaining capital go up and revenue holds. The market prices shrinking invested capital as decline and almost never separates shrinkage with improving economics from shrinkage with deteriorating economics, and those two have opposite forward returns.
Held now: TRLV, WOR, BTI, WLY
Held now: CRUS
A business repriced as cyclical whose earnings never cycled. The market marked the price down as if profits were cyclical; the realised multi-year operating profit series says they never fell below a level they had already reached. The mismatch is between the risk the price is discounting and the risk the filings have demonstrated.
Held now: ADBE
The float shrinking, paid for out of operations. A cheap name with no catalyst is inventory, not an investment. The most reliable catalyst is the one needing nobody’s permission: a company retiring its own shares out of operating cash while the stock is still below its high. Per-share value compounds whether or not the market ever re-rates it.
Held now: INMD, CRTO, ADBE, CRUS
Revenue growing while the capital base stands still. A company that grows meaningfully while its invested capital and its share count both hold flat has stopped being the capital-consuming business its filings describe. That transition is invisible in any single-year margin and usually gets attributed to cost discipline rather than to a change in what the business is.
Held now: MNST
Cash left after the capital the business had to absorb. Every cash-yield screen quotes a pre-capex number, because the data has no capital-spending line. The growth in invested capital over a multi-year window is the capital the business absorbed to stay where it is. Subtract it and what remains is closer to what an owner keeps.
Held now: CRTO
Two to three years after a company’s own capital peaked. The capacity is gone, the write-offs are taken, and the same or greater absolute profit is now earned on a materially smaller base. The market prices the news of the retreat; it does not price the arithmetic of the survivor.
Held now: CRCT, TRLV, WOR
Marked down in a dated panic, never marked back up. The names the tape marked down hard in a mechanically identified stress window and has never marked back up, whose underwriting did not change while the quote did. In a process a buyer prices the business; the public market in a stress month prices the exposure.
Held now: FICO, ADBE, FTDR, CRUS
Sold like a cyclical, earns like an annuity. A business repeatedly sold harder than the median in mechanically identified stress months, whose return on invested capital has barely moved across four fiscal years, is filed under the wrong heading: paid a cyclical’s discount for an income statement with no cycle in it.
Software-like margins with no laboratory. Near-zero R&D beside a 65%+ gross margin is the signature of an asset whose moat is already built and does not need re-buying every year: a data set, a registry, an embedded workflow. The classification system has no drawer for that combination, so the business gets filed as services or industrials and handed that sector’s multiple.
The best operator where the market says don’t build. Enterprise value below invested capital is the market saying a dollar committed here is worth less than a dollar, which is exactly the signal that stops new capacity being financed and starts existing capacity being retired. The quote gets read as a verdict on the company when it is a verdict on the industry’s appetite for capital.
Held now: INMD
Profitable for two years, still filed under losses. The market sorts companies into profitable and unprofitable from the reported earnings line and re-sorts them only after years of proof. A business with two consecutive positive fiscal years after two negative ones has changed category while its price still carries the old heading.
0-6 qualifying total, 0-2 entrants in a typical week; most weeks empty
Returned nothing this run. Several screens are built to be silent most weeks.
The names that did not fall when everything else did. Every other screen asks what a company is: its margins, its multiple, its capital base. This one asks how it behaved. It finds the market’s worst stretch mechanically from the data, then keeps the businesses that held up through it. Silent unless there has been a recent stress window.
Gate: Severity w1. worst 20-session universe median within the last 60 sessions is -5.38%, milder than -8%: no RECENT stress to decouple from. An older, deeper window may exist inside the 252-session lookback and is deliberately not used — W1 is recency-first by design
Not a picker, a veto that runs against every candidate from every other screen except the deletion basket. Statistically cheap, revenue going backwards and a negative six-month move at the same time gets thrown out, whatever the other screens liked about it. It would have caught most of what went wrong on the list I kept a year earlier, which is why it runs on everything.
Capital pouring in at returns below the incumbent base. When a firm is adding capacity at returns materially below what its existing capital earns, the industry’s returns are being competed away in real time, including its own. This suppresses a name from the week’s pitches rather than surfacing one.