scalpingclass.com • SMB • J3P LAB

EBAY

LEVEL BREAKOUT — LONG MOMO BURST
James Domino  •  7/1/2026 → 7/2 (2-day swing)
A+ · band_break archetype · level_breakout (long) · 2-day trend ride
Broke the 111.84 multi-day resistance shelf on the open, held every pullback above a rising VWAP, and rode the trend two sessions to 115.98 — a level-breakout momentum burst. One of my best trades: the discipline was HOLDING the runner.
LAB · UNVALIDATED A discipline + narrative playbook — not a validated $-edge. A-grade selectivity is the open hypothesis. The reconciled fills are real; the "this is repeatable edge" claim is not yet earned out-of-sample.
Prices and statistics are from reconciled fills and computed indicators; they do not reflect liquidity, slippage, or commissions. LAB material — not a validated edge.
scalpingclass.com • SMB
THE SETUP — LEVEL BREAKOUT, LONG
A+ — FRESH + ALIGNED The A+ on a breakout = fresh (a first clean break of a level that has held, not the 4th poke) + alpha-aligned (price above a rising VWAP, above the 2-day AVWAP, MAs stacking 5>20>50 up). When the break is fresh and everything points up, you get long and you hold. The edge here was never the entry tick — it was the discipline to ride the runner two sessions.

Anatomy — the momo-burst breakout

  • 1 — The level. EBAY had a multi-day resistance shelf at ~111.84 — the 6/30 high (111.84) capping the same zone that topped 6/24 (111.86). That shelf was the line.
  • 2 — Coil above a rising VWAP. Off the 7/1 open, price pushed up and coiled 111.4–111.9 (09:40–09:52) holding above the rising session VWAP, sitting on the +0.5σ band — a bull coil, not a fade.
  • 3 — The break. Price broke the shelf and the +σ bands (09:54→10:00: 111.89 → 112.06 → 112.25 → 112.42), expanding up and away from VWAP. The `band_break` archetype: coil above VWAP → break the upper band → ride.
  • 4 — Alignment made it fresh. Above rising VWAP, above the 2-day AVWAP (~110.85), MAs stacking 5>20>50 up (the J3P 2-min stack), breaking a level that had held twice. Alpha Trend env trace: Daily GREEN (bullish) at the 6/30 close ✓; 30-min cloud UNKNOWN at entry (indicator needs 86 completed 30-min bars — not available inside the kti window; first computable state 13:00 = GREEN). See the Environment panels. Family = level_breakout.
  • The ride: 111.84 shelf → 113.57 (7/1 high) → 115.98 (7/2 high). A clean two-session trend leg.

Honest read — what the engine actually saw

  • This was a quiet-tape breakout: RVOL sat below 1 all morning. So the J3P volume-gated breakout plays (band_break/open_drive need RVOL ≥ 1.2–1.5) did not fire — structurally a level-breakout, but without the volume signature the engine demands.
  • The engine's validated long that morning was the 09:48 vwap_pull_in (a VWAP recapture into the value zone) — and it won +1.21R. James's 09:37 entry front-ran that clean signal by ~10 minutes, buying the early strength directly.
  • The later trend_ride (11:20) stopped — the mid-day chop shook a fresh continuation entry.
  • Takeaway: the A+ here came from alignment + the level and, above all, the multi-day hold — not from a volume burst. A perfect example of a real winner that a strict volume-gated backtest would have skipped.
LAB material. Reconciled fills + computed indicators; no slippage/commission modeled.
scalpingclass.com • SMB
TECHNICAL ANALYSIS — DAILY
EBAY Daily — BB(20,2) + SMA 20/50/200 + 5-DMA env cloud (Alpha Trend Alignment V2) · trade date marked
RSI(10) + 10-MA + RSI-BB(10,2)
SMA(20) / Upper BB
Lower BB
SMA(50)
SMA(200)
5-DMA (env cloud)
RSI(10)
ENVIRONMENT · DAILY …
Shading = the indicator’s 5-DMA cloud states (green = rising 5-DMA & close above; yellow = mixed; red = falling & below), ported from the Pine source; drawn as full-height strips (lightweight-charts has no fill-between-series).
scalpingclass.com • SMB
INTERMEDIATE TIMEFRAME — 30-MIN
EBAY 30-min (RTH, ET; 1-min → 30-min) — VWAP + ±2σ + 5D(65) env cloud + key levels
Session VWAP
±2σ
5D = SMA(65) (env cloud)
PDH
ENVIRONMENT · 30-MIN …
Environment states are causal (COMPLETED bars only). UNKNOWN = the indicator’s inputs (SMA(65) + 21-bar slope, 86 bars of warm-up) aren’t available inside the kti data window — printed, never guessed.
scalpingclass.com • SMB
ANNOTATED REPLAY — EBAY 2-MIN (7/1 → 7/2)
EBAY 2-min — 7/1 & 7/2 RTH (the gap = the overnight hold) | ● Session VWAP   ● 2-Day AVWAP   ▬ 111.84 shelf (PDH)   ▲ my long entry   ▼ my scales/exits
Cumulative realized P&L (from reconciled fills) — +$152.53 on the 95 shares closed (5 still open)
Session VWAP
2-Day AVWAP
MA5
MA20
MA50
±2σ
long entry
scales / exits
RECONCILED · LIVE SWING Fills parsed from the FULL order log (ExecType∈{1,2}, TransactTime UTC→ET), reconciled to the matched-trades witness and cross-checked inside their 1-min bars. This is a multi-day swing that is still open (5 shares), so it is not a flat penny-closed round-trip — the realized reconciles; the trade isn't done.

My reconciled EBAY fills

When (ET)ActionSizePriceNote
7/1 09:37–09:38BUY (entry)100 sh~112.00Broke the 111.84 shelf on the open — the momo-burst entry.
7/1 10:49trim30 sh113.09First partial into strength; +$32.85 booked, 70 held overnight.
7/2 09:50–15:31distribute65 sh112.55 → 115.36Fed the runner out into the continuation; top tick 115.36.
—still open5 sh—The last runner piece — live swing.

Realized on the 95 closed shares: +$152.53 (matches the matched-trades witness +$152.50 to per-lot cent-rounding). Bought 100, sold 95, 5 still open. All 10 fills inside their 1-min bars.

LAB material. Reconciled fills + computed indicators; no slippage/commission modeled.
scalpingclass.com • SMB
THE TRIGGER — BAND-BREAK BAR MATH

band_break fire conditions (≤ current bar only)

  • Bull coil held above VWAP: over the last ~5 bars, ≥60% closed above VWAP (a coil, not a fade), and the range is contained (≤ 2·ATR — a range, not an already-running trend).
  • Fresh break: the prior bar was still inside the coil (so the break is new), and this bar closes above the coil high + 0.1·ATR, an up bar, in the upper band zone (close ≥ +0.5σ).
  • Context: close > VWAP and vwap_slope ≥ 0 (rising rail).
  • Volume gate: RVOL ≥ 1.2. — the one condition EBAY 7/1 did NOT meet (quiet tape); see the honest read.
  • Levels: entry = the break close; stop = below the coil low − 0.1·ATR; T1 = +2σ, T2 = +4σ partials, then RIDE the trend on the trail.
  • Concrete level here: the coil high ≈ the 111.84 multi-day shelf; the break of that shelf is the trigger regardless of the σ-band math.

EBAY 7/1 — the engine's long reads

TimePlayEntryOutcome
loading…

The volume-gated band_break/open_drive did not fire (RVOL < 1). The engine's validated long was the 09:48 vwap_pull_in (+1.21R) — James was already long by then. The lesson isn't the trigger; it's the hold.

No lookahead: every gate above uses only data up to the signal bar. Outcomes are a light 80/20 forward sim, not fills.
LAB material. Signal geometry uses only data up to the signal bar.
scalpingclass.com • SMB
TRADE MANAGEMENT — RIDE THE TREND

Track from VWAP for the trend

  • VWAP is the trend rail. On a breakout long, as long as price holds above the rising session VWAP (and above the 2-day AVWAP), the trend is intact — VWAP is where you ask "still my trade?"
  • +2σ partial: first scale at T1 (+2σ) into the initial thrust — take something off, de-risk the position.
  • +4σ partial: second scale at T2 (+4σ) when the move stretches.
  • Runner rides the trail. The last piece trails a rising higher-low (or the fast MA) — it does NOT come off at a fixed target. Trend-over = it loses VWAP / breaks the trailing higher-low.

What James actually did (and it worked)

  • Entry 100 @ 112 on the break; trimmed 30 @ 113.09 into 7/1 strength (the +2σ partial in spirit) — booked +$32.85, de-risked.
  • Held 70 overnight — the runner. The stock never lost the trend rail on 7/1's close weakness in a way that broke the swing.
  • Fed the runner out 7/2 from 112.55 up to 115.36 as it trended — distribution into strength, not a single fixed exit.
  • This is the textbook. Partial early, hold the runner on the trend, distribute into the continuation. The 2-day AVWAP never broke — the trend stayed on.
  • Nit: 5 shares still open — the very last of the runner hasn't been rung out. A defined trailing-stop rule (lose VWAP / the higher-low) would close the book cleanly.
LAB material. Reconciled fills + computed indicators; no slippage/commission modeled.
scalpingclass.com • SMB
MANAGEMENT TRACKING — ACTUAL vs J3P vs TMT
TMT · SHADOW-ONLY Specimen #1 of the Tiered MA-Trail (TMT) study — three management styles scored on the same entry. TMT trades nothing: promotion review at 20 specimens, full confidence at the 50+/family bar. Locked rule + decision log: Z:\trading\dev\tmt-tracker\TMT_SPEC.md.
+0.56R
Actual — blended thru 7/6
+0.56R
TMT — unrealized thru 7/6
5%
Actual exposure remaining
100%
TMT exposure remaining

Same scoreboard, opposite bets

As of the 7/6 close (113.605), Actual (+$152.53 realized on 95 sh, 5 sh still on) and TMT (all 100 sh held on max(floor, MA) stops) sit at an identical +0.56R — the actual average exit (113.60) matches the 7/6 close to the half-cent. The difference is entirely forward-looking: Actual keeps 5% of the position exposed to continuation, TMT keeps 100%. Whatever EBAY does next decides which style this specimen scores for — that is exactly the question the 20-specimen sample is built to answer.

Actual fills (order log × matched trades, reconciled)

Time (ET)SideQtyPriceR
7/1 09:37:57BUY50112.00—
7/1 09:38:00BUY50111.99avg 111.995
7/1 10:49:17SELL30113.09+0.38R
7/2 09:50:02SELL20112.55+0.19R
7/2 11:07:40SELL10113.12+0.39R
7/2 12:30:09SELL10113.79+0.63R
7/2 14:36:21SELL5115.32+1.16R
7/2 14:37:55SELL10115.11+1.09R
7/2 14:55:42SELL5115.36+1.17R
7/2 15:31:27SELL5114.95+1.03R

R basis: 1R = 111.995 − 109.13 = $2.865/sh. Open: 5 sh, live hard stop 111.32 (7/2 session low). Realized +$152.53.

TMT shadow state — 100 sh → 33 / 34 / 33

effective_stop(i) = max(109.13 floor, i-DMA) on the daily close; exit fills next open; the floor rests in the market intraday. SMA of daily closes; no ratchet, no buffer, no re-entry. Odd-lot remainder rides the 10 DMA.

DateCloseeff 5eff 10eff 20Exits
7/1111.280109.940109.130*109.130*none
7/2114.840111.308109.614109.138none
7/6113.605112.455110.151109.361none
TrancheShStatusR
5 DMA33OPEN+0.56R unrl.
10 DMA34OPEN+0.56R unrl.
20 DMA33OPEN+0.56R unrl.

* floor governs while the MA sits below 109.13. Generated by tmt-tracker/tmt_track.py from kti daily bars (+ Liberator patch for karb’s missing 6/9 bar).

J3P 60-20-20 σ-track (engine’s own 7/1 signals)

SignalEntryStopT1 (+2σ)T2 (+4σ)
#1 LONG111.52111.13111.89112.40
#2 LONG113.415113.00113.80115.29

60% off at T1, 20% at T2, 20% runner on the higher-TF higher-low trail; $50 risk basis. The engine trades its own entries — fills and outcomes render in the Annotated Replay above. Listed here as the third management yardstick per the TMT decision log.

Runner-rule fields (scorecard schema)

FieldThis trade
runner_stop_typeactual: htf_higher_low · shadow: tmt_5_10_20_dma
enum: prior_day_low · two_day_low · htf_higher_low · tmt_5_10_20_dma
ratchet_onlyclassic trails: true · TMT: false (pure MA tracking; the 109.13 floor is the catastrophe line)
breakeven_anchor_checkPASS no live stop within ±0.25% of entry — the debated 111.98 (entry − 1¢) was rejected in review; entry price is not structure
extension_pct_over_10dma+2.8% at entry (111.995 vs 10-DMA 108.919) — logged only; threshold to be fit from the sample, not guessed
CONVERGENCE NOTES — (1) The stop debated live on 7/6 (112.47) was the 5-DMA effective stop (112.455) to within $0.015: the discretionary instinct and the mechanical rule found the same level independently. (2) The 20-DMA crossed above the 109.13 floor on 7/2 by $0.008 — one session earlier than eyeballed; max(floor, MA) made the handoff invisible. (3) Jul 6’s 0.2x-RVOL red pause day required zero decisions from TMT — that is the point of the rule.
RE-GRADE 2026-07-06 — the 7/2 continuation entry, corrected (CORRECTION SPECIMEN). Previously called a volume-skip on Daily RVOL (0.64 EOD — “quiet tape”). The trigger-window metrics from Bella Vol MA say otherwise: same-bar RVOL 1.96 @ 10:08 and 2.27 @ 10:28 (trader-reported). Under the corrected gate spec (trigger-window RVOL gates entries; Daily RVOL is EOD classification ONLY — never an entry gate), the 7/2 entry re-grades to qualified-A. Family qualify also passed via the dollar-breakout branch (close−open +3.52) even though volume-vs-prior-day failed (0.82×) — the OR-gate exists for exactly this case. Logged as the correction specimen in tmt-tracker/data/signal_skip_log.csv; gate spec: TMT_SPEC.md § RVOL Gate Spec.
VERIFICATION UPDATE (computed 7/6 PM, feed-verified: Liberator ≡ karb, ratio 1.003). Recomputing the Bella same-bar formula from real 2-min buckets over the prior 20 sessions: 10:08 bar = 2.61 (trader-read 1.96) — a real, large burst (21.5K vs 8.2K avg) but ISOLATED (10:06 = 0.95, 10:10 = 0.45 → the “≥1.5 sustained, 2-of-3 bars” test FAILS); 10:28 bar = 1.51 (trader-read 2.27; no neighboring bar matches 2.27). And the primary gate: RVOL ToD = 0.21–0.22 at both reads — corroborated by this day's own overlay panel (.19x). Net: as the gate spec is currently written, the 7/2 entry does NOT qualify — and it made money anyway. So the correction specimen now cuts both ways: Daily RVOL wrongly skips this trade, and the provisional ToD thresholds skip it too; only the isolated same-bar burst flagged it. That is exactly what the 30-row threshold re-fit exists to resolve — this row is exhibit #1, not a rule change. Possible sources of the read-vs-computed gap: mid-bar label reads, extended-hours charts breaking the Pine's volume[i×barsPerDay] same-bar alignment (sparse premarket bars), or lenDays settings — take same-bar readings from RTH charts and verify against computed values before they gate size.

Auto-updated TMT state (daily post-close job)

TMT shadow track -- entry 2026-07-01 @ 111.995, floor 109.13 (1R = 2.87), 100 sh, tranches 5/10/20 DMA

DateClose5SMAeff5t510SMAeff10t1020SMAeff20t20
2026-07-01111.280109.940109.940open108.919109.130open108.837109.130open
2026-07-02114.840111.308111.308open109.614109.614open109.138109.138open
2026-07-06113.605112.455112.455open110.151110.151open109.361109.361open
TrancheSharesStatusSignalFill dateFillR
5 DMA33OPEN+0.56R (unrealized)
10 DMA34OPEN+0.56R (unrealized)
20 DMA33OPEN+0.56R (unrealized)

This block is rewritten by tmt_daily_update.py after each close (weekdays 16:30 ET). Tomorrow's exit thresholds land on the morning card in tmt-tracker/morning/.

LAB material. TMT is SHADOW-TRACKED (specimen #1 of a ≥20 gate): fill = next-day open, no slippage/commission. Actual fills reconciled from the FULL order log crossed against matched trades. Bars: kti price daily (karb) + Liberator daily_bars patch for 2026-06-09.
scalpingclass.com • SMB
THE LESSON — FRESH + ALIGNED, THEN HOLD

What made this an A+ — and why it's the one to replicate

This wasn't a loud, high-volume breakout — it was a quiet, perfectly-aligned one, and that's exactly why it's instructive. Everything pointed the same way and stayed that way for two sessions:

Fresh + aligned (the A+ checklist)

The discipline that made the money

The takeaway

When a breakout is fresh and everything is aligned, the job isn't to scalp it — it's to size it, partial it, and hold the runner on the trend rail. The honest asterisk: the volume wasn't there, so a strict volume-gated screen would have skipped this. The alignment + the discipline to hold is what turned a quiet breakout into one of the best trades. Replicate the process, not the luck.

LAB material. Reconciled fills + computed indicators; no slippage/commission modeled.
scalpingclass.com • SMB
SCORE CARD & DATA TRUST
+$152.53
Realized (95 sh closed)
+3.7%
111.84 → 115.98 leg
$0.03
Reconcile diff vs witness
A+
Setup grade (fresh+aligned)

Data-Trust provenance (honest, multi-day)

  • Fill parse: FULL order log, ExecType∈{1,2} & LastShares>0 (never --exec-type fill); TransactTime UTC→ET corrected.
  • Realized: average-cost trace across BOTH sessions — bought 100, sold 95, +$152.53.
  • Witness cross-check: matches the independent matched-trades EntryPNL (7/1 +32.70 + 7/2 +119.80 = +$152.50) to per-lot cent-rounding (diff $0.03).
  • Fill-in-bar: all 10 fills inside their 1-min karb bars — 0 outside.
  • Open position: net +5 matches the kti positions endpoint (CurrentQuantity 5). This is a live swing, not a flat book — no "penny-closed GATE PASS" is claimed; the realized reconciles, the trade isn't done.

Grade

CategoryGradeNotes
The readAfresh break of a real multi-day level, fully aligned
EntryB+bought the strength early — front-ran the clean band-break by ~10 min
SizingA100-share conviction on the break
ManagementA−partial + held the runner two sessions to 115+
Close-outB5 shares still open — no defined trailing-stop exit rule
Volume confirmationCquiet tape (RVOL < 1) — the honest asterisk on the breakout
LAB material. Reconciled fills + computed indicators; no slippage/commission modeled.
scalpingclass.com • SMB
HOW THIS WAS BUILT
  • Bars & bands reuse the verified J3P parity harness compute_2min() — real kti 1-min → 2-min, OHLC4 VWAP/σ, per-session 09:30 anchor, bands ±0.5/1/2/4σ, MA stack 5/20/50. The chart spans 7/1 + 7/2 (the gap = the overnight hold).
  • 2-day AVWAP from add_avwap2() — the OHLC4 spine run continuous across both sessions.
  • Classification from the TOD-scan engine's long plays (band_break / trend_ride / open_drive / vwap_pull_in) — reported honestly (the volume gate didn't fire; the validated long was the 09:48 pull-in).
  • Fills parsed with the DATA-TRUST order-log parser + reconciled against the matched-trades witness and the 1-min bars — before anything was drawn.
  • The page is the QQQ-3/31 playbook template + gen_playbook.py (data JSON → HTML); nothing hand-rolled. Sibling of the MU VWAP-Reject playbook.

Frame: LAB, not cockpit. Nothing here is promoted to the live Pine / HUD. A winning trade is evidence a process can work, not proof it has edge — that's earned out-of-sample.