The KWM Constitution

Why the name

USS Constitution was launched at Boston in 1797, named by George Washington for the founding document, and is the oldest commissioned warship afloat in the world. Her hull was live oak more than two feet thick. When British shot failed to penetrate it off Nova Scotia in 1812, a sailor called out that her sides were made of iron. She won three engagements against the Royal Navy, took thirty-three vessels across fifty-seven years of service, and was never defeated.

Two things about her are worth borrowing. She was built from deliberately chosen material rather than convenient material — and she is governed by, and named for, a written document.

What the portfolio is

A thirty-position equity portfolio assembled and monitored by a system of language-model agents working under a written doctrine. The doctrine is not the model's opinion: it was distilled from forty-nine years of Berkshire Hathaway shareholder letters, 1977 through 2025, and it is a versioned document the system reads on every decision and may never amend on its own. Specialist agents surface candidates each quarter and a dedicated Skeptic attacks each one before it is considered. A single Steward then weighs the case and writes a verdict, citing the principle and the passage it rests on. Nothing is acted on until Scott Zodin agrees with it. Positions are held up to 8%, and the Skeptic goes on monitoring every holding against the exit conditions recorded at purchase.

• Built of chosen material The doctrine is written, cited and versioned. Every verdict names the principle it rests on, so the reasoning can be reviewed months later.
• Built to take fire No position enters without a pre-mortem describing how it becomes a permanent loss, and a written statement of what would prove the thesis wrong.
• Built to stay in service About four replacements a year — an average holding period near seven years. Most weeks the correct output of this system is silence.

The doctrine

Ten principles. Every company is assessed against all of them and the assessment is written down — including for the companies that are turned away.

1. P1Franchise, not merely a good businessPricing power that survives a bad manager and a bad year.
2. P2A durable advantage that is wideningAn advantage that is merely wide is one that is being crossed.
3. P3Return on invested capitalIncluding on incremental capital, measured across a full cycle.
4. P4Owner earningsCash the owner may actually take out. EBITDA is not used.
5. P5Capital allocationWhat management does with a dollar it does not need.
6. P6Circle of competenceIf the business cannot be explained simply, it is not assessed.
7. P7Candor of managementHow they write about the year they got wrong.
8. P8Risk as permanent lossNot volatility. The chance the capital does not come back.
9. P9The price paidThe final gate, applied by the principal — never the reason to be interested in the first place.
10. P10Earnings that hold in any weatherRecession, inflation, stagflation. Pricing power, and products bought in every environment.

Price sits at the end of that list deliberately. A cheap price is not a reason to own a business; it is only ever the final question asked about a business already worth owning.

How a company gets in

The agents read filings, screen the universe, and write up the case for and against each name. Every recommendation arrives in writing, argued against the ten principles, and is signed off — or refused — by Scott Zodin personally. Nothing is bought automatically, and no model places a trade.

• 3,828 US-listed companies in the screening universe
• Doctrine screen Every name assessed against the doctrine before price is considered, with written findings
• Skeptic Attacks each survivor before it is considered, and keeps attacking it once owned
• Steward Weighs the case and writes the verdict, citing the principle it rests on
• Price P9 applied last, to what has already cleared the bar
• The principal Scott Zodin signs off, or refuses. The Steward recommends; it does not decide

Construction

Holdings	Up to 30 individual common stocks
Position size	Up to 8%, sized by conviction
Minimum size	$2.5 billion market capitalization
Universe	US-listed common stocks, all sectors, no index membership requirement
Expected turnover	About four replacements a year — an average holding period near seven years
Structure	Individual securities held directly in the client's own account. No fund wrapper.
Custody	Charles Schwab
Cash	A reserve managed by the adviser — see below

When we sell

A falling price is not a reason to sell. It may be a reason to buy more. Only the thesis can trigger a sale, and the case has to be written and argued against the doctrine before it is acted on. Four grounds are admissible:

• G1The thesis is broken — something recorded at purchase as disqualifying has actually happened.
• G2An error in the original judgment is recognized.
• G3A materially better use of the same capital, under the thirty-name cap.
• G4A valuation that can no longer be defended on any reasonable set of assumptions.

When a case is close, the incumbent holding stays. Every thesis is recorded at purchase together with the specific facts that would prove it wrong — written before there is a position to defend, which is the only time such things get written honestly.

Cash

Every account carries a cash reserve, and it does three jobs: it meets distributions and expenses so that we are never forced to sell a good business at a bad moment; it funds new purchases when something finally clears the bar, which tends to happen on the market's schedule rather than ours; and it plays defense when defense is called for.

The doctrine screens businesses and has nothing to say about cash. That side is managed by the adviser directly, at his discretion, and sized to the client's circumstances rather than to a fixed model weight. Held with intent, cash is not a drag on a portfolio — it is what allows the rest of the portfolio to be patient.

Cost and tax

You own the businesses directly, in your own account. There is no fund wrapper, no layered management fee, no distribution charge, and no bundle of underlying products each taking a bite. Tax lots stay visible and controllable: losses can be harvested and gains deferred deliberately, rather than arriving every December on someone else's redemption schedule.

Held as a standalone account, the portfolio carries an advisory fee of 0.65% a year. Held alongside other assets under our standard advisory agreement, Keel Wealth Management's tiered fee schedule applies instead. Nothing else is layered on beyond ordinary custodial and transaction costs. The full fee schedule is set out in our Form ADV Part 2A.

A note on artificial intelligence

AI is the tool here, not the theme. This is not an AI portfolio: there are no thematic bets, no story stocks, and no wager on which model wins. The models do what a very patient analyst would do if you could afford one who never slept and never fell in love with a position. The philosophy they serve is roughly fifty years old and does not mention computers.