← Validation Lab
Forward Alpha ↗v0.17 Robustness ↗
RESEARCH LEDGER / FORWARD OPS & INTEGRITY

CIF v0.21.3

Centralize the experiment chain, audit the prospective registry continuously, and prevent operational drift from silently corrupting forward evidence.

Mode
GOVERNANCE + OPS
v0.18 REGISTRY IS UNCHANGED. v0.19 observes and audits the forward experiment; it cannot rewrite its universe, factors, regimes, horizons, hash or gates.
LATEST RESEARCH-INTEGRITY AUDIT

PASS

2026-10-05T12:46:15
forward_registry_hash_consistentPASS
no_pre_freeze_snapshotsPASS
capture_not_before_checkpointPASS
no_duplicate_week_asset_snapshotsPASS
one_checkpoint_per_registry_weekPASS
no_premature_forward_outcomesPASS
mature_outcomes_have_returnsPASS
mature_outcomes_computed_after_maturityPASS
outcome_maturity_matches_checkpoint_plus_horizonPASS
forward_collector_recent_or_never_startedPASS
pipeline_version_currentPASS
collector_schema_currentPASS
core_bootstrap_coherentPASS
historical_path_maturity_integrityPASS
Forward snapshot weeks9
Due pending outcomes0
Forward collector freshness0.0 h
Historical premature paths0
FORWARD REGISTRY LOCK

alpha-forward-registry-v1-2026-08-08

8b5cd9d834ca9b0e…
BNBSOLXRPADADOGELINKAVAXLTCTRXDOTBCHSUI
Hypotheses6
Mutated by v0.19NO
BackfillFORBIDDEN
Production formula changedNO
EXPERIMENT LEDGER

How we got here

10 registered stages
VersionStageScopePrimary resultVerdict
v0.9hei-mos-validation-labhistorical_validationHEI MOS replay + forward shadowAggregate historical MOS replay did not show positive excess-BTC rank information.DESCRIPTIVE_ONLY
v0.10cross-asset-mos-validationcross_asset_replicationFrozen liquid-alt basketCross-asset diagnostics expanded the evidence base without authorizing score changes.NO_AUTO_TUNING
v0.11mos-candidate-holdoutcandidate_researchMOS candidate holdoutCandidate research remained exploratory and did not alter production scoring.NO_PRODUCTION_CANDIDATE
v0.12signal-decompositiondecompositionMOS signal decompositionDecomposition did not justify a replacement MOS formula.NEITHER_SUPPORTED
v0.13alpha-factor-screenfactor_screenEight simple primary factorsEMA200 Trend Distance screened in as time-series only.ONE_FACTOR_SCREENED_IN
v0.14ema200-factor-confirmationindependent_confirmationExact EMA200 Trend DistanceEMA200 failed independent confirmation with sign reversal.EMA200_NOT_CONFIRMED
v0.15alpha-hypothesis-expansionexpanded_factor_screenTwelve factors across six familiesNo universal factor passed the frozen multi-horizon + temporal-stability gate.NO_CANDIDATES_SCREENED_IN
v0.16conditional-alpha-regimeconditional_confirmationSix factor × BTC-regime interactionsVolume-Price Thrust 7D × BTC Risk-On passed the first conditional confirmation.ISOLATED_CONDITIONAL_ALPHA_ONLY
v0.17conditional-alpha-robustnesssecond_confirmationExact v0.16 winner on new assets + pre-discovery blockThe interaction did not pass the preserved second-confirmation gate.SECOND_CONFIRMATION_FAILED
v0.18prospective-forward-alpha-registryprospective_forwardSix frozen factor-regime hypotheses; weekly no-backfill snapshotsProspective registry activated; historical evidence cannot promote any hypothesis.FORWARD_ONLY_COLLECTING
FORWARD CLOCK

Prospective milestones

no shortcut permitted
Registry freeze2026-08-07
Earliest 7d from freeze2026-08-14
Earliest 30d from freeze2026-09-06
Earliest 90d from freeze2026-11-05
Earliest 26-week gate2027-02-05
Calendar gate reachedNO

These dates are only earliest calendar milestones. Actual hypothesis eligibility can be later because each target and opposite BTC regime still needs sufficient per-asset observations.

FORWARD OUTCOME OPERATIONS

Pending vs mature

full-horizon only
7dfuture outcomes

96 mature · 12 pending

30dfuture outcomes

60 mature · 48 pending

90dfuture outcomes

0 mature · 108 pending

DECISION FIREWALL

Rules that v0.19 enforces

no alpha tuning