Methodology library

Every figure below links to the methodology that produced it, the protocol it was measured under, the calendar that defined its sessions, and the number of configurations attempted before it was published.

● clear ◐ attention ✕ blocked ○ none reasoning │ execution │ judgment

Lineage status

Version	Inquiry	Findings	Claims	Assumptions	Methodology	Protocol	Calendar	Run	Assessment	Policy	Publication	Errata	Comparability	State
methodology/hrp@1	◐	✕	●	●	●	●	●	◐	·	·	·	●	◐	attention at inquiry, comparability · affected by 2 findings
methodology/hrp@2	◐	✕	●	●	●	●	●	◐	·	·	·	●	◐	attention at inquiry, comparability · affected by 2 findings
methodology/hrp@3	◐	✕	●	●	●	●	●	◐	·	·	·	●	◐	attention at inquiry, comparability · affected by 3 findings
methodology/xsmom@1	●	✕	●	●	●	●	●	◐	·	·	·	●	●	affected by 1 finding

A dot reports the state of that artifact class, not its presence — so a version with many artifacts still reads as blocked when one of them establishes a blocker. Hover any dot for its condition.

Hierarchical Risk Parity (capped)

◐✕●●●●●◐···●◐ methodology/hrp@3 Not assessed

attention at inquiry, comparability · affected by 3 findings

Allocate across a fixed multi-asset ETF universe using correlation-based hierarchical clustering, subject to a per-asset ceiling that prevents a near-zero-variance instrument from absorbing the portfolio.

Versions: @1 · @2 · @3
Configurations attempted: 0 (0 count toward multiple-testing correction)
Grounded in: doi:10.3905/jpm.2016.42.4.59

Cross-sectional momentum (12-1)

●✕●●●●●◐···●● methodology/xsmom@1 Not assessed

affected by 1 finding

Rank a fixed multi-asset ETF universe by trailing total return over a 12-month formation window, skipping the most recent month, and hold the highest-ranked assets in equal weight.

Versions: @1
Configurations attempted: 0 (0 count toward multiple-testing correction)
Grounded in: doi:10.1111/j.1540-6261.1993.tb04702.x

What this library does differently

A methodology here is executable data, not prose. Its lookback, covariance estimator, weight bounds and rebalance cadence are declared fields — so changing one necessarily changes the result, and two versions can be compared field by field rather than by eyeballing two curves.

A published figure names both the methodology and the evaluation protocol that produced it, because methodology + protocol = performance. The protocol carries the trading calendar, the transaction costs, the execution lag and the data snapshot — all of which move the number.

Corrections are published, not quietly applied. Superseded figures stay reachable and link forward to what replaced them.