Chapter 09 — Write the Thesis and Size the Position | Clarion: Buffett-Style Investing with AI on Zo Computer

Turn a candidate into a written investment thesis with a score, a target weight, and explicit kill conditions. Capital does not move until this exists.

9.1 Why write it down

A thesis is the document that justifies putting capital into a name. Writing it down does three things at once:

It forces you to be specific. Vague convictions do not survive being typed.
It gives you something to score — the system can grade a thesis only after it exists.
It gives you something to monitor and to update when the world changes — the topic of the next chapter.

Without a written thesis, you have an opinion. With one, you have a position you can defend, monitor, and eventually exit on facts rather than feelings.

The rule

No capital moves into a name until there is a thesis file. The system treats this as a hard rule. So should you.

A published Clarion thesis for DECK (Deckers Outdoor). Header shows the ticker, status (Watchlist), current price $102.72, and snapshot date. A metric strip below the header lists trailing P/E, free-cash-flow yield, market cap, operating margin, and gross margin. A bordered thesis-summary panel explains the call in one paragraph. Below it: two cards titled 'The market is using the wrong mental model' and 'The margin expansion is the tell'. — What a finished thesis looks like, published as a page. One verdict on the first line, the numbers that matter in a strip, the one-line summary, then the “what I believe” cards. The rest of this chapter is about how to get here. More examples at cis.zo.space.

9.2 What the system writes for you

The clarion-thesis-write skill scaffolds a markdown file in your workspace at ~/clarion/theses/<TICKER>.md with all the sections in the canonical order, pre-filled with what the system already knows (filings cited, fundamentals, regime context). Your job is to fill in the prose — with the help of the Clarion Thesis Architect persona, which is deliberately picky about vague claims.

In Zo chat
>Scaffold a thesis on KO.
Pre-flight:
· 10-K and 10-Q indexed.
· Buffett-lens evaluation already run (verdict: Watchlist at $66).
· No existing thesis file for KO.
· Bucket: value (confirmed).

Scaffolded ~/clarion/theses/KO.md with these sections ready to fill:
(1) What I Believe — the core thesis in 1-3 paragraphs.
(2) Why I Believe It — numbered evidence with filing citations.
(3) What It Is Worth — bear / base / bull scenarios with the math.
(4) Kill Conditions — three specific events that would break it.
(5) Why Now — the catalyst, or the patience argument.

Three filing citations pre-seeded as starting evidence. Ready to work through section 1 with you.

9.3 How the system scores a thesis

Once a thesis is written, the system grades it on six components. Each one gets a score on a comparable scale; the components combine into an overall score that maps to an action.

Component	What it measures
Valuation safety	Current price versus the base case fair value you wrote into the thesis. The bigger the discount, the higher the score.
Business health	Quality of the underlying operations: margins, returns on capital, balance-sheet strength, revenue trend.
Insider alignment	Are officers and directors buying or selling their own shares? Read from Form 4 filings.
Catalyst proximity	Is there a near-term event — earnings, a product launch, a regulatory decision — that could move the price toward fair value?
Thesis integrity	Is the original thesis still consistent with the most recent filings, or have new disclosures undermined it?
Risk environment	The current regime, applied to this kind of position. A cyclical name takes a bigger penalty in red than a staple does.

The six numbers are combined into an overall score. The overall score and any triggered kill conditions decide the action.

9.4 The action ladder

ADD

Score is strong and the regime is supportive. The position is undersized for what the thesis justifies.

HOLD

The thesis is intact and the price is in the range. No action.

REDUCE

Score has weakened. Trim the position rather than waiting for it to break.

EXIT

The thesis is broken or a kill condition has fired. Get out.

Kill conditions override the score

A triggered kill condition forces EXIT regardless of how good the score otherwise looks. That is what makes them kill conditions. The system enforces this in code — the AI cannot “reason it away” on you.

9.5 Sizing the position

Score tells you direction (add / hold / reduce / exit). It does not tell you how much capital to put in. That is a separate decision, and a simple one once the thesis is written.

Three constraints apply, in order:

Bucket cap. Long single-name equity ideas live in the Value bucket, which has a regime-adjusted target. In a friendly regime that bucket is roughly 55% of the portfolio; in danger it drops to 40%.
Single-position cap. No single equity position exceeds 10% of the total portfolio. Most positions land in the 2-5% range. The 10% cap is a ceiling, not a target.
Conviction tilt. Within those limits, conviction (the thesis score and the size of the discount to fair value) decides where in the range a position sits. A higher-conviction thesis gets a larger weight inside the cap.

In Zo chat
>If I open KO based on this thesis, what size should I run?
Thesis score 64 / 100 (HOLD-leaning). Regime ORANGE — Value bucket target 50%.

Single-position cap: 10% of portfolio.
Conviction tilt for a score of 64: roughly 3% of the portfolio, on the lower half of the typical range.

Recommended target: 3% of the portfolio. Build in two to three tranches rather than all at once, in case price moves in your favor.

The sizing answer is intentionally narrow. The system will not tell you to put 20% of the portfolio in a single name even if the score is excellent — that limit exists because every investor, including the best ones, has been wrong about a high-conviction idea at least once.

9.6 Ready to act?

Before any capital moves, the thesis should clear a short checklist:

The base case fair value is written into the thesis — this is what the monitor uses to recompute valuation safety later.
At least three kill conditions are listed, each measurable in a filing or a price chart.
Every claim in the “Why I Believe It” section cites a filing.
The overall score is at least 55. Anything lower means the thesis itself is too thin to act on.
The size is set: a percentage of the portfolio, not a dollar amount you happen to have lying around.

Then, and only then

Once the thesis is scored, sized, and the checklist is clean, you can place an order — or a limit order at a better price — and move on to the management chapter.

Key takeaways

A written thesis is the entry ticket. No thesis, no capital.
Six components feed the score: valuation safety, business health, insider alignment, catalyst proximity, thesis integrity, and risk environment.
The score plus kill conditions produce an action: ADD, HOLD, REDUCE, or EXIT.
Sizing comes from three constraints: bucket cap, single-position cap, and conviction tilt.
A thesis with a score below 55 is too thin to act on — strengthen it or set the name aside.