aggressive_bear confirmed at 10:30 — after the 10:10 entry. Indicator-supported grade: daily-aligned short, intraday mixed-at-entry. Family = vwap_pullback_continuation (short).avwap (the retrace touched the 2-day line).vwap_reject_short engine as the TOD matrix — no lookahead.
| Attempt | Window (ET) | Size | What happened | Result |
|---|---|---|---|---|
| 1 — the probe | 08:21–08:26 (pre-mkt) | 5 sh short | Shorted 1048.31 too early, no confirmation; covered 1058 on the pop. | −$48.45 stopped |
| 2 — timid | 08:30–10:08 | mostly 1 sh | Right idea, no size. A grind of 1-share scalps clawing the probe back to flat. | scratched back to +$4 |
| 3 — the A+ trade | 10:09–10:15 | 20 sh short | Fresh VWAP loss below the 2-day AVWAP — the reject-short fired; sized in 5 + 15 and covered into −2σ. | the real trade |
Full day: 66 fills / 140 shares, net flat, realized +$110.86. The RTH P&L curve above starts after the pre-market probe was already clawed back; the −$48 drawdown is attempt 1, told in The Lesson.
cross_dn — fresh loss of session VWAP from above: close < VWAP AND prev_close ≥ prev_VWAP (the exact mirror of the long recapture’s cross_up).down_ctx — vwap_slope ≤ 0 OR close < 2-day AVWAP. The AVWAP arm is what lets a valid short fire while the session VWAP is still ramping up.value_zone — lb2 ≤ close ≤ VWAP (the VWAP→−1σ short value zone; don’t chase a close already below −1σ).sigma_below — lb3 < close, i.e. the −2σ target still sits below the entry (mirror of ub3 > close).avwap (the graded version), else tier vwap.cross_dn gate only fires on the bar that loses VWAP — re-tests below VWAP don’t re-arm it until price reclaims and loses again.avwap) + 0.25·ATR. A reclaim of the AVWAP/high invalidates the short.lb3). T2 = −4σ (lb5). Runner on the verified trail.| Time | Tier | Entry | Stop | T1 (−2σ) | R:R | Outcome |
|---|---|---|---|---|---|---|
| loading… | ||||||
The engine independently confirms the discipline story: the two earlier attempts (09:38 / 09:46) stopped; the 10:10 & 10:14 A+ signals — the ones below the 2-day AVWAP where James sized into 20 shares — hit T1.
lb3) — the mechanical target and the highest-probability touch.lb5) when the move extends.The vwap_reject_short was on the board all morning, and the tape kept paying it — a broken slot machine. What changed trade to trade was not the setup; it was the size. Three attempts at the same idea:
Stop paying tuition on 5-share pre-market probes and stop leaving the A+ on 1-share size. Wait for fresh + alpha-aligned (A+), then size it. The edge on this tape was never a better read — it was matching size to the grade. The engine agrees: the sub-A attempts (09:38 / 09:46) stopped; only the A+ window (10:10 / 10:14, below the 2-day AVWAP) hit T1.
ExecType∈{1,2} & LastShares>0 (never --exec-type fill — that drops partials); TransactTime UTC→ET corrected.(exit−entry)×shares estimate anywhere.$0.00.[low−0.10, high+0.10] — 0 outside.| Category | Grade | Notes |
|---|---|---|
| The read | A | vwap_reject_short identified correctly — engine-confirmed, tier avwap |
| Selectivity | C | 66 fills / many sub-A probes — the A+ was 1 of several presses |
| Sizing | D→A | 5-sh probe & 1-sh churn, then finally 20 sh on the A+ |
| Risk | B | pre-market probe −$48 was the only real damage; net day green |
| Management | C | scalped −2σ touches; left the VWAP-trailed runner on the table |
compute_2min() — real kti 1-min → 2-min, OHLC4 VWAP/σ, 09:30 anchor, bands ±0.5/1/2/4σ.add_avwap2() — the same OHLC4 spine run continuous from the prior session’s open.vwap_reject_short play in the TOD-scan engine (the short-side mirror of the VWAP pull-in) — the same engine as the J3P TOD matrix, no lookahead.trust_gate.py) — penny reconcile + fill-inside-bar — before anything was drawn.gen_playbook.py (data JSON → HTML); nothing hand-rolled.Frame: LAB, not cockpit. Nothing here is promoted to the live Pine / HUD. A-grade selectivity is a hypothesis to earn out-of-sample, not a proven edge.