# ICT Analyzer Audit — Panel Charter

**Version**: 1.0
**Effective**: Upon formation of the first quarterly panel
**Owner**: Drive By Numbers (DBN)

---

## 1. Purpose

This charter governs the quarterly independent audit of DBN's ICT Analyzer — the automated system that tags every trade with ICT/SMC structural labels (Order Blocks, Fair Value Gaps, Market Structure breaks, Kill Zones, Premium/Discount zones) at entry time. The audit ensures the Analyzer's ground-truth tags remain accurate, credible, and aligned with expert consensus.

The audit is designed to be transparent, repeatable, and independently verifiable. All results are published publicly with no redactions.

## 2. Panel Composition

Each quarterly panel consists of **3 independent ICT/SMC experts**. Panelists are selected based on demonstrable expertise in ICT and Smart Money Concepts, including but not limited to: published chart analysis, verified trading records, educational credentials, recognized community standing, or professional trading experience.

### Eligibility Requirements

- Demonstrable ICT/SMC expertise (minimum 2 years of active practice)
- No current employment by or financial relationship with DBN
- Willingness to complete 200 trade reviews within the 2-week review window
- Agreement to blind review protocol (no access to Analyzer output or other panelists' tags during review)

### Conflicts of Interest

Panelists must disclose any financial relationship with prop firms, trading platforms, or education companies that could create a conflict of interest. Disclosed conflicts do not automatically disqualify a nominee but will be considered during selection.

## 3. Rotation Mechanism

Panel continuity is maintained through staggered rotation:

- **Panel size**: 3 panelists per quarter
- **Rotation**: 1 panelist rotates off and 1 new panelist joins each quarter
- **Maximum tenure**: 12 months (4 consecutive quarters)
- **Minimum tenure**: 1 quarter (panelists may voluntarily step down after completing their first review)
- **Inaugural panel**: All 3 seats filled simultaneously from the nomination pool. In Q2, 1 rotates off. Rotation continues thereafter.

### Rotation Selection

When a seat opens, the next panelist is selected from the nomination pool based on:

1. Diversity of expertise (ICT vs SMC specialization, different instrument focus)
2. Credential strength
3. Geographic/timezone diversity (to avoid review-window bottlenecks)
4. Nomination date (earlier nominations given slight preference, all else equal)

The outgoing panelist is the one with the longest continuous tenure. In ties, the panelist who has completed the most quarterly reviews rotates off first.

## 4. Nomination Process

Nominations are accepted year-round through the public nomination form at `drivebynumbers.com/audit.html#nominate`. Self-nominations are welcome and encouraged.

### Nomination Fields

- Nominee name (required)
- Nominee email (required)
- Credentials / qualifications (required) — free text explaining why the nominee is qualified
- Nominator name (optional)
- Nominator email (optional)

### Nomination Review

Nominations are reviewed on a rolling basis. Status progression: `pending` → `reviewed` → `accepted` | `rejected` | `waitlisted`. Accepted nominees enter the pool for the next available seat. Waitlisted nominees remain in the pool for future quarters.

## 5. Quarterly Review Process

### Timeline

Each quarterly audit follows this schedule:

| Phase | Duration | Description |
|-------|----------|-------------|
| Sample generation | Day 1 | 200 trades randomly selected from that quarter's Bridge, Sim, and Paper trades |
| Review window opens | Day 1 | Panelists receive access to the 200 sampled trades via the Audit Portal |
| Blind review period | 14 days | Each panelist independently tags all 200 trades |
| Review window closes | Day 14 | Submissions locked — no further edits |
| Agreement computation | Day 14-15 | Per-pattern and aggregate agreement rates calculated |
| Publication | Day 16 (48hr after close) | Results published on `audit.html` — no redactions |

### Sample Selection

- **Sample size**: 200 trades per quarter
- **Sources**: All trades with autoTags generated that quarter — Bridge (live venue), Sim (prop challenge, backtesting), Paper (paper trading)
- **Method**: Cryptographically random selection (seeded PRNG for reproducibility)
- **Frozen state**: Each sampled trade includes the ICT Analyzer's full output at entry time: autoTags array, kill zone, trend, premium/discount zone, active order blocks, unfilled FVGs, recent structure points, and 100 candles of chart context around entry

### Blind Tagging Protocol

Each panelist reviews every sampled trade independently:

1. **No Analyzer output visible** — panelists see only the chart context (100 candles around entry), entry price, exit price, direction, and symbol
2. **No other panelists' tags visible** — tagging is fully blind until the review window closes
3. **Independent assessment**: For each trade, the panelist provides:
   - Their own set of ICT tags (OB, FVG, BOS/CHoCH, etc.)
   - Kill zone identification
   - Trend assessment (bullish/bearish/ranging)
   - Premium/Discount zone
   - Order block identification (with approximate price levels)
   - FVG identification (with approximate price levels)
   - Market structure points
   - Agreement/disagreement flag with the Analyzer (computed post-submission)
   - Disagreement notes (required when disagreeing)
   - Confidence rating (1-5)

### Agreement Computation

After the review window closes:

1. Each panelist's tags are compared against the Analyzer's frozen tags for each trade
2. Agreement is binary per pattern type per trade: agree or disagree
3. **Per-pattern agreement rate** = (agreements / total comparisons) for each pattern type (OB, FVG, BOS, KZ, PD)
4. **Aggregate agreement rate** = weighted average across all pattern types
5. **Unanimous disagreements** = count of trades where ALL 3 panelists disagreed with the Analyzer on the same pattern

## 6. Publication

Results are published publicly 48 hours after the review window closes. The published report includes:

- **Aggregate agreement rate** (single headline number)
- **Per-pattern agreement rates** (OB, FVG, BOS, KZ, PD)
- **Total trades sampled**
- **Number of panelists**
- **Unanimous disagreements** count and descriptions
- **Panel roster** (names and credentials of that quarter's panelists)
- **Quarter identifier** (e.g., "2026-Q3")

No results are redacted, edited, or withheld. If the agreement rate is low, it is published as-is.

## 7. Retrain Trigger

If the **aggregate agreement rate falls below 85% for two consecutive quarters**, an automatic retrain flag is raised:

- A public alert banner appears on `audit.html`
- The engineering team initiates a ground-up review of all ICT Analyzer detection modules (OB, FVG, BOS, KZ, PD)
- The retrain review process is documented and its outcomes published
- The flag is cleared when a subsequent quarterly audit achieves ≥85% aggregate agreement

This mechanism ensures the Analyzer cannot silently degrade — experts enforce accountability.

## 8. Stipend

Panelists receive a quarterly stipend for completing their review:

- **Amount**: To be determined based on platform revenue. Initial target: $500-1000 USD equivalent per panelist per quarter.
- **Payment**: Issued within 30 days of publication, conditional on completing all 200 trade reviews within the review window.
- **Partial completion**: Panelists who review fewer than 200 trades receive a pro-rated stipend based on completion percentage, provided they reviewed at least 100 trades (50%).

The stipend compensates for the time and expertise required to review 200 trades with the rigor this audit demands. It does not create an employment relationship.

## 9. Confidentiality

- Panelists may publicly state that they serve on the DBN Audit Panel
- Panelists must not share their individual tags, disagreement notes, or internal communications before publication
- After publication, panelists may discuss the published results freely
- DBN will not share individual panelist tags publicly — only aggregate and per-pattern rates are published

## 10. Amendments

This charter may be amended by DBN with 30 days' notice to active panelists. Substantive changes (panel size, rotation rules, retrain threshold) require notification to all past panelists and public posting of the updated charter.

---

**Document history**

| Version | Date | Change |
|---------|------|--------|
| 1.0 | 2026-04-19 | Initial charter |