CAPTURE-AND-LIQUIFY SWING ENGINE

Long-only · unhedged · fully-liquidated rotation swings · dual-engine (tech-internal + macro regime) · v2.0

LIVE-ANCHORED · 26 JUN 2026

0 · Capital Map & Live Anchors

SPAXX buffer

$261,068.54

Core cap / cycle

$200,000

Hard reserve (locked)

$61,068.54

Deployable pool

$200,000

buffer − reserve

Underlying exposure reality — why this engine exists

Sleeve	Current exposure	Names	Engine role
Tech physical infra	$112,000	NVDA · VRT · AMD · AVGO · LRCX · ANET	Already overweight. Numerator side (the leg this engine fades, never adds to).
Enterprise software	$3,300	ORCL only	Structural gap. Denominator side — the leg the engine accumulates on extension, then liquidates.
Defensive (macro)	$4,700	XLV (Individual TOD)	Baseline. Logical accumulation target for the macro engine's denominator side.

Live ratio anchors (self-calibrating — no absolute level is ever hardcoded)

SMH / IGV · tech-internal

≈ 7.3

SMH ~$620 · IGV ~$85

QQQ / XLV · macro regime

≈ 4.6

QQQ ~$713 · XLV ~$155.6

QQQ / XLP · macro alt

≈ 8.4

QQQ ~$713 · XLP ~$84.6

Capital conflict — resolve before running both engines. Core cap $200k already consumes the entire $200k deployable pool. The macro engine's $100k is additive on paper → a combined $300k draw that breaches the $61,068.54 hard reserve. The engines must share one pool. Pick one: (A) single shared $200k ceiling, macro draws only from core's undeployed tranches; (B) hard split, e.g. $130k core / $70k macro; (C) strict mutual exclusion — only one engine live at a time. Default below assumes (A): total deployed across both engines ≤ $200k, reserve untouched.

1 · Ratio Construction

Identical mechanics on either pair. The Pine engine (§6) builds the ratio internally via request.security() so it runs on any chart.

Build A — synthetic symbol

TradingView: type SMH/IGV or QQQ/XLV directly in the symbol box.
TrendSpider: comparison chart → type = Ratio → numerator ÷ denominator.
All studies run on the ratio series, not either leg.

Build B — computed (engine default)

Chart anything; pull both legs, divide. Decouples signal from the visible symbol.
Lets you watch the laggard ETF's own tape for execution while the ratio drives signals.

Overextension boundaries (computed on the ratio)

Layer	Setting	Function
Basis	SMA 20	rolling fair value; the full-liquidation line (Z = 0)
Dispersion	StdDev 20	σ unit; bands at 2.0 / 2.5 / 3.0 / 3.5σ drive the deploy ladder
Normalizer	Z = (ratio − basis)/σ	single scalar; every entry and exit keys off it
Partial line	±1.0σ	the 50% unwind trigger

Validate the σ grid against your own TrendSpider/TV backtest of each ratio's tag-frequency. The 2.0 → 3.5σ grid is the deploy schedule; confirm the historical ratio actually visits those bands at a cadence you want to trade before committing the cap.

2 · Capture-and-Liquify Swing Rulebook

Deploy flat into the laggard as the ratio stretches; liquidate the entire position back to SPAXX in two mechanical steps as it reverts. No averaging logic, no hold-through, no discretionary trims.

Deploy ladder — FLAT sizing, core engine

Tranche	Trigger (daily close)	Size	Cumulative	Action
T1	ratio crosses ±2.0σ	$50,000	$50,000	BUY laggard
T2	ratio crosses ±2.5σ	$50,000	$100,000	BUY laggard
T3	ratio crosses ±3.0σ	$50,000	$150,000	BUY laggard
T4	ratio crosses ±3.5σ	$50,000	$200,000	BUY laggard · cap hit

Default direction (your config): the ratio sits high when semis are rich and software is the discount → tranches buy IGV at the +σ bands. The engine is symmetric and will instead buy the numerator at −σ bands if you flip the direction input — but with a $112k hardware overweight, the −σ (buy-SMH) side stays disabled for the core engine.

Liquidation ladder — two-step, full exit to SPAXX

Step	Trigger (daily close)	Action	Destination
L1	ratio mean-reverts to ±1.0σ	SELL 50% of total position	lock core profit, clear half the risk
L2	ratio hits 20-day SMA basis (Z = 0)	LIQUIDATE remaining 50%	→ SPAXX. Cycle complete, flat, await next extreme

SWING LIFECYCLE │ ├─ ratio stretches to +2.0σ ─▶ T1 $50k IGV ├─ +2.5σ ─▶ T2 $50k ├─ +3.0σ ─▶ T3 $50k ├─ +3.5σ ─▶ T4 $50k (cap $200k) │ │ … ratio peaks, institutional flow rotates back into software, spread compresses … │ ├─ reverts to +1.0σ ─▶ L1: SELL 50% (de-risk, bank the move) └─ reverts to basis / Z=0 ─▶ L2: LIQUIDATE remaining 50% → SPAXX └─ position flat · engine re-arms for the next excursion

3 · 24-Hour Execution Protocol

Signal on the confirmed RTH close; fill on the overnight gap. You are the resting bid into institutional flow, not the chaser at the cash open.

Window	Engine action	Instrument
16:00 ET close	Ratio bar confirms; tranche/liquidation alert fires (alerts set "Once Per Bar Close"). Z-score and stage read off the HUD.	ratio
Post 16:00 / pre 09:30	Stage the armed tranche as extended-hours limit / GTC buy at the band-implied laggard price. Overnight gap fills at the level — no live screen required.	laggard ETF in EXT, or its liquid megacap components
Intraday flush	Resting GTC limits at the deeper σ bands absorb the washout.	laggard ETF
Reversion	L1/L2 sell limits pre-staged at the ±1.0σ and basis-implied prices; liquidation executes into the bounce.	laggard ETF

Binary catalyst circuit breaker — intact, non-negotiable. All resting / extended-hours orders are cancelled for any session carrying a top-component earnings print (NVDA, AVGO, MU, MSFT, ORCL, CRM, NOW, ADBE for the tech engine; sector-mover prints + JNJ/UNH/LLY, PG/COST for the macro legs), FOMC, or a CPI/PCE release. Through a binary the engine drops to alert-only; execution is manual after the print clears — review pre-market futures, the sector ETF tape, and the laggard's own pre-market behavior before placing. The 24-hour automation governs non-event drift only.

4 · Operational Guardrails

Velocity filter

If the laggard's standalone volume on the trigger bar is below its 20-day average, the extension is moving on thin participation — institutions aren't behind it yet. Halve the next tranche ($50k → $25k). Restore full size when volume reclaims its 20-day average. The engine flags this live in the HUD and fires a VELOCITY FILTER alert.

Hard time-stop

If a swing stays armed for more than 25 trading days without reaching Z = 0, the reversion thesis has failed — the spread is consolidating or trending, not snapping back. Liquidate at market. This is the regime backstop (see below). The engine counts SWING AGE in the HUD and fires a HARD TIME-STOP alert on the 26th bar.

Unhedged long-only risk — own it explicitly. This is not a market-neutral spread. You hold full beta on the laggard ETF; the ratio only times entry/exit. If the ratio extends past +3.5σ and keeps going (structural regime, not oscillation), you are long $200k of a falling instrument until the 25-day time-stop liquidates. Max adverse outcome = the drawdown on $200k of the laggard over ≤25 trading days. The time-stop caps duration, not depth — there is no σ-band that suppresses deployment, because the schedule is deterministic by design. Size the cap to a drawdown you will accept on that worst case.

Regime context as a manual overlay (not a gate): the current SMH/IGV spread is structurally elevated — SMH ≈ +138% / IGV ≈ +20% trailing year, IGV pinned near 52-week lows on margin-compression fear. A reversion swing here is betting against a strong one-year trend. The 25-day time-stop is what protects you if that trend doesn't break. Watch the 200-day ratio slope; if it's steepening with the extreme, expect the time-stop to do the work.

5 · Dual-Engine Map

Same statistical core, same two-step liquidation, two independent pendulums on different time signatures. One Pine script (§6) drives both — only the symbol inputs and the sizing config change.

Engine 1 · Tech-Internal Pendulum

Ratio	SMH / IGV
Live anchor	≈ 7.3
Cap	$200,000
Tranche	$50,000 × 4
Bands	2.0 / 2.5 / 3.0 / 3.5σ
Buys	IGV (software laggard)
Liquidation	50% @ ±1σ → rest @ Z=0

Captures the semis ⇄ software flush you documented — memory-led hardware washouts vs broad SaaS green days.

Engine 2 · Macro Regime Pendulum

Ratio	QQQ / XLV (alt: QQQ / XLP)
Live anchor	≈ 4.6 (XLP: 8.4)
Cap	$100,000 *
Tranche	$25,000 × 4
Bands	2.0 / 2.5 / 3.0 / 3.5σ
Buys	XLV / XLP (defensive laggard)
Liquidation	50% @ ±1σ → rest @ Z=0

Risk-on vs risk-off institutional flow. Baseline $4.7k XLV already held → denominator accumulation is additive to an existing position. * shares the $200k pool — see §0 conflict.

Macro direction logic: QQQ/XLV high = tech euphoric, defensives are the discount → buy XLV at +σ. QQQ/XLV low = risk-off overshoot, tech is the discount → buy QQQ at −σ (enable the "Lower → Long Numerator" or "Both" direction input). Unlike the core engine, both sides are legitimate here since you're not structurally overweight either leg.

6 · Pine v6 Engine — Deployment

Full copy-pasteable script in the companion .pine file. One indicator, both engines.

To run…	Numerator input	Denominator input	Tranche $ / Cap $ (echo)	Direction
Tech-internal	SMH	IGV	50000 / 200000	Upper → Long Denom
Macro (XLV)	QQQ	XLV	25000 / 100000	Both (symmetric)
Macro (XLP)	QQQ	XLP	25000 / 100000	Both (symmetric)

// core math — self-calibrating, no hardcoded level
ratio = num / den
basis = ta.sma(ratio, 20)
dev   = ta.stdev(ratio, 20)
z     = (ratio - basis) / dev

// deploy: flat tranches on band crosses (direction-gated)
dep1 = ta.crossover(ratio, basis + 2.0*dev)   // "SWING BUY ARMED - DEPLOY TRANCHE 1"
dep4 = ta.crossover(ratio, basis + 3.5*dev)   // "...TRANCHE 4 - cap reached"

// two-step liquidation
partial = ta.crossunder(math.abs(z), 1.0)        // "PARTIAL PROFIT TAKE: UNWIND 50%"
full    = ta.cross(z, 0)                       // "STRATEGY COMPLETE: LIQUIDATE REMAINING"

// guardrails
lowVol   = tradedVol < ta.sma(tradedVol, 20)     // halve next tranche
timeStop = swingAge > 25                       // liquidate at market

Repaint discipline: request.security shows the developing value intrabar. Signals are valid only on the confirmed daily close — set every alert to "Once Per Bar Close." On-chart shapes settle at the close; do not act on intrabar flips. The HUD table (top-right) shows live pair, Z, deploy stage, position, velocity, and swing age.

Anchored to live prints 26 Jun 2026: SMH ~$620 / IGV ~$85 → 7.3 · QQQ ~$713 / XLV ~$155.6 → 4.6 / XLP ~$84.6 → 8.4. Intraday — re-pull before sizing. σ grid, the $200k/$100k caps, and tranche sizes are execution parameters for you to validate against your own ratio backtests; this is a system you operate, not an allocation instruction. Confirm each ratio's historical band-tag frequency and your accepted worst-case drawdown before committing capital.