# `/strategies/` Page vs `strategies.json` — Coverage Audit

**Generated:** 2026-05-16
**Status:** Reference doc — produced during the family-tag migration audit (Contradiction #6 in [UNIFIED-FRAMEWORK-v1.md](UNIFIED-FRAMEWORK-v1.md))
**Source files:**
- Page: [strategies/index.html](../strategies/index.html) — CATALOG object, lines 255-468
- Page mirror: `/scorecards/_scientific_overlay.js` (per page footer at line 243)
- Registry: [system/strategies.json](strategies.json) — 9 entries, post-family-migration 2026-05-15

---

## TL;DR

| Surface | Count | Source of truth |
|---|---|---|
| `/strategies/` page (the trader's reference UI) | **19 strategies** | `_scientific_overlay.js` CATALOG |
| `strategies.json` (the evaluator's registry) | **9 strategies** (8 + 1 off-book sentinel) | The schema-driven evaluator |
| Both | 8 strategies (after mirror reconciliation) | — |
| Page-only | **10 strategies** | — |
| Registry-only | **1 strategy** (`trend-trade-short`) | — |

**Implication:** the trader can see and pattern-match against 19 setups, but the live evaluator can grade only 8 of them. Every trade tagged with a page-only setup is silently OFF-BOOK in the evaluator.

---

## Strategy-by-strategy inventory

### ✅ In both surfaces (8 strategies)

| Strategy id | Page name | JSON id | Family (canonical) | Notes |
|---|---|---|---|---|
| `vwap-pull-in` | VWAP Pull-In | `vwap-pull-in-long` | vwap_pullback_continuation | Aligned |
| `vwap-pull-in-short` | VWAP Pull-In (short) | `vwap-pull-in-short` | vwap_pullback_continuation | Aligned |
| `td-vwap` | TD VWAP | `td-vwap-long` | vwap_pullback_continuation | Aligned |
| `td-vwap-short` | TD VWAP (short) | `td-vwap-short` | vwap_pullback_continuation | Aligned |
| `trend-trade` | Trend Trade | `trend-trade-long` | vwap_pullback_continuation | Aligned |
| `backside-scalp` | Backside Scalp | `backside-scalp-short` | counter_trend_scalp | Aligned |
| `backside-long` | Backside Long ★ featured | `backside-long-long` | counter_trend_scalp | Aligned |
| (off-book) | (not on page) | `off-book` | off-book | Sentinel — correct that it's not on page |

### ❌ Page-only — 10 strategies needing schema entries

Sorted by priority (sample size × family fit × business case). Each row includes the type-catalog additions needed.

| # | Page id | Name | Family | n (page) | New types needed | Priority |
|---|---|---|---|---|---|---|
| 1 | `vwap-continuation` | VWAP Continuation | vwap_pullback_continuation | **24** | `level_breach` (PDH/PDL targets), existing types only | **P1** (highest n, full-band-driven, simple schema) |
| 2 | `vwap-pull-in -> trend` | Pull-In → Trend (hybrid) | vwap_pullback_continuation | 20 | `band_slope_threshold` (regime flip), `mfe_threshold` (trigger) | **P1** (proven sample, captures regime transition) |
| 3 | `opening-range-break` | Opening Range Break | level_breakout | 20 | `time_window` (OR boundary), `volume_above`, `range_compression` | **P2** |
| 4 | `back-through-open` | Back Through Open | gap_fade | 18 | `time_window` (first 15min), opening-print structural level (new) | **P2** |
| 5 | `gap-give-and-go` | Gap Give and Go (GGG) | vwap_pullback_continuation | 18 | `time_window`, mini-consolidation structural range (new) | **P2** |
| 6 | `day-3-breakout-down` | Day 3 Breakout/Down | level_breakout | 16 | Multi-day OHLC reference (D1 H/L), `time_elapsed_session_days` | **P3** (requires multi-day state) |
| 7 | `breaking-news` | Breaking News | event_driven | 16 | News-event proximity check (new), catalyst severity (new) | **P3** (requires news ingest wiring) |
| 8 | `earnings-day-1` | Earnings Day 1 | event_driven | 14 | Earnings event proximity (new), PM range vs AH range comparison (new) | **P3** (requires earnings calendar wiring) |
| 9 | `htf-breakout-down` | HTF Breakout/Down | level_breakout | 14 | Daily/weekly S/R reference (new), `time_elapsed_session_days` | **P3** |
| 10 | `swang-rubberband` | Swang's Rubberband | volume_profile | **0** | `extension_atr` (new), `volume_climax_bar` (new), top-N-volume-bar check (new) | **P4** (zero history; new strategy) |
| 11 | `volume-gap-flush` | Volume Gap Flush | volume_profile | **0** | `vp_node_proximity` (LVN pocket), `level_breach` | **P4** (zero history; depends on MTF VP wiring) |
| 12 | `fl2-fash-late` | FL2 Fash Late | vwap_pullback_continuation | **0** | Grade-based discriminator (A/B/C from band depth), `level_hold` on 9 EMA | **P4** (zero history; needs grade taxonomy first) |

Note: I counted 12 page-only items above; #10-12 have n=0 (newly added to page, no live history yet).

### ⚠️ JSON-only — 1 strategy

| JSON id | Why not on page? | Action |
|---|---|---|
| `trend-trade-short` | Short mirror of `trend-trade`; page shows only the long. Likely an oversight when the page was built — `td-vwap` and `vwap-pull-in` both have short mirror entries, so the convention is to surface both directions. | **Trivial fix:** add a `trend-trade-short` entry to the page CATALOG (lines 255-468 in `strategies/index.html`). Mirror of `trend-trade` with direction inverted. |

---

## Recommended sequencing

### Sprint 1 (this/next week — purely additive, no infrastructure needed)

1. **Author `vwap-continuation-long` and `vwap-continuation-short`** in `strategies.json` — uses existing type catalog (no v1.2 types needed). n=24 page sample makes the empirical block populatable from day one.
2. **Author `vwap-pull-in-trend` hybrid** in `strategies.json` — uses existing types + adds a regime-flip trigger that uses `band_slope_threshold` (already in catalog). Captures the "started as Pull-In, becomes Trend Trade" promotion logic.
3. **Add `trend-trade-short` to the page CATALOG** — mirrors `trend-trade` with direction inverted. 2-minute task.

After Sprint 1: 11 strategies in registry, 20 on page, gap closes from 11 to 9.

### Sprint 2 (after v1.2 type catalog promoted)

4-6. **Author `opening-range-break`, `back-through-open`, `gap-give-and-go`** — each needs 1-2 new types (`time_window`, opening-print/mini-consolidation structural levels). These are MORNING-WINDOW strategies that share infrastructure.

After Sprint 2: 14 strategies in registry. Gap closes from 9 to 6.

### Sprint 3 (needs event-driven infrastructure)

7-9. **Author `breaking-news`, `earnings-day-1`, `htf-breakout-down`, `day-3-breakout-down`** — each needs event-driven primitives (news proximity, earnings calendar, HTF S/R). The earnings calendar already exists in MGP infrastructure; need to expose it as an evaluator primitive.

After Sprint 3: 18 strategies in registry. Gap closes from 6 to 2.

### Sprint 4 (after MTF VP scanner wired to evaluator)

10-11. **Author `swang-rubberband`, `volume-gap-flush`** — both volume_profile family, both n=0 currently. Need MTF VP scanner output piped to evaluator. Defer until VP infrastructure is consumed.

12. **Author `fl2-fash-late`** — depends on the A/B/C grade taxonomy being formalized as a new schema layer.

After Sprint 4: full coverage; page and registry aligned at ~20-22 strategies.

---

## Side findings

1. **Page CATALOG comment claims "kept in sync with /scorecards/_scientific_overlay.js"** (line 253). ✅ **Verified 2026-05-16**: `_scientific_overlay.js` already uses canonical snake_case family values (`vwap_pullback_continuation`, `counter_trend_scalp`, etc.) — it was always aligned with the page. The 2026-05-15 `strategies.json` migration brought the registry into line with what was already there. No further alignment work needed on the JS file.

2. **`backside-long` is the FEATURED card** (line 335: `featured: true`). The feature flag isn't currently a field in `strategies.json`. Suggest adding `featured: bool` to the canonical schema — small extension, surfaces directly on the page.

3. **The `aka` field on the page** is rich — e.g. `swang-rubberband` lists "Exhaustion-Climax Reversal / SMB Swang Rubberband". The registry has no `aka` field; trades tagged with the alias name won't link to the canonical strategy. Suggest adding `aliases: string[]` to schema and a lookup function.

4. **Page entries have per-strategy `expectancyR / winRate / avgMaeR / avgMfeR / n` blocks** that duplicate what should live in `empirical:` in the registry. After the v1.2 promotion, the page should READ from the registry (or both should read from a common source — Fix-It-Once rule #1). Today they diverge silently.

---

## What this means for the framework

This audit confirms Contradiction #6 from `UNIFIED-FRAMEWORK-v1.md` Part 12 as concrete and tractable. The path to full coverage is **incremental, sequenced over 4 sprints, and most P1 items need ZERO new infrastructure** — they reuse the existing type catalog.

Most importantly: **the trader is currently flying with 19 setups but only 8 live-graded**. That gap is silent in scorecards (every page-only setup gets the `off-book` sentinel grade — which then collides with all 3 meanings of "Off-Book" from Contradiction #1). Closing this gap also clarifies which trades are LEGITIMATELY off-book vs which were tagged off-book purely because the schema didn't cover them.
