Exposure: the contributions table follows the Lookback control

[mspinola]

The headline, the band, the %ile scale and the caption all ranked over whatever the Lookback control resolved to. The contributions table under them ranked every member against all history whatever the control said, because contribution_table called expanding_pct_rank unconditionally. So on "52 weeks" the page said "higher than 97% of the last 52 weeks" three lines above a column headed "%ile" measured against twenty years, with both on screen at once and nothing on either saying they were different questions.

Live, Gold alone on Lookback 52 weeks:

	before	after
headline	62nd percentile of the last 52 weeks	same
table row, %ile	99	61.5

Why threading it through rather than explaining it

The expanding rank has a real claim on this page. No unit here is stationary through time, and it is the only form that can say "the most ever". But that is an argument for keeping All history as the default, which it is, not for ignoring a control after the reader has moved it. Copy can describe two bases; it cannot make them comparable, and the question the table answers is "which market is driving the number in the headline".

Three columns move together or none of them do

• the two dollar percentiles, via a new window parameter on cotmetrics.exposure.contribution_table (Rank a contribution table over a window, not always all history cotmetrics#32)
• the Contracts %ile column from Exposure: the contract percentile, per market #64, which was written to match whatever the dollar columns did. Moving one without the other puts the row back to mixing histories by the other route.

Copy that goes false the moment the arithmetic lands

All in here rather than in a follow-up:

• the dollar column's header tooltip, "Against all of it, whatever the Lookback control says", written one commit earlier and true when written
• the table label, "Percentiles are each market against its own history"
• the standing explanation, which told a reader who moved the control that the band moved with it and said nothing about the table

The label also absorbs the second imprecision #64 left for whoever next edited it: a member's "own history" is the weeks the SET covers, since aggregate_exposure restricts every member to the weeks the total can price. Gold reads 88 alone and 81 in a set starting in 2002 because Russell does. Both restrictions are now named, and the set clause is dropped on a single market where it has nothing to qualify.

Floor

cotmetrics[options]>=0.9.0, a correctness floor of the loudest kind: against 0.8.0 the keyword does not exist, the call raises TypeError on every render, and no setting of the control avoids it.

Stacking, now resolved

Was stacked on #64 and blocked on mspinola/cotmetrics#32. Both have merged, so this is rebuilt as a single commit on main (3d3f1c6) with no merge commits, and check_dep_floors.py is green: cotmetrics 0.9.0 satisfies '>=0.9.0'.

Verification

492 tests pass, ruff check src tests clean on 0.15.22, floors green. Run against the real cotmetrics 0.9.0 editable install with no PYTHONPATH shadowing. Verified against the live store by rendering the page callback headlessly rather than in a browser. Nothing about the layout changed, only text and which weeks the ranks run over.

Reasoning recorded in docs/design/positioning-report-view-ideas.md under "The Lookback control reaches the table (2026-08-24)".

🤖 Generated with Claude Code