Three of this week's five are priced as though a pause were permanent: a European building trough, a half-year of cash swallowed by inventory, a wobble in Australian buyer confidence. In each case the money has not turned up yet, and the price is treating that as the business. The two SubprimePicks are the mirror image — companies we admire, where the price already banks something that has not happened either.
“Priced as though Europe's building trough were permanent. It is a rate cycle, and rate cycles turn.”
Saint-Gobain makes the dull parts of a building: mineral-wool insulation (Isover, CertainTeed), plasterboard (Placo, Gyproc), Weber mortars, flat glass, and the asphalt shingles on American roofs. It sold €46.5bn of those in 2025 at a 10.7% operating margin — up from 7.3% in 2020, after years of shifting the mix toward higher-margin specified products like construction chemicals and performance insulation. What protects those earnings is the width of the range. Rockwool and Kingspan compete in insulation, Knauf in plasterboard, James Hardie in American siding; each can go at one line, but none is positioned to attack the whole portfolio at once. And Europe's building-energy rules, which oblige member states to keep upgrading the energy ratings of existing buildings, write insulation and glazing into decades of mandated retrofit work.
Revenue fell roughly 9% from the 2022 peak of €51.2bn to €46.5bn in 2025 — mortgage rates choking off European new-build starts and deferring renovation, not share lost to a rival. Buyers today get about a 9% free-cash-flow yield on the €3.5bn the business generated in 2025, the sort of return usually reserved for a company in trouble. This one isn't: net debt is 1.1 times EBITDA, roughly a year's earnings before interest, tax and write-downs, and interest is covered 6.5 times. So the price is assuming the trough is the permanent condition of European building. Rate cuts have been in place across most of its markets since 2024 and are beginning to lift renovation enquiries, and the share count has come down steadily for twelve years, from 560m in 2014 to 494m in 2025, with the dividend up from €1.24 to €2.30. A rate cycle is not a permanent condition, and the price has confused the two.
What would end the argument: If European residential construction stays depressed through 2027 on affordability, operating income could slip from €4.95bn in 2025 toward €4.2–4.4bn, a 10–15% fall, with free cash flow declining toward approximately €2.9–3.1bn and the earnings recovery pushed out 18–24 months. The asbestos and PFOA litigation is provisioned at approximately €420–422m at mid-2025, but the eventual cash cost is uncertain. The …read the full thesis →
“Half the cost of a 155mm shell is powder and explosive. CSG is buying those plants.”
CSG makes the things that go bang and the vehicles that carry them: 105mm to 155mm artillery rounds, tank ammunition, mortar bombs, Pandur 8×8 armoured vehicles and military radars. Defence Systems was about 81% of revenue in the first half of 2026, €2,620m, and the armoured-vehicle line grew 96% year on year to €445m. The customers are NATO governments refilling stockpiles emptied by the war in Ukraine, and €17bn of the work is already contracted — about 2.4 times a year's revenue. The edge is chemistry. Propellant and explosive are 50–60% of the cost of a 155mm round, and CSG is buying its own supply: nitrocellulose capacity in Germany, TNT through a Greek joint venture, nitroglycerin in Slovakia from 2028. Management says that will cut propellant costs by up to 50% and explosive costs by up to 70% once the plants run, which both widens the margin on every shell and lets CSG bid under rivals in tenders and still make money.
The pricing power is already in the accounts: group operating margins rose from 21.4% in FY2023 to 23.6% in FY2025 through sharply rising input costs, and Defence Systems earned 28.8% in the first half of 2026 while raw-material costs rose 24% year on year. What the price will not credit is cash, and not unfairly, because the cash isn't there yet: working capital stood at €2,895m in June 2026, 40.1% of the last twelve months' revenue, and pre-tax free cash flow was −€411m in the half. A buyer today pays a price implying roughly a 4% free-cash-flow yield, and that rests on management releasing the trapped cash — it guides working capital below 20% of revenue by the end of 2026. It has happened before: FY2024 produced €739m of free cash flow when the inventory cycle converted. Against what this business is likely to earn over the next decade, the shares are genuinely cheap.
What would end the argument: Ukraine was 17% of group revenue in the first half of 2026, around €558m, and a ceasefire or peace settlement would expose an estimated €900–1,100m of annualised revenue — roughly 12–15% of projected FY2026 revenue — to repricing or deferral, concentrated in the highest-margin ammunition line. The working-capital release is the other one: if production delays or slow customer payment prevent it, debt rises toward 2× …read the full thesis →
“The wobble is in confidence. The $830 million sales book is contracted out to FY2029.”
Cedar Woods buys land in Western Australia, Queensland, South Australia and Victoria, gets it approved, subdivides or builds on it, and sells the lots, houses, apartments and a little office space to owner-occupiers and investors. Nothing reaches the accounts until a sale settles. In the year to June 2026 that meant 1,326 settled lots, homes and offices, $502.4 million of revenue, up 7.8%, and net profit of $65.6 million, up 36%. There is no moat here and it is worth saying so plainly: the company owns no land a rival could not buy and no brand a buyer pays extra for. What it has is skill at picking sites and getting them through planning, built up since 1987, and that shows in a 20.6% operating margin against a subsector median of 11.1%, and an 11.75% return on invested capital against a 2.0% median.
Those margins came from land bought years earlier settling at today's prices, walking the operating margin up from 14.8% in FY2021. The price today is discounting a wobble management named itself: buyer momentum "moderated" late in FY2026 as interest rates rose, living costs bit and a federal tax-change announcement made buyers pause. But the wobble is about people signing new contracts, and $830 million of contracts are already signed, worth about 1.65 times a full year of FY2026 revenue and settling out to FY2029, against an Australian housing shortage nobody disputes. Management is targeting 15% profit growth in FY2027 off that book. The company sits in net cash, still buying land in WA, Victoria and Queensland during FY2026 into a pipeline of more than 9,600 dwellings, lots and offices, and lifted the fully franked dividend 34% to 39.0 cents a share alongside that spending rather than instead of it. The softening is an argument about when the money arrives; the price has made it an argument about the business.
What would end the argument: If the buyer softening persists into FY2027, both the 15% profit-growth target and the pace at which the $830 million presale book converts into cash are at risk, because buyers can still walk away or fail financing between signing and settlement. Cedar Woods builds through third-party fixed-price contractors and flags contractor financial viability as an elevated risk after a period of …read the full thesis →
“Share count up 53% since 2014, no buybacks since 2017: the enterprise compounds, the share lags.”
Axon sells the kit an American police officer carries and the place the footage ends up. TASER 10 stun weapons, body cameras, in-car cameras and counter-drone systems brought in $507 million in the June 2026 quarter, 56% of revenue, growing 35% year on year. The other 44% is software — Axon Evidence, which stores the footage, and Draft One, which drafts officers' reports — at $398 million in the quarter on a 75.1% adjusted gross margin. The moat is genuine, and it is switching costs rather than brand. Axon Evidence holds years of body-camera video, use-of-force records and chain-of-custody documentation that prosecutors and courts depend on, and moving that archive to a rival mid-contract while keeping it admissible in court is effectively prohibitive. Agencies do not leave, they expand: existing customers spent 126% of what they had spent a year earlier in the June 2026 quarter, and only about 30% of the installed hardware base has taken the premium software plans.
None of that is the objection. At today's price a buyer is paying for every lever to fire at once and keep firing — new agencies signing, existing ones upgrading, artificial-intelligence products selling — with no margin disappointment anywhere. The per-share arithmetic runs the other way. Stock-based compensation, shares handed to staff instead of cash, was $634 million in FY2025, 23% of revenue, guided at $590–620 million for FY2026, and the share count grew 53% between 2014 and June 2026 with no buybacks since 2017. Free cash flow over the last twelve months was roughly $133 million — $266 million of operating cash flow less $133 million of capital spending — against the $710 million of adjusted EBITDA the company reports, and that stock-based pay is the gap between the two. Net debt was $1,115 million in June 2026 against a statutory operating loss. Exceptional franchise; a price sitting far ahead of the evidence.
What would make us wrong: The way this goes wrong for us is that the upsell simply keeps working: with only about 30% of the installed base on premium plans and the AI Era Plan booking approximately $750 million in its first full year, several more years of retention like the 126% recorded in the June 2026 quarter would grow the business into its price. Stock-based compensation falling durably below 15% of revenue would partly restore the …read the full thesis →
“A $1.4bn step-up in next year's bookings rests on one game arriving in November.”
Take-Two makes video games. Roughly 60% of revenue comes from console and PC — Grand Theft Auto and Red Dead Redemption from Rockstar, the annual NBA 2K and WWE 2K — and 40% from the Zynga portfolio of free phone games: Toon Blast, Match Factory!, Empires & Puzzles. The money mostly arrives after the sale rather than at it. In the three months to June 2026, 84% of bookings came from players buying virtual currency and items inside games they already own. Grand Theft Auto is why that works. GTA V came out in 2013 and was still one of the top contributors to bookings in the fourth quarter of FY2026, through the GTA Online multiplayer world where players buy in-game cash. One open-world release paying out for over a decade is something no competitor has repeated, and it is what keeps the earnings coming between launches.
The objection is what November 2026 has to carry. Management guides FY2027 net bookings of $8.0–8.2bn against $6.72bn in FY2026, and almost the whole $1.4bn step-up comes from Grand Theft Auto VI; that one launch is what a buyer at today's price is paying for. The balance sheet leaves no room for the date to move. Gross debt is $2.5bn against $624m of operating cash flow over the last twelve months, interest cover is negative at −0.73 times, meaning interest is paid out of cash reserves rather than earnings, and $1.4bn falls due in FY2028. A six-month slip to spring 2027 would push $1.3–1.5bn of the guided step-up into that same year, just as lenders price the refinancing. Behind it sits the $9.5bn Zynga purchase: $5.9bn of goodwill impaired, shareholders diluted 61% from 115m to 185m shares since FY2022, and return on invested capital of −1.7% across four straight years of operating losses, against Electronic Arts' 14.9% on similar revenue. The risk is not the game. It is the calendar.
What would make us wrong: If Grand Theft Auto VI ships on 19 November 2026 to the reception the franchise usually gets — past titles have scored 97–98 on Metacritic — and GTA VI Online monetises the way its predecessor did, the FY2027 guidance lands and the deleveraging runway comes with it. The mobile side would help: Zynga contributed $739m of net revenue in the three months to June …read the full thesis →
Our approach was built with equity research analysts from leading investment banks: fundamental analysis on a company's filings written into sharp investment reports, rolled out across our database of 40,679 companies and made accessible via a screener.
The companies above (and their reports) were taken from that screener, and it's the same one available for free on our website.
Browse all companies, their research and our target prices.
Every one of your positions analysed like above.