Is it a buy or a sell?

Three things settle it, and only one is about the company. The first is whether the business can deliver what its price already assumes. The second is how it sits against what you own: a superb company that duplicates your three largest positions adds concentration, not diversification. The third is your horizon, because the same volatility that is survivable over ten years is ruinous over one.

A page can help with the first. Only a view of your whole account can answer the second, which is the gap these pages deliberately leave open rather than paper over.

If you are already reading about a stock, you have the bullish argument. It is why you clicked. The version you are missing is the one that explains how it goes wrong, and the failure mode is rarely disagreeing with the bear case, it is never having read it.

So each page sets out both at full strength and names what would break each one. A thesis you cannot describe the failure of is not a thesis yet, and knowing in advance what would change your mind is the difference between deciding and reacting.

A price-to-earnings ratio is the market telling you how many years of current profit it is willing to pay for a claim on future ones. A high multiple is not a warning and a low one is not a bargain. Both are statements about what the market expects next, and the useful question is whether those expectations look achievable.

That is why comparing a multiple to the market average is close to meaningless. A software company with recurring revenue and 80% gross margins should trade at a different multiple from a steel producer, and always has. The comparison that carries information is against the company’s own history and against direct peers with similar economics.

The trap at both ends is symmetrical. An expensive stock can keep compounding for years if it grows into the price, and plenty have. A cheap one can stay cheap indefinitely if the business is slowly deteriorating, which is the classic value trap: the multiple looks attractive precisely because earnings are about to fall.

Every argument for owning a company reduces to a small number of things that have to happen. Revenue grows at some rate for some period. Margins hold or improve. The competitive position does not erode. Capital gets allocated sensibly. Writing those down turns a vague sense of optimism into a set of claims you can check against reality as it arrives.

The discipline this creates is knowing in advance what would change your mind. An investor who has written down that the thesis depends on margins staying above a certain level has a decision rule when they do not. An investor who has not written anything down has only a feeling, and feelings adjust themselves to whatever the price has recently done.

Each page below states the bull case as a set of conditions rather than as enthusiasm, which is what makes it checkable later.

The bear case is usually presented as a token counterpoint, a paragraph of risks nobody expects you to weigh. That is worse than useless, because it creates the impression of balance while providing none.

A real bear case names the mechanism. Not "competition could increase" but which competitor, doing what, and why the company’s defence might not hold. Not "regulation is a risk" but which rule, in which market, and what it would cost. Specific arguments can be checked as events unfold; general ones cannot, which is exactly why general ones feel safe to write.

Reading it first is a useful habit. By the time you are reading about a company you already have the bullish argument, because it is why you are here. The failure mode is not disagreeing with the bear case after weighing it. It is never having gone looking for it.

Everything above assumes the business can defend its position, and that assumption does most of the work in any long-horizon case. A company earning high returns attracts competitors; whether those returns persist depends on what stops them.

The durable defences are few and recognisable. Scale that makes the incumbent structurally cheaper to run. Network effects where each new user makes the product better for the others. Switching costs that make leaving expensive even when a rival is better. Regulatory or patent protection with a known expiry. A brand people will pay more for out of habit rather than analysis.

What is not a moat: being first, being biggest right now, having good current margins, or having a better product today. Those describe the present rather than protect the future, and they are the things most often mistaken for durability in a bull case.

The single most useful question before buying anything widely followed is what you believe that the price does not already reflect. A company everyone agrees is excellent is usually priced as though it is, and the quality is not the opportunity; the gap between the quality and the price is.

For heavily covered large caps, that gap is rarely informational. Thousands of professionals read the same filings within minutes. Where an individual investor does have an edge is horizon: institutions are judged quarterly and often cannot hold through a bad two years, and someone who can is being paid for patience rather than for insight.

That reframes the whole exercise. The question is not usually whether a company is good. It is whether you are being compensated for holding something others cannot, and whether you will still be holding it when the compensation arrives.

Whether a company is a good business and whether it deserves 15% of your portfolio are separate questions, and the second one determines far more of your outcome. A correct call at 1% barely registers. A wrong one at 30% can undo years of good decisions.

Concentration also arrives without being chosen. Winners grow into an outsized share of a portfolio precisely because they did well, and the position that has worked best is usually the one hardest to trim. That is how portfolios end up with a single name carrying most of their risk, assembled by inaction rather than by conviction.

None of that is visible from a page about one company. It only becomes visible when something reads the whole account and shows you what fraction of your outcome now depends on one argument being right.

Three hold up. The thesis broke: something you were relying on turned out to be false, and the argument for owning it no longer stands. The position outgrew its place: it now represents more risk than you would choose to take deliberately. Or you need the money for something specific and dated.

The price falling is not on the list, though it is the most common trigger in practice. A lower price with an intact thesis is the same investment at a better entry, and selling into it converts a paper decline into a permanent loss. The price rising is not on the list either, though "taking profits" sounds prudent enough that it rarely gets examined.

Tax makes this concrete rather than theoretical. Selling a long-held winner in a taxable account triggers a bill that a rebalance inside a retirement account does not, so the same decision has different costs depending on where the shares sit. Each page argues the case; where the shares live is a question about your accounts.

A buy or sell rating compresses everything above into one word, and the compression destroys precisely the information that mattered. It hides the horizon, the assumptions, the position size, and the portfolio it would join, all of which change the answer.

It also implies a certainty that does not exist. The same company is genuinely a sensible holding for someone with a decade and no overlapping exposure, and a poor one for someone who needs the money in two years and already owns three competitors. A rating cannot be right for both and is usually published as though it were.

So each page argues both sides at full strength and names what would settle it. That is more work to read than a rating. It is also the only version that stays useful once you know what else is in your account.