How the analysis is produced

Sources

Analysis is built on primary sources: regulatory filings, earnings-call transcripts, company presentations, and disclosed market data. Secondary commentary is not an input. Where a figure appears in the output, it traces back to the document it was disclosed in and the date of that disclosure.

Structure

Company analysis follows a fixed template of eleven sections, applied identically every time. The template is not optional, and this is deliberate: when sections can be dropped, the ones that go are reliably the uncomfortable ones — management credibility, and what would falsify the thesis.

The fixed structure also makes two analyses comparable to each other, rather than comparable only to how persuasively each was written.

Multiple perspectives

Several independent analytical perspectives are run over the same evidence. Where they agree, the agreement is worth something. Where they disagree, the disagreement is surfaced rather than resolved by averaging, because an averaged answer discards precisely the tension that should prompt a closer look.

A minority bearish reading is surfaced prominently rather than outvoted. This is an intentional bias in the method: the case you least want to read is the one most worth reading.

Dating and revision

Every piece of analysis is stamped with the date it was produced and is not silently revised afterwards. Where a view changes, the change is recorded as a new dated statement rather than an edit to the old one. Analysis published on this site refers to a specific point in time and is not maintained as a current view.

Limitations — read this part

Machine analysis makes mistakes. It can misread a disclosure, miss context that is obvious to a person familiar with a company, or state something with more confidence than the evidence supports. Everything here should be treated as a well-organised starting point that you are expected to check, not as a conclusion to adopt.

Scenarios are not forecasts. Bull, base, and bear paths describe possible outcomes and the conditions attached to them. They are not predictions, they carry no probabilities we would defend, and the fact that one of three scenarios resembles what happened is not evidence of skill.

Coverage is uneven. Disclosure quality varies enormously between companies and jurisdictions. Analysis of a company that discloses little will be correspondingly thinner, and thinness is not always visible in the output.

Nothing here is personalised. The analysis knows nothing about your objectives, time horizon, tax position, existing exposure, or risk tolerance, and therefore cannot be suitable or unsuitable for you. That judgement is not one it is equipped to make.