Every market in the benchmark valued against its own economy. Every sector modelled the same way. Every company put through three unrelated disciplines. Published in full, every month.
Most equity research begins with a company and works outward. This begins with the economy and works in, because that is where most of the answer lives.
The framework has run in its essentials for many years. What follows is all of it: the five macro factors and why each was chosen, the arithmetic that turns a model reading into a portfolio weight, the ten screens and the two things ranking costs you, and the four fundamental models built from the accounts. There is nothing withheld for the paid section, because a method you cannot examine is not a method, it is a claim.
Benchmark as at 08/Sept/2026. Every figure on these pages is current model output and moves when the models are re-estimated.
The observation the whole framework is built on.
Which market a company is listed in explains more of its price movement than anything about the company.
What it does, in global terms, rather than where it is domiciled.
The part most research spends all of its time on, and the smallest of the three for a large listed business.
The proportions are approximate and they shift as you move down the capitalisation scale. Read as a statement of where analytical effort should go for large global equities, they are hard to argue with, and almost universally ignored.
Valuing a whole equity market against the economy underneath it.
Read the chapter IIThe same discipline applied sector by sector, and how the two become an allocation.
Read the chapter IIINarrowing thirteen hundred companies to the ones worth the accounting work.
Read the chapter IVThree unrelated disciplines, asked the same question about the same company.
Read the chapterThe whole chain, in order, with nothing running on cached results.
Constituents and weights are taken from the MSCI World. Companies entering are added; companies leaving keep their models but are flagged so that they drop out of every benchmark calculation from the date they left.
National statistics agencies and central banks, per country. Where a series is stale, it is not used; a source that has stopped publishing is not a source.
One global, twenty three country, eleven sector. Each produces a model value, a mispricing and a goodness of fit.
Weights must sum to the whole from both directions. If they do not, the run stops rather than publishing a matrix that does not add up.
Econometric models, ten screens, four fundamental models, forensic panel, then the composite.
Written commentary is generated for each card on each company page, and for the allocation as a whole.
Each run is archived at its own permanent address, so a figure quoted in a document sent last month still resolves to the figures that document quoted.
Models are replicable. Anyone with the data and the patience can build them, and we would rather say so than pretend otherwise. What is not replicable is what gets written next to them: an explanation of where the models disagree, which reading the evidence supports, and what would have to change for the conclusion to change.
That is what the client area is for, and it is the only part of this that is not described in full above.