Capstone Lab: The Initiation Report

The note, section by section

An initiation has a fixed spine. First the business: what it sells, to whom, how it earns money, and the two or three drivers that decide the outcome — not the history of the company. Then the financial summary: the three statements compressed to the lines that matter, with the links shown, because a forecast that does not articulate is a set of assertions that will not reconcile when one driver changes. Then the returns analysis: return on invested capital against the cost of capital, decomposed so that margin, turnover and leverage are each visible, and the reinvestment required to produce the growth you forecast (F6, F7). Then the valuation, and it should be two methods that can disagree — a discounted cash flow with a sensitivity table on the two inputs that matter, and a comps table on the multiples the market uses for that industry. A single method produces a number; two produce an argument, and the gap between them is information about which assumption is doing the work. Then the scenarios: a base case, a bull and a bear, each with probabilities and each with the driver values that define it. Then the catalysts on a timeline, because a thesis that is right in five years and wrong in two is a thesis about financing as well as about the business. Finally the risk section and the position. The risks are the conditions that would falsify the thesis, stated as quantities, and the invalidation is the specific observation that ends it — a margin below a level for two quarters, a market share that stops growing, a discount rate that changes the base case. The position is then sized from that invalidation and from the distance the bear case implies, under the risk policy, so the note ends with a fraction of the book rather than an adjective. Everything above it exists to justify that fraction. • Business and drivers — not the corporate history. • Three statements, articulated, with the links visible. • ROIC against the cost of capital, with reinvestment, not just growth. • Two valuation methods that can disagree, with a sensitivity table. • Bull, base and bear with probabilities and driver values. • Catalysts on a timeline, and financing as a risk. • Invalidation as a quantity, and a position sized from it. The note is written for a reader who will check it. Assume every number will be recalculated by someone with the filings open, and write it so that being wrong is specific enough to be useful.

How this capstone is graded

Not on the return, and not on whether the shares move your way. The assessment is on the four things that make an analyst improvable: whether each claim was converted into a number with a source, whether the valuation used two methods and reported the sensitivity, whether the scenarios were probability-weighted with driver values rather than adjectives, and whether the position had a stated invalidation sized under a policy. A note that gets the direction wrong and does all four of those is a good note; a note that gets it right without them is luck with prose attached. The second half of the assessment is calibration. Before the note is filed, record what you expect for the two drivers you consider most important, with a confidence level: "19% operating margin in two years, 70% confident". Six months later, compare. The record of those forecasts is the only way to learn whether your 70% means 70%, and the pattern — overconfidence in margins, say, or a habit of assuming consensus-level growth — is the most valuable output of the whole subject. That is the same discipline that Market Psychology P16 applies to trade outcomes, moved upstream to the forecasts themselves. The capstone closes the subject by being self-contained: everything in it was built in an earlier lesson. The statements came from F2 to F5, the returns arithmetic from F6 and F7, the red flags from F8, the discounting and cost of capital from F10 and F11, the model from F12, the comps from F13, the scenarios from F15, and the question of what the price already assumes from F21. What the capstone adds is the assembly — and the discipline of writing down, in advance, what would make you wrong. The one-page test — Every claim: a number, with a source · The valuation: two methods, a sensitivity table, and the gap explained ← · The scenarios: three cases, probabilities, and the driver values that define each · The position: sized from a stated invalidation under the risk policy · The record: your two key forecasts with confidence levels, filed before the outcome A defensible note can still be a bad position: excellent analysis on a company whose price already reflects it earns nothing. The last check before sizing is whether the price disagrees with you, not whether your numbers are internally consistent.

The variant perception

A note that concludes the stock is cheap has not said anything the price does not already contain. The part that carries the analysis is the **variant perception** — the specific, stated difference between what your numbers imply and what the market’s numbers imply. “The market is modelling 3% growth; I think 7% is achievable, and here is the evidence and the mechanism” is a thesis. “The company is a strong franchise trading below intrinsic value” is a description of a price. A professional reader looks for the former and discards the latter. A variant perception implies its own falsifiers. If the disagreement is about growth, there is a growth rate that would resolve it, a date by which evidence should arrive, and a set of observable markers — unit economics, a backlog, a competitor’s share — that tell you early whether you are right. Writing those down before the position is what separates a thesis from a hope, and it makes the review honest: you are checking the mechanism, not the price. Finally, state what the market is getting right. Most of a company’s story is usually priced correctly, and the honest version of a thesis says which part is not. A note that disagrees with the market about everything is not a variant perception, it is a confession that the analysis has not engaged with why the price is where it is — and the price is the best single summary of the consensus you are betting against. If your note cannot name the number the market is using that you think is wrong, it is not a variant perception. It is a preference.

After the note: monitoring and the sell decision

The note answers what the business is worth and what the price is. It does not answer what happens next, and that is where most theses are actually lost — not because the analysis was wrong, but because the holder had no way to tell whether it had become wrong. A research note without a monitoring plan is a document that can only be read once. The plan has four parts and is written at the same time as the note, while the position is theoretical and the judgement is unclouded. First, the **milestones**: three or four specific operating numbers the thesis depends on, with the level expected and the date it should appear. Not “growth continues” but “segment margin above eighteen percent by the third quarter”. A milestone that cannot be missed is not a milestone. Second, the **falsifiers**: the specific observations that would show the thesis is wrong, defined before any of them occur. This is the most valuable sentence in the whole note, because it moves the decision out of the moment when the position is down and the explanation is scarce. Third, the **cadence**: how often each piece of evidence is checked and from what source — accounts quarterly, operating metrics monthly, and a firm rule against looking at the price more often than the plan requires, since a rising or falling quote is not evidence about a business. Fourth, the **sell discipline**, which is three separate triggers that are often mistaken for one. The thesis can be *achieved*, meaning the value has been realised and the remaining question is what the money does next. It can be *broken*, meaning a falsifier has occurred. Or it can be *pre-empted by a better use of capital*, meaning the discount closed because the price rose rather than because the value did — and the same business at a fair price after a re-rating is a different holding from the one that was bought cheap. The trap these four parts exist to prevent is the quiet rewriting of a thesis to fit new evidence. A falsifier does not update a thesis; it either confirms it or breaks it, and a broken thesis should release the capital rather than host a fresh argument that happens to keep the position open. A genuinely new thesis requires a new note, with its own milestones and its own falsifiers, written down at the time it was formed. The final piece is the record. Most investors cannot say afterwards whether they sold for a good reason or a bad one, because the reason is reconstructed from the outcome. Logging each decision at the moment it is taken, against the specific part of the plan it corresponds to, is what makes that question answerable later — and what turns a portfolio from a set of opinions into a process that can be improved. • Milestones are specific numbers with dates; anything unmissable is not a milestone. • Falsifiers are written before the position exists, so the judgement is not made under pressure. • Three sell triggers — achieved, broken, better use — are different decisions. • Log the decision when it is taken, or the reason will be reconstructed from the outcome. The one-page test this capstone is graded against asks whether a stranger could act on the note. The monitoring plan is the part that lets a stranger hold the position as well as open it.

Sources

• The structure of an initiation report and what a reader looks forStandard sell-side and buy-side research formats
• Sensitivity analysis and scenario weighting in practiceMcKinsey, "Valuation"; Rappaport & Mauboussin, "Expectations Investing"
• Position sizing from a stated invalidationRisk & Sizing R4–R7 — the position is sized to the condition that ends the thesis
• Base rates and calibration: recording what you expected against what happenedStandard forecasting-practice literature on calibration and journaling