The Latticework Monday Morning Briefing is our weekly “Guide to the Markets” for long term-oriented investors. It is sent on a separate mailing list (complimentary to members). If you do not wish to receive it, opt out here.
We have a great agenda for Latticework 2026 (November 10-11). The summit features fireside chats, panel discussions, idea pitches, and a group dinner. Capacity is strictly limited. Secure your seat.
Listen to the Latticework Monday Morning Call:
Ideas from our member newsletter
We highlight a few stock write-ups from the latest Weekly Inspiration:
Close Brothers Group (London: CBG) is a UK specialist lender whose restructuring is running ahead of plan: FY26 adjusted operating profit of £120.3 million beat consensus by 8.4%, annualised cost savings came in at £36 million against an original £20 million target, management now promises more than £60 million by the end of FY27, and net loan growth turned from a 1% decline in 1H26 to 4% in 2H26. The beat came from costs (£430.9 million against £448 million expected) rather than margin; NIM was 6.9% against 7.0% expected, and credit impairments of £91.7 million ran above the £83 million consensus as Property-book provisions rose. Reported RoTE was 5.5% against 4.4% expected, with management guiding to double digits by FY28, and CET1 of 14.1% leaves about 380bps of headroom over the minimum. SeaPoint Insights reads the 15.5% share-price jump on results day as relief that no further motor finance provision was taken in 4Q, leaving the total at £318.5 million. That provision remains the main risk: the FCA redress scheme faces legal challenges, no FY26 dividend was declared because of it, and Viceroy Research’s March short report argued the top-up could run from £272 million to £932 million, which the company strongly disputes. The stock trades at 0.52x TNAV against consensus FY28 RoTE of 9.0%, an implied cost of equity of 17.4% versus 15.8-17.7% for OSB, Paragon and Shawbrook, so the discount has narrowed but not closed. The author also sees the group as a plausible takeover candidate once the redress costs and the Aldermore sale outcome are known.
Green Landscaping Group (Stockholm: GREEN) is a decentralised owner of 59 local landscaping, grounds-maintenance and winter-service contractors across Sweden, Norway, Germany, Finland, Lithuania and Switzerland, about 60% of whose revenue comes from public-sector framework contracts of three to five years with extensions to eight, and it has grown net sales from SEK 3,139 million in 2021 to a SEK 6,632 million run-rate at Q2 2026. Operators in smaller municipalities often hold quasi-monopoly status, route density lowers drive time per job, and ISO, bonding and union requirements keep small independents out of the larger tenders. Deals were done at 4.5-6.5x EV/EBIT with 20-40% of consideration as earn-outs tied to EBIT hurdles. The problem is debt: leverage reached 3.4x pro-forma EBITDA against a 2.5x internal ceiling, net debt including earn-outs, leases and minorities is SEK 2,859 million, acquisitions are paused, Norway posted negative EBIT in Q1 and a 4.1% margin in Q2, and the CEO of 11 years has left with an interim in place. The Hermit values the equity at SEK 41.59 a share in a base case (9% cost of capital, margins recovering to ~6.5%, 152% upside) and SEK 6.40 in a bear case where margins stay near 5%, with a path to zero if debt rollovers fail; by that arithmetic the price implies ~82% odds of the bear or zero outcomes. The Continental European unit earns 15-22% margins and Lithuanian renewals came in above segment averages.
OMS Energy Technologies (Nasdaq: OMSE) makes the conductor connectors, surface wellheads and premium-threaded casing that hold shallow-water and onshore wells together in the Middle East and Southeast Asia, holds $154.3 million of cash and restricted cash with no bank debt against a market cap of about $185 million, and generated $52.5 million of adjusted free cash flow on $155.9 million of FY26 revenue with capex of just $1.64 million. Specialty connectors were 61.6% of FY26 revenue, 95-99% of it from the Dammam plant to Saudi Aramco under a 10-year agreement running to 2033; premium threading under licences from VAM, Tenaris, JFE, Hunting and NOV Grant Prideco is the recurring, highest-margin line at 28.3% to 40%+ gross margin. Revenue fell 23.4% from $203.6 million in FY25 as Aramco call-offs normalised after an overlap between the old and new contracts, and the 12-month backlog dropped from $102.0 million to $60.7 million, which management attributes to timing between tranches rather than lost tenders; an $11.0 million Aramco call-off booked in March 2026 was the first replenishment. The Hermit, who spoke with senior management, counts the risks as Aramco at 57% of sales, zero minimum-volume guarantees, and a disclosed material weakness in internal controls. With cash at ~$3.60 a share and the stock near book value of $4.03, the author wants a dividend or buyback, puts base-case fair value at $9.72 (13% discount rate, revenue recovering toward $200 million at 21.5% operating margins, 126% upside), and a bear case at $3.50.
Braskem (NYSE: BAK) is a petrochemical producer earning its way through a restructuring: Q2 EBITDA was roughly $1.0 billion on the Hormuz-driven spike in polyethylene spreads, the Strait has stayed largely disrupted into Q3, and the 90-day creditor window that opened in late August closes around November 22. At the last count only about 40% of creditors had signed the plan. Braskem has sent terms that now include a debt haircut, absent from the original out-of-court proposal; creditors have reportedly asked the controlling shareholders for around $3 billion of fresh capital, and offshore bondholders again want a capital backstop from Petrobras, which says it expects a deal, will not raise its stake, but will “increase its support.” A court has also blocked transfers to the Mexican unit Idesa while talks run. Coherence Research sees three broad outcomes: a negotiated deal with limited dilution, which the stock is nowhere near pricing in; a deal with a meaningful equity component, where existing holders survive but are diluted; and a breakdown into judicial recovery, which Petrobras’ comments suggest nobody at the table wants. A fourth, less likely path is that strong spreads let Braskem build enough cash to weaken the case for a heavy restructuring. PE-naphtha spreads have averaged about $600/t this quarter against ~$770/t in Q2, close to double the ~$316/t consultants forecast, and the author estimates Q3 EBITDA of $550-700 million.
uniQure (Nasdaq: QURE) has in AMT-130 a Huntington’s disease gene therapy whose 36-month analysis, the timepoint the FDA said could serve as the primary basis of an accelerated-approval BLA, showed a statistically significant 75% slowing of cUHDRS decline (p=0.003) and 60% slowing of Total Functional Capacity decline (p=0.033), and three more high-dose patients reaching 36 months strengthened those to 80% and 67%. The stock sold off on the four-year data, where cUHDRS slowing fell to 44% (p=0.144) among twelve high-dose patients. Napkin Bios argues that number answers the wrong question: cUHDRS at three years is an intermediate clinical endpoint whose purpose is to predict later functional benefit, and the functional measure itself, TFC, still showed 61% slowing at four years (nominal p=0.008), against an updated ENROLL-HD natural-history comparator that by then appeared populated by slower progressors. The cUHDRS softening was driven by the SDMT cognitive test and the clinician-rated Total Motor Score, with one participant’s TMS reading having a material effect in a twelve-patient cohort. Dose response is the other underweighted fact: at 48 months high-dose completers had declined about 0.9 cUHDRS points against 1.9 for low-dose, a separation no flaw in the external control can manufacture. CSF NfL does not line up neatly (about 4% above baseline on high dose, 8% below on low dose), which the author acknowledges. The next test is the Q-Motor presentation at the EHDN congress plus imaging; convergence with TFC would make the four-year composite hard to read as lost efficacy.
Liquidia (Nasdaq: LQDA) sells Yutrepia for pulmonary arterial hypertension (PAH) and PH-ILD, the launch more than 15 months ago has produced roughly half of revenue from each indication, and the court ruling that went to United Therapeutics on two of six claims of the ’327 patent applies only to PH-ILD, so the PAH half is untouched whatever the remedy. The remedy is not yet set: Judge Andrews gave the parties a week to submit an agreed form of final judgment, Liquidia will ask the FDA to remove PH-ILD from the label so PAH treatment continues uninterrupted, and United will likely seek a permanent injunction and damages. Arquitos Capital argues a royalty is the rational outcome under the four-factor eBay test: United is still in the market, the judge earlier refused a preliminary injunction on the ground that money damages suffice, Tyvaso pricing has not eroded, and a documented share of patients cannot use United’s devices. A settlement is unlikely given United’s litigation record, so the judge may decide, which could take weeks to six months. The author had Liquidia on track for $1.4 billion of 2027 revenue at net margins near 50%; with PH-ILD removed, $700 million of revenue and $350 million of cash flow support a stock value of $44-55, and a 30% royalty on continued PH-ILD sales lifts that to $75-94. The twice-daily L606 formulation, due in 2030 or earlier if marketed to PAH only, could be worth at least the current market cap. Off-label PH-ILD prescribing has historically been covered by insurers. Arquitos calls today’s price a compelling entry point.
As always, the above theses reflect the linked authors’ views (available here), not Latticework recommendations.
Articles worth your time
Who Gets Paid When AI Does the Shopping?, by Alex Immerman and Santiago Rodriguez, argues that AI shopping agents threaten the advertising and commission profits of marketplaces far more than their fulfilment businesses, and that Amazon’s 2025 advertising revenue of $69 billion, set against $34 billion of operating income excluding AWS, shows how much is at stake. At an assumed 70% contribution margin, advertising contributed more than Amazon’s reported operating income, so a layer that intermediates discovery attacks the most profitable line even if Amazon still packs the box. The authors offer two questions for any platform: how much incremental demand can an agent bring, and how much of the platform’s profit depends on controlling discovery and the customer relationship. DoorDash sits on the wrong side of the second: an agent can bring the restaurant a customer without routing them through the marketplace, pressuring commissions and ads even when DoorDash still handles the order. Shopify, Toast and Square sit on the right side, since they earn software and payments revenue rather than ads, and giving agents access to merchants’ inventory and checkout simply opens another channel. Instacart, with no comparable ad business to protect, has leaned in and reports that orders placed through its AI assistant include more items and exceed its typical basket value. The authors close on the open question: whether agents create new purchasing volume or merely redirect existing orders decides whose economics survive.
Before you fade Muse, use it for 72 hours, by Andrew Walker, tells value investors who dismiss consumer AI agents to download Meta’s Muse, give it email access and use it for three days before forming a view, because the skeptics he meets have not actually tried it. His own three days supply the evidence. Buying a new phone and AirPods, Muse found Amazon cheaper than Apple and saved ~$50, then found a better trade-in worth ~$200 more. He photographed a pair of pants; it identified the brand and found cheaper alternatives. Asked for undershirts, it searched retailers, loaded a JCPenney cart and, once connected to his credit card, checked out on its own. It compared Knicks ticket prices, built eight itineraries for a boys’ weekend in seconds, replaced his MyFitnessPal subscription with photo-based calorie logging and did the game reviews he used to pay Chess.com for. He then follows the money. If an agent price-compares every ticket resale fee, what happens to the sums Ticketmaster and StubHub pay Google to top a search for Knicks tickets? Retail brand loyalty erodes when the agent, not the shopper, picks the store; hotel booking is not far off, and the online travel agencies are absent from that scenario; small subscriptions lose their reason to exist when the agent does the job free. He is long Meta, in a small position, and expects that what the bleeding edge does with agents today is how everyone will use them in about 18 months or sooner.
Conceptual Gaps in How Investors Think About AI, by Mallika Paulraj, names five ways investors’ mental model of AI lags the technology, among them that enterprise agentic workloads went from near zero in summer 2025 to 64% of tokens used by summer 2026. The first gap is convergence: Gemini, GPT and Claude now perform within a narrow band, so the edge shifts to reliability and security, and capability can be copied through distillation. The second is scope: alongside interactive AI (a human asks, the model answers) and agentic AI (a human delegates an outcome), physical AI in vehicles, robots, drones and biomedicine is built on world models; she cites AMD paying $8 billion for Fei-Fei Li’s World Labs, valued at $3 billion two months earlier. The third is price: intelligence has fallen in cost about as far in five years as electricity did in 80 or DNA sequencing (a million-fold) in 20, so yesterday’s frontier capability erodes fast and Nvidia gives models away to sell chips. The fourth is geography: training stays centralised while inference moves to devices, local servers and regional and colocation facilities depending on the workload. The fifth is the one for allocators: AI demand is not the same as AI investment returns. A data-centre investment rests on the probability the facility actually gets built (permits, contractors, delays), the probability customers use that specific site, the asset economics of construction cost, operating expense and competitive longevity, and a capital structure whose debt terms decide what equity keeps. Being right about demand, she writes, is not the same as being right about the investment.
The Problem of Goliath, by Devendra Bambardekar, walks the reader through sixteen years of an Indian full-service brokerage’s decisions, from 2010 to 2026, to show why a well-run incumbent loses to a discount broker while making rational choices at every step. The thought experiment puts you in the incumbent’s chair. In 2010 a startup launches a ₹20 flat fee per trade against your percentage-based commissions; you ignore it. By 2014 it has 35,000 accounts, concentrated among active traders you are happy to lose, and matching its price would, in Bambardekar’s words, decimate the profit and loss statement overnight; besides, you cannot fire the research teams or dismantle the branch network, and as part of a listed group you cannot justify the margin cut to analysts. In 2015 the startup drops brokerage on delivery trades to zero, and its accounts grow from 70,000 in December 2015 to 200,000 in December 2016. The pandemic brings a wave of new retail investors and several more discount brokers, and by 2026 some of them out-earn you. The startup won by rewriting the rules: open-source technology with minimal infrastructure, word-of-mouth acquisition at near-zero cost, and fully digital Aadhaar-based onboarding. He frames this with Christensen: giants fail not through stupidity but because their financial models tell them to run away from low-margin business, and the remedy is a separate autonomous organisation with its own incentives, cost structure and customers, insulated from the legacy priorities. One peer did overhaul itself into a major discount player; the rest did not.
A €35 Million Loss — And What It Taught Us, by Reimar Scholz, dissects how Russian shares that cost about 2% of the portfolio, and had grown to roughly 5% through gains, became worthless almost overnight, a loss of about €35 million. Scholz lists four mistakes: underestimating the probability of military action, underestimating the speed and severity of sanctions, assuming political leaders act in their economic self-interest, and confusing “irrational” with “impossible.” The mechanism is what makes the case instructive. Liquidity vanished, market access disappeared and depositary receipts were suspended within days, so shares that looked extraordinarily cheap on conventional measures offered no protection; some risks are binary, and a stock can go from tradable to inaccessible with no intermediate price at which to sell. Two sizing lessons follow. Position size should be measured against current portfolio value, not historical cost, so a 2% purchase that compounds into 5% is a 5% risk; and he now spreads emerging-market bets more widely and takes some profits along the way, a deliberate departure from concentrating in the best ideas and letting winners run. Before any valuation work on a foreign holding he now asks whether foreign investors can enforce ownership rights, whether capital can leave freely, whether the legal system is predictable, whether sanctions could render the shares unsellable, whether the expected return justifies risks that cannot be diversified, and whether the position is sized so the firm survives a total loss. The question “can we survive being completely wrong?” comes before the model.
New in this week’s deck
We add eight charts on the economics of the AI build-out and the market around it, drawn mainly from Andreessen Horowitz’s research with underlying data from BofA, Vanguard, JPMorgan Asset Management, Citi, Citadel and Goldman Sachs. They cover hyperscaler forward free cash flow turning into semiconductor free cash flow, hyperscaler capex heading past $1 trillion a year, hyperscaler borrowing for the build-out, S&P 500 net margins heading for a record 17.5%, a record 93.6% of S&P 500 companies meeting or beating EPS estimates, the ~75% of public software companies that are profitable against the ~30% growing 20% or more, leveraged ETF assets near $200 billion with 61% in tech and semis, and asset-heavy stocks erasing a decade of underperformance against asset-light ones.
The deck
The full Monday Morning Briefing is available to members. It spans our weekly scoreboard, idea-generation screens (this week including quality names near their 52-week lows, micro-cap “tiny titans,” recent spin-offs, activist campaigns, superinvestor holdings, owner-operators, share-count shrinkers, and the broad US valuation screens), market valuation and positioning, macro and fixed income, and equity valuation screens across international markets (Canada, the UK, Germany, Australia, Japan, Korea, India and Sweden, alongside a cross-country P/E table and European REIT discounts to net tangible assets). We welcome your feedback.
Feedback on the Briefing
“Most of what I monitor, all in one place. Great value add.” —Brad Lummis
“Loving these Monday briefings!” —Jon Bartel
“Tightly presented and easy to digest. I just spent 20 minutes going through it, and it’s helped to level set me for the week ahead.” —Michael Loftis
“A great piece and thoughtfully assembled.” —Brian Wolf
“I have never seen more valuable content in one place.” —Bill Coleman
“Worth its weight in gold.” —Shree Viswanathan
Featured Events
Latticework 2026 (AGENDA) (REGISTER), Chicago (Nov. 10-11, 2026)
Ideaweek 2027 (FULLY BOOKED), St. Moritz (Feb. 1-4, 2027)
The Zurich Project 2027 (COMING SOON) (Jun. 1-3, 2027)
