PAS — the Pythia Academic Score

What the number means

PAS is a composite from 0 to 100 where higher means stronger than peers, not stronger in the abstract. That distinction is the whole design.

Two companies in different sectors can both score 72. It does not mean they are equally good businesses. It means each is stronger than most companies in its own GICS sector and country — a software company priced attractively against other software companies, and a utility that is safe against other utilities.

A score near 50 is neutral against peers. Above 70 is strong, below 30 is weak.

The five pillars

PAS is built from 22 raw inputs, grouped and weighted:

How the arithmetic works

Each input is z-scored within (GICS sector × country) — measured in standard deviations from its own peer group's mean, so a 40% gross margin is judged against software rather than against grocery.

Scores are then winsorized at ±3σ so one extreme outlier cannot dominate a pillar, averaged within each pillar, and mapped onto 0-100 through the standard normal CDF. The final number is therefore a percentile-like position, not a raw ratio.

What happens when data is missing

A pillar without enough data is dropped from the weighted average, not filled with a neutral value. Filling it with 50 would drag every newly-listed company toward the middle and quietly assert something we did not measure — a company with no momentum history has unknown momentum, not average momentum.

Companies with no GICS sector — or in a sector-and-country bucket too sparse for reliable statistics — get no PAS at all, rather than a score computed some other way. There is no peer group to be neutral against, and a number produced without one would not mean what every other PAS on the site means. Their raw inputs are still stored, so the drill-in can show what was measured and say why it stopped there.

Arguing with it

Every PAS on a company page opens into its pillar breakdown and the exact inputs behind it. If you disagree with the weighting — and there is no objectively correct weighting — the point of showing the pillars is that you can see which one is carrying the score and form your own view.