The Thesis Template
A fill-in template you can use today, a worked example, how the thesis feeds a simple DCF — and how its kill criteria become rules Telumin watches around the clock.
The template
Nine sections, one page. If you can’t fill in sections 2 and 6 — variant perception and kill criteria — you don’t have a thesis yet; you have a feeling.
Investment Thesis — [Company] ([TICKER])
Date: [YYYY-MM-DD] · Version: 1.0
1. One-sentence thesis
[Clear, falsifiable: why you own it and what drives the return]
2. Variant perception
[Why is the market wrong or incomplete right now?]
3. Business quality & moat
[Model · advantages · unit economics · management]
4. Valuation
[Method used (DCF, multiples) · intrinsic value · margin of
safety · what the market is already pricing in]
5. Catalysts
[Specific events that close the price/value gap, with timing]
6. Risks → kill criteria (break-conditions)
[Risk 1 → measurable break condition]
[Risk 2 → measurable break condition]
[Risk 3 → measurable break condition]
7. Expected return & horizon
[Base-case annualized return · horizon · up/base/down cases]
8. Position sizing & role
[Weight and what job it does in the portfolio]
9. Monitoring plan
[What you track, how often you formally review]A worked example (abridged)
Illustration, not analysis
Investment Thesis — Costco (COST) [illustrative example]
1. One-sentence thesis
A high-quality compounder whose membership model can grow free
cash flow mid-to-high single digits for many years — still
underappreciated at today's price.
2. Variant perception
The market treats it as a fully-priced, low-growth retailer and
underweights the membership flywheel, fee pricing power, and the
warehouse-expansion runway.
6. Risks → kill criteria
Renewal rates begin a sustained decline → review immediately
Operating margins compress structurally → reassess thesis
Large value-destructive acquisition → exit
Comparable sales slow with no temporary cause→ formal review
9. Monitoring plan
Membership metrics, comps, margins, fee moves — every quarter.
Full review yearly, or the moment any kill criterion fires.Notice the shape of section 6: every risk ends in a measurable condition and a predefined action. That is what makes a thesis monitorable — by you, or by a machine.
Linking the thesis to a simple DCF
The thesis is the story; the DCF turns the story into numbers. Each qualitative judgment maps to a specific assumption:
Thesis section → DCF assumption it feeds
─────────────────────────────────────────────────────────────
Variant perception + → growth rate, margin stability,
business quality and a defensible discount rate
Catalysts → higher near-term growth years
Risks & kill criteria → the downside / sensitivity cases
Valuation → where the DCF's output lands
Expected return → (DCF value ÷ price) − 1, annualizedWrite the qualitative thesis first
Judgments before numbers — otherwise the model launders your bias into precision.
Translate judgments into assumptions
Durable moat → steadier growth and a lower discount rate you can defend. Catalysts → the years that carry above-trend growth. Keep the terminal rate boring.
Run the base case — then run the kill cases
Re-run the same model with each break-condition triggered: lower growth, compressed margins, a higher discount rate. If even the kill cases are survivable, the thesis is robust; if not, you have learned your true margin of safety.
Write the output back into the thesis
The DCF’s value fills section 4; the spread to today’s price fills section 7. When a kill criterion later fires, the first thing you do is re-run the model with the new reality — the second is decide.
Wiring it into Telumin
Telumin is built for exactly one stage of this process: the monitoring. When you add a position, you write the reason in one line — the engine structures it into break-conditions and watches them continuously. Section 6 of your thesis is not a document you revisit yearly; it is a set of live rules.
- · Kill criteria → the thesis drift sweep re-checks them against filings, prices, and news.
- · Price-level risks → stop-loss watchers hold the line 24/7; the order stays yours.
- · The decision log → every accept/reject lands in the journal with your reason, building the review trail section 9 asks for.
- · Connected to Claude via MCP, you can draft the thesis in conversation and save the finished analysis — verdict, scored dimensions, evidence — back to Telumin, shareable by link.