J3P β€” James Three Plays Β· Home Hub

Single source of truth. Every tool findable, every play's current rules in one place. Consolidation only β€” no new analysis here.
πŸ“‡ Today's Cards β€” START HEREThe one page you trade from: pre-committed cards (entry/stop/target/size/the 3 orders) decided before the moment, so there's no risk left to accept when you're live. Everything else kept one click away.
πŸ§ͺ LAB β€” validated here (backtest / replay / sweep). The rules below are lab-validated on a ~2-week, 86-signal window. Directional, not deep-data-proven.
πŸ•ΉοΈ COCKPIT β€” the live TradingView Pine (J3P_v2.2). Nothing below is promoted to the cockpit yet β€” the Pine still runs the older exit model until the winners hold on deeper data.

β‘  Control Ledger β€” the current, verified rules

PlayDirEntryStopExit (verified)Best timeStatus
vwap_pull_inLONGVWAP recapture into the VWAPβ†’+1Οƒ value zone (+2Οƒ above entry)below the shakeout lowladder: 60% +2Οƒ / 20% +4Οƒ / 20% trailedOPEN / morningBROAD n=338: NO EDGE (βˆ’0.06R) β€” thin n=42 was overfit
vwap_reject_shortSHORTexact MIRROR of the pull-in: price LOSES session VWAP from above (cross-down) into the VWAPβ†’βˆ’1Οƒ value zone, below the 2-day anchored VWAP (multi-day resistance capping overhead)above the shakeout high / 2-day AVWAPβˆ’2Οƒ / βˆ’4Οƒ partials, runner trailsany (fires on the reject)NEW (MU-7/2 short mirror) Β· fires at his 10:10 20-share short Β· UNVALIDATED (no OOS yet)
backside_rollSHORTextended ~2–3Β·ATR above VWAP, then a rising 5-MA is lost (rollover)above the failed highcover 100% at VWAP (trail HURTS the fade)MID / POWER / afternoonBROAD n=376: OVERFIT (IS +0.16 β†’ OOS βˆ’0.12)
backsideSHORTβ‰₯2Β·ATR extended + a rejection candle (strict)above the failed highcover into VWAPβ€”THIN Β· n=2 Β· parked
open_driveLONGopen-window inside-bar breakout above VWAPunder the inside candle+2/+4Β·ATR measured moveOPEN onlyTHIN Β· n=1 Β· needs looser trigger + data
band_breakLONGbull coil held above VWAP β†’ fresh up-bar breaks the coil high in the upper band zone (β‰₯+0.5Οƒ)below the coil low+2Οƒ / +4Οƒ partials, then RIDE the trend on the trailtrend days (any window)NEW (AMZN-7/1 play) Β· BROAD n=380: βˆ’0.07R β€” catches trend days, breakouts mostly fail
v_bottomLONGflush into the lower band (≀ βˆ’1Οƒ) β†’ 5-SMA slope turns up (a V) β†’ reclaims the fast MA (quiet bar)below the V lowrevert to VWAP β†’ +1Οƒ, runner trailsopen flush / any reversalNEW (AMZN-7/1 open) Β· BROAD: ~βˆ’0.05R, IS≀0 β†’ OOS noise β€” no edge on ~2wk
trend_rideLONGup-regime (5>20>50, VWAP rising) β†’ pullback to the 5-SMA β†’ GO (break the pullback high)below the pullback low+2Οƒ / +4Οƒ partials, runner rides the trailtrend days Β· mid/powerNEW (AMZN-7/1 trend) Β· BROAD n=358: βˆ’0.09β†’βˆ’0.13 OOS β€” no edge on ~2wk
+4Οƒ (T2): the optimal fixed scale-out uses 0% at T2 (a pre-committed T2 slice underperforms riding to the EVENT). But +4Οƒ is NOT a phantom β€” verified frozen at the entry bar (not moving), it tags ~33% of pull-ins and is the best exit when reached (+5.2R avg). It loses in the fixed sweep because a fixed % can't tell the 33% taggers from the 67% non-taggers. Conditional +4Οƒ tested 2026-07-02 & rejected: "+4Οƒ-if-first, else ride" = +0.207R and targeting +4Οƒ = βˆ’0.037R β€” both lose to riding the event (+0.251R). +4Οƒ doesn't help in any form; ride to the rollover. Sizing on all plays = shares at your Risk $ (round(Risk$ / |entryβˆ’stop|)).

🟒 β‘‘ Use Daily

β–Ά Interactive Replays β€” type a ticker + dateThe main tool: any liquid ticker on any recent day β†’ annotated 2-min chart with every play firing (entry/stop/target + why + interactive scale-out).
NVDA 2026-07-02NVDA 2026-07-01MU 2026-07-02EBAY_2026-07-01_06 auditEBAY 2026-07-01AMZN 2026-07-01
Request another (ticker, date) β†’ runs py j3p_replay.py SYMBOL YYYY-MM-DD (HOME-side, kti ~14-day window).
⭐ NEW β€” first "me vs strategy" adherence case (2026-07-02, trust-gate pilot). The MU and NVDA replays now overlay your REAL fills β€” pulled from the Gr8Trade order log and run through the Data-Trust STEP-0 gate: reconciled to the matched-trades P&L to the penny + every fill inside its 1-min bar (no silent estimate). MU = plan-right, execution-timid (your 10:10 20-share short WAS the new vwap_reject_short signal; you covered it in minutes). NVDA = no setup, idea-traded (zero vwap_reject_short signals; you traded it hardest of the day, both ways).
πŸ“˜ Full playbook β†’ MU Β· VWAP Reject β€” Short Continuation β€” the QQQ-template writeup of this exact setup: anatomy (broke session VWAP β†’ retraced UP into the 2-day AVWAP β†’ rejected β†’ fresh cross back below = trigger), the exact vwap_reject_short bar-math, management (track from VWAP, βˆ’2Οƒ/βˆ’4Οƒ partials, runner covers on VWAP reclaim), and the sizing lesson β€” with the annotated 2-min replay + reconciled fills. Upd 7/6: both playbooks now carry the full library spine β€” Daily BB/RSI + 30-min env panels with the Alpha-Trend cloud; environment INDICATOR-TRACED (MU: Daily RED / 30-min MIXED at entry; EBAY: Daily GREEN / 30-min UNKNOWN-at-entry).
πŸ“˜ Full playbook β†’ EBAY Β· Level Breakout β€” Long Momo Burst β€” a winning LONG "best trade" writeup (band_break archetype, family level_breakout): broke the 111.84 multi-day shelf, held above a rising VWAP with MAs stacked 5>20>50, and rode the trend TWO sessions to 115.98 (+$152.53 reconciled, 5 sh still open = a live swing). Honest read: quiet tape (RVOL<1) so the volume-gated band_break did NOT fire β€” the engine's validated long was the 09:48 pull-in (+1.21R); the A+ was alignment + the hold discipline. Grades DISCIPLINE, not dollars β€” nothing promoted.
Data-Trust Spec ⭐ (foundation for scoring)Before any me-vs-strategies scoring: provenance Β· reconcile-vs-independent-witness Β· drill Β· fail-loud, per fact type (fills/signals/grades) + strategy versioning + scorecard audit. Reuses this replay's provenance/drill model.🎯 Setup Analog Target ⭐ (think-like-me targets)Replaces 'revert to the 200-DMA.' Plug a live setup (direction / entry vs VWAP / Οƒ) β†’ the target = what setups LIKE IT actually ran, forward-measured in Οƒ so it works on ANY stock (ALOY with no 200-DMA exactly like MU). Reveals the reclaim entry is a ~2-hour trend trade (long +2.2Οƒ/77%, short +3.4Οƒ/78%), not a 3-min scalp; extended entries = burst spent (the Lynch line). LAB, cached-subset.🎯 Bracket Planner ⭐ (Step 3 tool)Type the trade the moment you're in (symbol/dir/entry/stop/type/target/risk) β†’ it outputs the exact 3 orders to place: bulk target order (+R), the % runner trail, and the stopβ†’breakeven trigger, with share size + R-ladder. Trend-vs-reversion aware. Mechanizes bracket-and-leave so you can't give the winner back. Client-side only β€” does NOT route orders (live hotkey wiring pending firm sign-off).πŸƒ Cheat Sheet ⭐ (glance before you trade)One screen, four picture-cards: the context tells you the plan. Almost no words β€” under a falling 200-day=trim into resistance Β· above a rising 200-day=buy the reclaim Β· a parabola=staircase out Β· an intraday trend=ride the EMA.πŸ“Š Earnings Pipeline Replay (the deployed system, replayed)James: see it find the setups and execute BEFORE going live. The exact live funnel replayed 6/23-7/10: 13 fired, 46% win, avgR +0.35, +4.6R; A-grades (gap>=5%) = 5 fired / 60% win / +0.77 avgR. Per-day cards.🎯 Exit Study ⭐ (how to close a trend trade)Same trades, only the EXIT changes: cap-at-band vs ladder vs ride-the-EMA vs partial-then-ride vs the daily-MA swing. Verdict: RIDE the trend (+$1,471/100 fresh); partials cost profit (buy only smoothness); the swing 'exit' is a lottery that fails OOS. Plain-English report + recommendation.Exit Param-Surface (thin sample)Earlier 3-name exit sweep β€” superseded by the OOS verdict; the thin sample was overfit.Experiment Log ⭐ (what works / what didn't)Living lab notebook β€” every change we've tested toward profitability, what worked and what didn't. The running record.🧭 The Daily Flow β€” map ⭐ (how the pieces connect)The one path as a map: WHAT to trade (cockpit) β†’ HOW the play works β†’ MANAGE it (bracket-and-leave + the trend-vs-reversion exit fork) β†’ GRADE it (adherence). Names the canonical page per step; Step 3 (management) is where the money leaks. Day-to-day you live in Today's Cards β€” open this when you want the whole map.Exit Ladder β€” OOS Verdict ⭐Your 60/20/20 ladder graded HONESTLY on 28 names with an out-of-sample hold-out. Read this first β€” no edge held out-of-sample.πŸ”΄ LIVE Signals (today) ⭐Auto-refreshing morning scanner: today's SIP names, the OOS-validated plays only (revbreak / trend_ride / vwap_reject_short), morning + daily-trend gates applied. Tells you when a working play triggers. LAB β€” trade small by hand.MU Management / R:R Lab ⭐ (n=1)The MU 10:10 A+ short managed 4 ways on the REAL path β€” ACTUAL (+0.06R, rung out) vs playbook-rule held (+1.54R) vs ladder sweep vs MFE ceiling. Reuses ladder_blended_R; Python↔JS reconciles exactly. Quantifies the MU playbook's flagged management gap. Now includes MA-trail runners (ma5/ma20/ma50, switchable on the chart) + all automated J3P shorts on the tape. LAB/n=1 β€” discipline, not a validated number.🧩 What Makes a Stock Pay ⭐ (nuances)Dead-simple green-light/red-light: which in-play characteristics make the setups win, from 3,520 real setups (OOS-gated). Headline: get paid catching a move that's still coming, not one that already happened β€” quiet-open + wild-day pays 2x; chasing a big gap / already-ripped / over-extended name pays a quarter.Playbook #1 β€” VWAP Reject Short ⭐ (held OOS)The one setup that survived out-of-sample (+0.19/+0.11) = your MU-7/2 short. Full spec: SIP + morning + daily-aligned gates, trigger, entry/stop/targets, discipline don'ts. LAB β€” trade small to confirm.The Plays ⭐ (charts + specs + code)Every play documented: click one to expand a labeled chart of the setup + the exact specs + the real code. Ranked by out-of-sample strength. The place to verify each setup is coded right.🎬 Replay Practice ⭐ (muscle memory)Reveal a real trade one day at a time; make YOUR call at each decision (enter/trim/add/stop/exit); get scored vs the playbook. 3 trades, 3 lessons: IBIT defense (trim at the ceiling, +4.47R) Β· JNJ offense (buy the pullback, ride it, +8.51R) Β· BTDR discipline (staircase saves a round-trip, +0.69R vs -0.35R held).πŸ€– Robot Report ⭐ (what if a robot traded today?)Every signal, $100 risk each, exits by the rules β€” three robots compared: take-everything vs $500-daily-stop vs short-only. Click any trade for its chart (VWAP, bands, MAs, in/out marked). Rebuilds daily at 4:15.πŸ—ΊοΈ Runbook ⭐ (a day, end to end)The whole loop on one screen: 8:00 watcher β†’ 8:45 SIP+earnings β†’ 9:25 scanner β†’ alerts (what YOU do) β†’ manage by the card β†’ 4:15 scorecard β†’ Trade Check. Week-one rule: A-grades only, tiny.Play SetupsPlain-English entry / stop / target spec for every play.90-Day SIP Backtest ⭐⭐ (holds OOS!)The big one: playing daily stocks-in-play (not a fixed list) makes the edge SURVIVE out-of-sample for the first time. Selective stack TRAIN +0.36R β†’ TEST +0.12R; revbreak / trend_ride / vwap_reject_short hold strongly. LAB, trade small first.🎬 SIP Demo (a real earnings morning)The SIP page as it looks on a real earnings day (7/10: DAL/PEP/ABBV...): earnings badges, beat/miss, pop-up details, the also-reported section.πŸŽ›οΈ COCKPIT β€” Today's Decisions ⭐⭐ (pick names here)The one trader surface: each in-play name gets an opinionated GO / WAIT / SKIP with the reasons shown (real news, right-kind-of-day for the direction, trend agrees, not chasing) β€” each check cites an OOS-validated number. Reads live from the SIP pipeline. Click a symbol for its chart. Everything else is a drill-down from here.Swing DMA-Trail Management ⭐ (topping shorts)The SWING/DAILY companion: how to stay with + de-risk a multi-day short as it reverts to a DMA and keeps falling. Case = AMZN 6/1 (lost 20-DMA β†’ imploded to 227; a 20-DMA trail held it for +7.5%, you bailed early). Applied LIVE to MU (short from 7/2, lost 10/20-DMA 7/1, +5.8% riding toward the 50-DMA @853 = +18%; trail the falling 20-DMA). Honest AMD read: it RECLAIMED its 20-DMA = bullish, NOT this setup. Re-run build_swing.py to track MU forward.πŸ“‡ TODAY'S CARDS ⭐⭐⭐ (START HERE)The one page you trade from. Each in-play name gets a pre-committed CARD β€” entry / stop / target / gradeβ†’size / the 3 bracket orders β€” decided before the moment so there's no risk left to accept when you're live. Live AMD + MU cards, the blank-card link (Bracket Planner), and everything else kept one click away grouped by when you need it. This is the front door; the flow map and the rest fold underneath.Time-of-Day Γ— Trade-TypeWhich play works in which session window (open / mid / power).βœ… Trade Check ⭐ (you vs the plays)Your REAL fills graded end-of-day: SIP name? with the trend? morning window? 7/2 proof β€” the 2 SIP names made +$132 (MU short +$111, fully on-plan), the 13 off-book names lost βˆ’$165 (NVDA long, off-book+against-trend, βˆ’$87). Accountability loop.πŸ“ˆ Trending-Stock Playbook ⭐ (the 200-DMA lens)How to enter + manage a stock starting to trend: context (price vs the 200-day + slope) β†’ grade A/B/C/Pass β†’ size by grade (20-30%/10-15%/pilot/0) β†’ MANDATORY trim at the 200-DMA test (+5.22R vs +0.93R on IBIT). IBIT worked phase-by-phase from the MTF-VP autopsy + a live board (JNJ, BBIO, BTDR, ALOY) with clickable charts.πŸ”¬ Trust Report ⭐ (verify before you build)Every load-bearing claim adversarially re-checked from source: 2 CONFIRMED / 4 WEAKENED / 0 REFUTED. Edge is REAL but tail-driven and smaller than advertised (true OOS +0.17R not +0.12); only the short is durable; open_drive unproven; LIVE scanner watches different stocks than the backtest proved (1-4/15 overlap).πŸ”” Management Watcher (Phase A) ⭐Read-only daily guardrail: watchlist names (JNJ/BBIO/BTDR/ALOY) -> pullback-zone entry alerts; open positions (positions.json) -> 200-day TRIM-NOW + staircase 5/10/20-day sell-1/3 alerts. Alerts only, never executes. Reminds you of the plan.πŸ”” Management Watcher β€” Spec (review)The review-first spec for the guardrail watcher: 4 alerts (pullback-zone, ceiling/200-DMA TRIM-NOW, staircase-leg sell-1/3, selected-management-line) + a staged build plan (daily read-only β†’ plan-per-position β†’ hotkey enforcement). Nothing built until reviewed.

πŸ”΅ β‘’ Reference

Backtest SummaryThe original edge-read summary (older; superseded by the surface/TOD reads).Parity Viewer (2-min)OHLC4 band + signal parity vs the Pine β€” the verification proof.Runs RegistryEvery backtest run logged (variations + results).State of the Project ⭐Where we started Β· what's done Β· where we are (no edge holds OOS yet) Β· what's left Β· the one decision (the Polygon key). Includes a scanned Links Index + orphan flags.πŸ“‹ Trades By Day (backtest)The stocks TRADED each day in the 90-day backtest: per-day universe watched + the tradeable strategy trades (stock, play, P&L). 2,183 simulated trades on 110 stocks. Downloadable CSV. NOT real fills β€” not in the trading DB.

πŸ†• Unclassified β€” review

These J3P pages exist in /system/ but aren't in the hub registry yet β€” classify them in j3p_hub.py:
J3P_ALERTS_CHEAT.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_EXECUTION_REPORT.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_LIVE_DRILL.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_REPLAY_BTDR.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_REPLAY_IBIT.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_REPLAY_INTRA.html⚠ new / unclassified β€” added since the hub registry was last updated.J3P_REPLAY_JNJ.html⚠ new / unclassified β€” added since the hub registry was last updated.
βšͺ β‘£ Archive β€” superseded v2.2 / v2.3 / v2.4 iteration pages (10)
Static SVG charts β€” Non-interactive chart dump (2.8 MB).Backtest β€” SIP edge β€” SIP-names edge read.Backtest β€” market bucket β€” Fixed market-stock bucket read.Static SVG charts (market) β€” Non-interactive chart dump (3.8 MB).5-min interactive viewer β€” Superseded by the 2-min replays + parity viewer.Backtest β€” WIDE (v2.2/2.3) β€” Wide-universe edge read.v2.4 experiment plan β€” The regime-filter options doc.SIP vs bucket β€” In-play-daily vs static-bucket comparison.Today page (2026-06-29) β€” One-off dated page; superseded by the replays.v2.4 filter matrix β€” Regime-filter results grid.

🧭 β‘€ Roadmap β€” what's next (not built yet)

  1. Broaden the name set to firm up the optima (more liquid tickers through the TOD scan + exit sweep).
  2. Port the winners to the Pine (cockpit): pull-in β†’ event-exit, fade β†’ 100%-at-VWAP, drop the +4Οƒ leg.
  3. Deep-data adapter (Polygon / Databento) β€” parked, needs an API key; the real unlock for trusting magnitudes past the kti ~14-day window.
Generated by j3p_hub.py (self-updating: re-scans /system/ each run). ← /system/ hub