Composite

One number summarizing a score card — the harmonic mean of the trait scores.

The composite is the single 0–100 number that summarizes a score card. It is the harmonic mean of the per-trait scores, so a weakness in any one trait pulls it down.

Definition

For trait scores $s_1 \dots s_n$ above zero, the composite is their harmonic mean:

$$\text{composite} = \frac{n}{\sum_{i=1}^{n} \frac{1}{s_i}}$$

The harmonic mean is dominated by the smallest values, so it penalizes an uneven card more sharply than an arithmetic average would.

Mechanism

The composite carries three signals — and deliberately no tier label:

Signal	Derivation
composite	Harmonic mean of the trait scores.
confidence	The weakest per-trait confidence.
headroom	The largest per-trait headroom — the bottleneck trait.

A composite has no training distribution of its own, so there are no breaks to grade it against and no composite tier.

Interpretation

Read the composite first, then the traits that drag it down. Because the mean is harmonic, a card with one Weak trait is not “mostly fine” — that trait dominates the composite. The weakest-link confidence and the bottleneck headroom point at the same place: the trait to fix first.

Edge cases

• A single-trait model still returns a composite; it equals that trait’s score.
• A trait that scores exactly 0 is excluded from the composite rather than forcing it to 0 — the composite reflects the traits that carry signal.

Next

Score card

The full result the composite summarizes.

Tiers & breaks

Tiers, the breaks that define them, and the headroom the composite reports.