iteration 02 · feature #38

cumulative_residual_extropy — how spread-out are the recent returns?

A bin-free dispersion / concentration measure (CREx). It asks: across the last ~200 bars of gap-aware log-returns, how spread-out (vs concentrated) is the distribution — without ever choosing a histogram bin. Computed on the shared gap-aware substrate.

✅ PR OPEN — awaiting squash-merge — PR2 (stacked on bootstrap)

Status: PR OPEN — kernel ported, wired, all gates GREEN. PR2 #126 · feat/odb-ext/cumulative_residual_extropy · base=feat/odb-ext/rqw-crex-bootstrap.
Plain-language one-liner: how spread-out are the recent returns? (a bin-free dispersion/concentration measure).
Substrate: gap-aware logret — trailing 200-bar window → diff(log(close)) → null gaps > 8h → drop NaN → require ≥ 8 finite.
Tier / TOL: ε 1e-9 (VALUE-within parity vs pinned numpy 2.5.0 / scipy 1.18.0 oracle).
Parity: 11,200 finite compared (== oracle finite, non-vacuous); max-abs-dev 1.4e-17 batch / 8.8e-16 streaming vs TOL 1e-9.
Challenge-and-Held: 10 attacks, 10 held, 0 conceded.

🟡 Hardening witness — scale-witness

CREx is scale-sensitive: the raw-input value and the standardized-input value differ by roughly 1/std. A well-meaning reviewer who z-scores the returns "to normalize" would silently change the answer.

The witness asserts that z-scoring the input FAILS parity — that locks the RAW-input convention into the test suite, so the scale convention cannot drift in a later refactor. On the frozen fixture the RAW-scale CREx is ≤ 0 throughout. scale-witness

The §5 pipeline (complete — all stages done)

stage	what happened	state
worktree	branch stacked on the bootstrap branch (base of the stack).	done
gap-aware oracle	pinned numpy CREx reference on the gap-aware logret window; ε 1e-9.	done
kernel port	causal Rust kernel; RAW-input (no z-score).	done
wire	BarRow · COLUMN_NAMES · BarDerivedV2 · schema.sql · feature_metadata — one no-split-brain PR.	done
SOTA §6 + §6.RG	parity · property · crash-stress · batch≡streaming · golden + the scale-witness. 11,200 finite compared (== oracle, non-vacuous); max-abs-dev 1.4e-17 batch / 8.8e-16 streaming vs TOL 1e-9.	done
all-gates	fmt · clippy · schema-contract · drift — GREEN.	done
no-split-brain	schema + metadata + BarRow + cohort + sha256 landed together.	done
challenge-held	adversarial Attacker+Defender — 10 attacks, 10 held, 0 conceded; SHIP.	done
PR	PR2 #126 OPEN, stacked on bootstrap — awaiting squash-merge.	OPEN

← back to the running dashboard · next: #7 right_quantile_weight_tailweight