The 25-Year Hold Thesis

Filter 1 — Survival First

A business that does not survive cannot compound. Survival analysis comes before any assessment of upside.

The criteria: debt must be manageable at stressed operating conditions with no fragile maturity wall. No single customer, regulator, payer, or counterparty should account for more than 30% of revenue. No clear substrate-change risk from a better alternative already at scale. Governance must be clean — no restatements, no undisclosed related-party transactions. The business should be able to withstand a 30% unit-volume decline over two years without permanent balance-sheet impairment.

Candidates that do not pass survival analysis do not advance to deeper work. The analysis is not a formality.

Filter 2 — Name the Mechanism

"Quality" is not enough. High ROIC is an output, not an explanation. The process requires a named structural mechanism: a specific, identifiable source of competitive defense that a well-capitalized competitor could not replicate within five years.

The recognized mechanism types: operational switching cost — the cost to the customer of changing suppliers is embedded in their processes, not just their contracts. Regulatory enshrinement — the business occupies a position that regulation makes difficult or impossible to replicate. Network effect — the service becomes more valuable as adoption grows, creating a compounding user-side barrier. Physical density — local network depth that requires years of capital deployment to build. Keystone data asset — a proprietary dataset that improves with use and cannot be reconstructed from public sources. Contractual tollbooth — recurring revenue from a position in a value chain that is difficult to bypass. Workflow integration — deep embedding in customer operations that makes the cost of switching visible and high. Benchmark embedding — the service has become the standard reference that competing products must be evaluated against.

Silent substitutability is not a moat. If a competitor can replicate the customer experience without replicating the structural position, the mechanism has not been named.

Filter 3 — Reinvestment Runway

An excellent business with no place to deploy capital is a cash cow, not a 25-year compounder. The process requires evidence of a plausible 15-plus-year reinvestment runway where incremental capital can earn returns above the cost of capital.

Incremental returns on incremental capital — ROIIC — matter more than legacy ROIC. A business with 30% ROIC on its existing asset base but declining returns on new investment is consuming, not building, its competitive position. The unit-growth model is required: how many more units can be added, at what incremental return, over what timeframe.

Mature capital return — buybacks and dividends — should be additive to the thesis, not a substitute for missing reinvestment. A business that returns capital because it has no better use for it is different from a business that returns capital after exhausting its highest-return reinvestment opportunities.

Filter 4 — Underwriting

Quality does not cancel price. Every candidate must clear the valuation gate on two independent tests: the base-case 25-year IRR must be at least 8% on EPS per share, and the base-case 25-year IRR must be at least 8% on free cash flow per share. Both tests must pass independently. Strong earnings numbers do not compensate for weak free-cash-flow numbers.

The underwriting must include bear, base, and bull cases that are explicitly different from each other. The terminal multiple must decline from the current multiple — assuming today's multiple persists for 25 years is not a model, it is an assumption dressed as analysis. The thesis must include an ex-ante falsification condition: a specific, observable event that would indicate the thesis is broken.

The No-Pick Rule

The quarterly cadence is a commitment device, not a quota. If no candidate clears all four filters and the underwriting bar, the correct name is no name. The ledger records the skipped quarter and the reason.

A skipped quarter is not a failure. It is evidence that the process is not being forced.

The Good Business / Bad Entry Rule

A business can pass the quality work and fail the price work. That is not a near miss. It is a watchlist item with a trigger price. Naming it at a price that does not clear the IRR hurdle would make the underwriting meaningless.

Good businesses at the wrong price appear in the subscriber short list with the trigger price noted. They do not appear in the public ledger.

The Insufficient Underwriting Rule

Missing evidence is not rounded up into conviction. If primary-source evidence is absent on a material point — customer retention data at scale, unit economics under stress, management track record through a full cycle — the candidate cannot be named regardless of how compelling the narrative is.

The narrative is a hypothesis. The evidence is the test. A compelling hypothesis without the evidence to support it is a watchlist item, not a name.