Skip to content

01 · Methodology

From sub-factor metrics to a single Global Rank

The ALB Core Model is a systematic ranking framework: a documented three-step ladder turns granular metrics into a single, comparable score for every name in the universe, for a single monthly reference date.

02 · The ladder

86

Sub-factor metrics

Fundamental, valuation, quality and market metrics computed for every name in the universe.

9

Macro-families

The sub-factor metrics are grouped and normalised cross-sectionally into nine macro-families, each scored 0–10 per security and published in full.

1

Global Rank

The macro-families are combined, by a documented function, into a single Global Rank — shown here as score_0_100.

Every level from the macro-families down is published per security: the nine family scores, the four style scores and the resulting decile. What is not published is the level above them — which 86 sub-factor metrics feed each family, how many, at what weight and in which direction. Nine numbers for a company say nothing about what produced them.

03 · Display scale & decile bands

One 0–10 scale, and an honest precision

Every score in this interface is shown on a scale where a higher reading is a better one — inverted from the engine's internal convention, where 0 is best. The overall score is a decile: exactly ten distinct values, roughly 95 securities to each. It places a security in a band rather than ranking it against every other name, and it is shown as a plain integer for that reason — stretched across a 0–100 axis those ten buckets would read as a percentile they are not. The four style scores use a continuous 0–10 scale and carry the finer ordering.

0
9
WorstBest

The overall score is a decile — one of ten bands, roughly 95 securities to a band, so it places a security rather than ranking it against every other name. The four style scores use a continuous 0–10 scale and carry the finer ordering.

04 · Universe & cadence

A monthly cross-section, a single release

The model is intended to run monthly. This interface currently carries one release, covering 978 large caps across United States · Canada · Developed Europe, organised into 10 GICS sectors and 60 GICS sub-industries. There is no historical series behind it yet — no prior release to compare against, no month-over-month deltas.

05 · The four styles

Growth, Valuation, Quality, Momentum

Alongside the Global Rank, the engine publishes four style scores per security. These carry default weights that combine into the Global Rank's own composite, and a manager can re-weight them from the equity screens — the universe re-ranks locally, on the same scale, without changing the underlying data.

Growth

Revenue and earnings growth relative to the rest of the universe.

Default weight 23%

Valuation

Price relative to fundamentals — earnings, cash flow and book value.

Default weight 30%

Quality

Profitability, margins and balance-sheet strength.

Default weight 23%

Momentum

Recent price and estimate trends relative to peers.

Default weight 25%

07 · Product boundaries

What ALB analyses, and what it never holds

ALB analyses investment universes. It is not a client-facing system, and it is not a place where information about your clients is stored, processed or inferred. That is a deliberate boundary rather than a feature that has not been built yet.

Inside the product

  • Listed-company data and the model’s own scores.
  • Universe definitions, strategy weights and constraints you configure.
  • Portfolios and reports you compose from that analysis.

Outside the product

  • No CRM, client portal or end-investor data platform.
  • No end-client identities, KYC files or suitability profiles.
  • No personal financial data belonging to your clients.

Suitability, appropriateness and the client relationship stay entirely with the professional. Nothing in ALB is designed to carry them.

06 · What this is not

A snapshot, not a forecast

The Global Rank and the four style scores describe how names compare to one another as of the current reference date. The model does not forecast returns, does not imply an expected return, and has not been back-tested for this interface. There is no buy/sell signal and no automated decision anywhere in this product — every screen is a read of where the universe currently stands, for a human to interpret.