The variants

Variant	Rows	What it is	Height @1440	Empty %	Open
A	16	cap=4 — what STRAT publishes today	1,072px	10%	open
B	47	cap=12	1,935px	6%	open
C	96	uncapped — every scored row	3,299px	3%	open
D	29	one row per finding, horizons inline — all 96 readings, none dropped	1,369px	7%	open

What the scan found

Density is not the constraint. Every variant scores sideways: false, leaks: 0, and empty share falls as rows rise (10% at 16 rows, 3% at 96). Nothing in the measurement argues for a cap.

The row was the wrong unit. The 96 rows are 29 findings — each document's reading of one sub-theme, emitting up to four near-identical rows that differ only in horizon. A flat table repeats the constant half four times and scatters the one thing that varies. cap=4 spends its whole budget inside a single sub-theme, which is why the panel read repetitive rather than thin.

15 of the 29 carry a term structure that disagrees across horizons — short=bear, medium=neutral, long=bull, future=bull. That relation is the finding, and it is invisible in A, B and C.