Our team

Why there is a method at all

I did not start with a system. I started the way most people do: reading widely, forming views, and backing the ones I felt strongest about. Some of that worked. What it never gave me was a way to tell, in advance and without flattering myself, which conviction was worth acting on and which one I simply liked.

The expensive lesson is that judgement is not the problem — unexamined judgement is. A view formed once and never re-tested will find reasons to survive long after the facts have moved. So the five structural tests ask the same questions of every company, in the same order, whether or not I like the answer, and every piece ends with the conditions that would damage its own argument. The method is not cleverer than a person. It just does not get bored, does not get attached, and cannot quietly change its mind about what it was looking for.

Where the tests came from, and what is worth reading: my influences and the reading list.

What this is, and what it is not

It is an assessment of business quality. Every piece puts one company through the same five tests and says plainly what came back. The ones that clear all five join the Exceptional Company Series — which is a judgement about the business and not a buy list. Exceptional can also be expensive.

It is never a view on the price. There are no price targets here, no model portfolio, no buy and sell alerts, and no performance claims. A company can be exceptional and far too expensive; those are two different findings and this site publishes only the first. What to pay is a separate discipline and it stays yours.

There is nothing to buy, and nobody to become a client of. I do not manage money, take clients, run a fund, or sell a service, a course or a set of signals. The only thing this site ever asks of you is an email address, and only if you want the writing sent to you.

It is not advice. I am not a financial adviser and nothing here is a personal recommendation. I do not know your circumstances, your tax position, your horizon or what else you own — which is most of what would actually matter.

What I get out of this

Nothing, financially. This makes no money at all. No paid tier, no paywall, no sponsorship, no advertising, no affiliate links — not even on the reading list, where they would be easiest to justify and would quietly turn a recommendation into an advertisement. No company has ever paid for coverage, been offered it, or been given sight of a piece before publication. The domain, the hosting and the market data are costs I pay, and nothing comes back.

What I get instead is the discipline. Writing it down in public, with dated conditions that could prove it wrong, stops me quietly revising a view later. I would be doing this work anyway; publishing it is what keeps it honest.

The site runs no analytics and sets no cookies. If any of that ever changes — a paid tier, most likely — this page will say so before it happens rather than after.

Positions, and how they are disclosed

I invest my own money, and I write about companies I may own. Pretending otherwise would be worse than the conflict itself.

So the rule is disclosure rather than abstinence. Every piece states my position in that company at the time of writing, including when it is nothing. I do not trade around publication to make a piece look better, and a holding is never a reason to soften an assessment — the whole record of what would prove each piece wrong is public and dated, which is the check on that.

How it is made, including the parts people usually leave out

The figures come from company filings and market data, pulled and computed by Legend, a system I built and run myself. Where a number is checkable it comes from the filing rather than a data provider, and the page says which.

The assessment itself is deterministic. Legend applies the same rules to the same numbers and returns the same answer every time. No model decides which companies clear the bar, and running it twice on one set of filings cannot produce two different grades.

I use AI in the research and the writing. It does not read the filings — Legend does, deterministically, as above. What it does is argue with my reasoning, check my numbers and help me draft. I am telling you because you would want to know, and because a publication that lectures companies about disclosure has no business being coy about its own.

What that does not change: every judgement here is mine, every published word has been read and approved by me, and I am accountable for all of it. A tool that drafts well is still a tool. Nobody gets to blame the software for a view that turned out to be wrong, least of all me.