Buffett's line about brilliant managers and businesses with bad economics gets used to end arguments rather than settle them. We ran it on Baseline to test it. Wetherspoon scores 4 for management. Five of its other six business quality components score 2 or less, including a 1 for barriers to switching. Tim Martin has beaten the industry on like-for-like sales for 42 consecutive months, and incremental ROIC still fell from 66.2% in FY23 to 5.8% in FY25. Across the universe, 80 companies pair good management with a bad end market, and 55% of them still reach a business quality score of 3 or better. The reverse, bad management in a good end market, does better at 67%. So Buffett is right, by twelve percentage points.

Lockheed Martin makes the only exportable fifth-generation fighter in the West. It scores 3 out of 5 on pricing power. TransDigm makes valves. It scores 5. Both are sole-source on parts that cannot fail. The difference is who they invoice. Lockheed hands the Pentagon certified cost data before the contract is signed, and the negotiated margin sits inside a band that federal cost-accounting rules define. TransDigm sells to several hundred airlines and maintenance shops, none of whom has any right to see the cost base. We score mission criticality and pricing power as separate fields on separate evidence. Of the 246 companies in our coverage scoring 5 on criticality, 79 score 3 or below on price. 26 of those 79 are healthcare, which is about 10% of what we cover. Ten are defence primes. Six make automotive parts. Every one of them sells to a buyer that either is a government or negotiates like one. Ten companies out of 3,676 hold a 5 on both. The caveat: this could be telling you how our prompts read healthcare and defence rather than how healthcare and defence actually work. Three unrelated industries reaching the same explanation is what makes me think it isn't, but it's one universe scored by one prompt library.

Delighted to see Baseline featured on Behind the Balance Sheet, with Stephen Clapham giving an honest assessment of what the platform can and cannot do. It published the same weekend our pay gates went live, so Baseline also took its first paying subscribers. Worth a read if you invest on fundamentals and have wondered what AI is actually useful for in that process. behindthebalancesheet.substa…

Stock of the week. Abercrombie & Fitch closed up 36% on Wednesday after a Q2 beat and a raised full-year outlook. Our headline score on the name has been a 4/5 since 27 March, written with the stock down more than half from its high. The rationale on file that day: "This is not a deteriorating business, it is a normalizing one, and the market appears to be over-discounting that normalization." Before anyone calls that a moat story, the card underneath says the opposite. Mission Criticality 2. Barriers to Entry 2. Barriers to Switching 1, the lowest score the rubric gives. The 4 rests on Management Quality 4, Unit Economics 4, Benchmarking 4. A well-run retailer in a structurally unprotected sector, priced in March as a deteriorating one. The print deserves the same honesty. Comps were flat. Roughly $1.75 of the $4.17 quarterly EPS came from refunds of IEEPA tariffs already paid, and the raised guidance leans on the same relief. Underlying EPS grew about 4%. And the March record got one thing wrong. It leaned on Hollister momentum against Abercrombie brand softness. The June quarter ran the other way: Abercrombie comps up 4, Hollister down 3. Every score carries its generation date so you can see exactly what was known when. Same rubric, same week, across the peers: URBN 4, GAP 4, AEO 3. All four score 2 or below on switching costs. In this sector the separation lives in the execution fields. Baseline surfaces and screens candidates. It does not predict returns.

Screen of the week. Three Baseline fields at maximum, all at once, across 3,700 scored global companies: Mission criticality 5/5 Barriers to entry 5/5 AI resilience 5/5 19 names clear all three. 0.5% of the universe. Almost none of them are software companies. ASML, TSMC, Shin-Etsu. GE Aerospace, Safran, TransDigm, BAE, Northrop, Huntington Ingalls, BWXT. Aena, Copenhagen Airports, ENAV, Norfolk Southern. Constellation Energy. Japan Exchange Group. CK Infrastructure. VeriSign. And Perimeter Solutions, which makes the retardant the US Forest Service drops out of aircraft. For Constellation Energy. 22.1 GW of nuclear against 6.3 GW at the next merchant operator, running at a 94.7% capacity factor. Microsoft and Meta both signed twenty-year power agreements rather than build their own generation. That is the most capitalised industry on earth conceding it cannot replicate the asset. Physically indispensable, capital-heavy businesses tend to score well on all three, so the screen finds a category rather than an anomaly. What is scarce is the triple maximum.

Using Baseline, one thing I've found useful is finding non-obvious names on an investment theme. For instance, Abercrombie has low customer lock in, but also high AGI risk (given its dependance on the algo). This wasn't priced, and provided a neat short opportunity.