SAMPLING + STOPPING INTEGRITY · 0.2-ST1

Freeze the finish line.
Then hide the score.

A team must not watch W and decide that today is the perfect day to stop. Profile 0.2-ST1 binds sampling units, exposure, cutoff, interim access, operational overshoot, and final analysis unlock to a prospective result-independent rule.

Verify a closure ↓Download template Open contract Preregistration gate

1FROZEN CUTOFF

7INTEGRITY GATES

0EARLY W LOOKS

0RESULT-DRIVEN STOPS

01 / SEQUENTIAL INTEGRITY

The result cannot
choose its sample.

W is descriptive, but a leaderboard still creates selection pressure. A fixed, independently attested closure boundary prevents favorable interim noise from deciding enrollment or follow-up.

01FREEZE THE FINISH LINE

Commit the unit target, exposure floor, calendar cutoff, sampling frame, and executable closure rule before activity.

02LOCK W UNTIL CLOSE

No sponsor, operator, or analyst sees primary outcomes or aggregate W while enrollment and follow-up decisions remain open.

03MONITOR WITHOUT PEEKING

Independent reviews may inspect safety, privacy, operations, and completeness—but not primary outcomes or W.

04KEEP THE OVERSHOOT

Every unit activated before closure and all of its observed exposure stay in the evidence and analysis denominators.

02 / DETERMINISTIC CLOSURE

Count first.
Unlock outcomes last.

The frozen unit target, exposure floor, and calendar cutoff must all be satisfied. Primary outcomes and W remain inaccessible until the observation cutoff has closed.

CLOSURE IDENTITYN = N_plan + N_overshoot

analysis units = every activated unit

FIRST PRIMARY ACCESS ≥ FINAL CUTOFF

03 / LOCAL REPRODUCTION

Replay the close.
Prove nobody peeked.

The browser verifier checks prospective timing, target minima, the exact cutoff, access logs, result-blind monitoring, target and cutoff changes, overshoot retention, public claims, and independent evidence.

LOCAL · ZERO UPLOADSAMPLING + STOPPING 0.2-ST1 VERIFIER

PROSPECTIVE CLOSURE MANIFEST

{
  "profile_version": "0.2-ST1",
  "target_certification": "WANTED_WILD",
  "protocol": {
    "closure_rule": "close_only_when_frozen_unit_target_exposure_floor_and_calendar_cutoff_are_satisfied",
    "primary_outcome_access_rule": "locked_until_final_observation_cutoff",
    "permitted_monitoring_purposes": [
      "safety",
      "privacy",
      "operational_integrity",
      "data_completeness"
    ],
    "efficacy_or_futility_stopping_permitted": false,
    "outcome_adaptive_enrollment_permitted": false,
    "outcome_adaptive_followup_permitted": false,
    "overshoot_rule": "retain_every_activated_unit_and_all_observed_exposure",
    "shortfall_rule": "no_certification_claim_until_every_frozen_minimum_is_met",
    "emergency_rule": "safety_or_privacy_stop_remains_visible_and_never_becomes_a_favorable_cutoff"
  },
  "study": {
    "study_id": "SYNTHETIC-STUDY-001",
    "preregistration_sha256": "b0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcde",
    "claim_revision_id": "claim-r1",
    "plan_registered_at": "2025-12-01T00:00:00Z",
    "first_benchmark_activity_at": "2026-01-02T00:00:00Z",
    "submission_at": "2026-08-28T19:00:00Z"
  },
  "plan": {
    "unit_type": "independent_environment",
    "planned_units": 24,
    "planned_exposure_hours": 120000,
    "planned_cutoff_at": "2026-08-28T17:00:00Z",
    "decision_rule_sha256": "930123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd",
    "sampling_frame_sha256": "940123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd"
  },
  "execution": {
    "actual_units": 24,
    "actual_exposure_hours": 120000,
    "operational_overshoot_units": 0,
    "first_activity_at": "2026-01-02T00:00:00Z",
    "enrollment_closed_at": "2026-01-31T00:00:00Z",
    "observation_cutoff_at": "2026-08-28T17:00:00Z",
    "first_primary_outcome_access_at": "2026-08-28T17:15:00Z",
    "primary_outcome_access_events_before_cutoff": 0,
    "unscheduled_primary_analyses": 0,
    "result_informed_extensions": 0,
    "result_informed_early_stops": 0,
    "target_changes_after_activity": 0,
    "cutoff_changes_after_activity": 0,
    "all_activated_units_retained": true,
    "all_overshoot_units_retained": true
  },
  "monitoring_reviews": [
    {
      "review_id": "MON-001",
      "purpose": "safety",
      "reviewed_at": "2026-03-01T12:00:00Z",
      "reviewer": "Independent Monitoring Statistician",
      "independent_of_sponsor": true,
      "primary_outcomes_visible": false,
      "aggregate_w_visible": false,
      "recommendation": "continue",
      "source_sha256": "910123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd"
    },
    {
      "review_id": "MON-002",
      "purpose": "data_completeness",
      "reviewed_at": "2026-06-01T12:00:00Z",
      "reviewer": "Independent Monitoring Statistician",
      "independent_of_sponsor": true,
      "primary_outcomes_visible": false,
      "aggregate_w_visible": false,
      "recommendation": "continue",
      "source_sha256": "920123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd"
    }
  ],
  "claimed": {
    "planned_units": 24,
    "actual_units": 24,
    "operational_overshoot_units": 0,
    "planned_exposure_hours": 120000,
    "actual_exposure_hours": 120000,
    "monitoring_review_count": 2,
    "primary_outcome_access_events_before_cutoff": 0,
    "unscheduled_primary_analyses": 0,
    "result_informed_extensions": 0,
    "result_informed_early_stops": 0,
    "target_changes_after_activity": 0,
    "cutoff_changes_after_activity": 0,
    "excluded_activated_units": 0,
    "excluded_overshoot_units": 0
  },
  "evidence": {
    "controlled_plan_uri": "https://example.org/wanted-sampling-stop-plan.json",
    "controlled_plan_sha256": "950123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd",
    "outcome_access_log_uri": "https://example.org/wanted-outcome-access-log.json",
    "outcome_access_log_sha256": "960123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd",
    "closure_report_uri": "https://example.org/wanted-closure-report.json",
    "closure_report_sha256": "970123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd",
    "public_history_uri": "https://example.org/wanted-sampling-stop-history.json",
    "public_history_sha256": "980123456789abcdef0123456789abcdef0123456789abcdef0123456789abcd"
  },
  "assessor": {
    "name": "Synthetic Stopping Assessor",
    "organization": "Independent Example Assurance",
    "independent_of_sponsor": true,
    "attested": true,
    "signed_at": "2026-08-28T18:30:00Z"
  }
}

READY TO VERIFY CLOSURE

Paste aggregate targets, access logs, review history, and evidence digests only. No participant data leaves this browser.

04 / DESIGN BASIS

Prospective stopping,
made executable.

These statistical and reporting sources inform the evidence topology. WANTED is not a clinical trial and does not claim regulatory conformity; it adopts the prospective-control principle because the same data-dependent stopping bias can affect a public benchmark.

ICH E9Place interim-analysis schedules and stopping guidelines in the protocol before unblinded access.PRIMARY SOURCE ↗ICH M11Specify interim purposes, responsible parties, decision criteria, and stopping or adaptation rules.PRIMARY SOURCE ↗SPIRIT 2025Prospectively state interim analyses, access, decision makers, and termination guidelines.PRIMARY SOURCE ↗CONSORT 2025Report planned size, interim looks, stopping rules, decision makers, and reasons for early closure.PRIMARY SOURCE ↗

NO FROZEN FINISH LINE, NO CERTIFICATION CLAIM

Safety can always stop a deployment. A favorable W cannot decide when the evidence is complete.

OPEN AUDIT PACK →

Freeze the finish line.Then hide the score.

The result cannotchoose its sample.

Count first.Unlock outcomes last.

Replay the close.Prove nobody peeked.

Prospective stopping,made executable.

Freeze the finish line.
Then hide the score.

The result cannot
choose its sample.

Count first.
Unlock outcomes last.

Replay the close.
Prove nobody peeked.

Prospective stopping,
made executable.