PANDORA Data: close of 19/11
Catalogue

Every condition tested in the lab (359).

✅ validated out of sample (16) · 🟡 measured once, without correction for the number of conditions tested (57) · ⚪ in research (188) · ❌ tested without edge (76). « on PANDORA » = computed every day on the site (see Signals). Open a row for its rule and measure.

A1Bullish (bearish) primary trend⚪

Rule · the last 3 daily swing lows are rising (falling)

Role · context / bias

Measure · ⚪

A2Log fib discount✅

Rule · log depth of price within the leg (H1, H4 or D) between 0.5 and 0.786: (ln H − ln P)/(ln H − ln L)

Role · location

Measure · ✅ foundation (alone = 0) · detector v1.8: at 10 d lows, does not pinpoint the pivot (lift 1.05) → stays a location, never a trigger

A3Nested fibs⚪

Rule · true discount on two timeframes (e.g. H1 and D) at the same price ±0.25σ

Role · location

Measure · ⚪ · see A21 (multi-anchor cluster ❌ with price alone)

A4Below prior-day high (long) / above prior-day low (short)✅

Rule · long: entry price < PDH · short: entry price > PDL (absolute rule, veto K7)

Role · location

Measure · ✅ (BASE rule)

A5Marked liquidity level⚪

Rule · double bottom (2 lows within ±0.3% at least 5 sessions or 20 H1 bars apart), PDL, overnight / pre-market / Asia low, IB low, swing low

Role · location

Measure · ⚪

A6Round number🟡

Rule · multiple of 5 (< 500), 10 (< 2000) or 50 within ±0.3% of the level

Role · location

Measure · 🟡 (+0.15 R, n=98, positive in the 2 periods)

A7V-band⚪

Rule · wick that launched a clean V on 1H / 15 min, not retested

Role · location

Measure · ⚪

A8IB golden zone⚪

Rule · 61.8-79% of the 09:30-10:00 range, drawn from the side formed first

Role · location

Measure · ⚪

A9Same-price cluster⚪

Rule · at least 2 items among {fib, liquidity level, GEX node, VEX node, DP, day/week/month σ band} within ±0.5σ of each other

Role · location

Measure · ⚪

A10First test of the level❌

Rule · no touch of the level since it was created (or since the morning)

Role · location

Measure · ❌ (tap #1 ≈ tap #2 on fibs) · v2.13 (guide 37) on SPY / QQQ / SPX gamma nodes, 8,380 touches: tap 1 47%, tap 2 48%, tap 3 50%, 4+ 44% → no decay (the community 66% / 33% rule does not hold)

A11Participation (relative volume)⚪

Rule · RVOL = day volume / 20-day median. Constructive for a long: pullback to the level with average RVOL < 0.8 after a reversal day with RVOL ≥ 1.2 (CRCL)

Role · context / bias

Measure · ⚪ · detector v1.8: ❌ (lift 0.67 at lows)

A12Worn-out zone⚪

Rule · the zone has been tested ≥ 5 times since it was created → reference mark, no longer support or resistance (TALON CRCL)

Role · context / bias + veto

Measure · ⚪ · v2.13: 5th tap and beyond 43% vs 48% (taps 2-4) → slight erosion, not a break

A13Old range retested (polarity)⚪

Rule · old daily consolidation range (≥ 5 sessions, width ≤ 4%) broken to the upside ≥ 3 months ago, retested from above for the 1st time (XLU)

Role · location

Measure · ⚪

A14Volume absorption⚪

Rule · RVOL ≥ 1.5 for ≥ 3 consecutive sessions at the level while closes stay within ±1σ daily (XLU 28/09-02/10)

Role · location + trigger

Measure · ⚪

A15Wave 2 at 0.5-0.66⚪

Rule · first pullback after a major low: retraces 0.5-0.66 log of the 1st impulse (≥ +20%) without breaking the origin (PLTR 28/07)

Role · location

Measure · ⚪ · Elliott 21/10/2026 , 538 assets 2018-2026, 0 credits): wave 2 / wave 1 median 0.61 (k = 4) / 0.59 (k = 8), 0.618-0.786 zone the most frequent but only 22-25% of cases; buying the 0.618-0.786 pullback of the 1st swing after a 60 d low +0.38R vs +0.32R buying right away → the pullback adds almost nothing

A16Strong-trend pullback (wave 4)⚪

Rule · pullback of 0.236-0.382 log of the last impulse, RVOL < 0.8 during the pullback, stable IV (PLTR 10/09)

Role · location

Measure · ⚪ · v2.8: daily version « bullish regime + −0.5σ month + 3 d RVOL < 0.8 » ❌ (−0.57R, guide 33) · Elliott 21/10/2026: wave 4 / wave 3 median 0.37 / 0.35, 0.236-0.382 zone the most frequent (42-44%) then 0.382-0.5 (24-29%)

A17Spring below a range (from a few months to several years)⚪

Rule · break of the bottom of a ≥ 3-month range on RVOL ≥ 5 with DP prints ≥ 10% ADV the same day, then a base of ≥ 5 sessions at the range bottom with RVOL ≤ 1.3, then a recovery close above the range bottom (GME, 2-year range)

Role · location + trigger

Measure · ⚪ · daily version: 60 cases, lift 1.18; + RVOL ≥ 1.5: +1.39R on 11 trades, 9 of them already S10b · Wyckoff 21/10/2026 , 534 assets 2021-2026, SC + AR range coded): spring W1 +0.01R vs −0.04R for a sweep + reclaim of a 10 d low without a range (gain +0.05R, 393 trades), halves +0.14 / −0.06 → ⚪; + dark anomaly in the range (M1) unstable (+0.42 / −0.50); spring with VIX 20-30 +0.25R (descriptive)

A18Volume climax at the low🟡

Rule · red day with RVOL ≥ 2 on a 10-session low

Role · trigger

Measure · 🟡 detector: pivot 37.5% vs 24% (lift 1.56) · v2.8: outside S10b days −0.03R → S10b bonus

A19Wick rejection below a bearish 0.618🟡

Rule · bearish leg ≥ 25%; the day high retraces 0.618-0.786 of the leg in log; close in the bottom 30% of the range (IREN 13/08, 08/09, 23/09)

Role · trigger

Measure · 🟡 = S11 trigger; ❌ if the band drops to 0.5 (−0.01R)

A20Equal highs swept⚪

Rule · new high within ±1.5% of a rejected top of the same leg (≥ 3 sessions earlier), then close in the bottom of the range (IREN; IBM 29/01/2026)

Role · trigger

Measure · ⚪ (S11b: 2 cases, +1.33R)

A21Multi-anchor fib cluster (Léo's doctrine)❌

Rule · ≥ 2 anchors (200-session swings, leg ≥ 20%) have a 0.5 / 0.618 / 0.66 / 0.705 / 0.786 (log) level inside the day's rejection wick; bonus FVG / liquidity grab (IREN: 4 to 7 anchors)

Role · location

Measure · ❌ price alone: short −0.25R (n 79), long −0.50R (n 64); + FVG or liquidity: no better; + top 3 node within ±0.25σ: short +0.15R vs −0.79R without node, unstable → ⚪

A22Bounce / rejection on a moving average❌

Rule · long in a bullish regime: low ≤ MA ≤ close (EMA21, 55, 144, 233, SMA200, weekly EMA21 and SMA50); mirror short in a bearish regime

Role · location

Measure · ❌ on daily: long −0.25R (232 trades); short −0.18R (161)

A23Off-hours liquidity hunt up to the King⚪

Rule · in pre / post-market, price exceeds a multi-month range high up to the GEX or VEX King strike, then the session opens below and rejects (IBM 01/06/2026, spike 341 = King 350, −40% afterwards)

Role · trigger

Measure · ⚪ (1 case)

A24Retest of an accepted breakout⚪

Rule · close above the 63-day high 2-10 sessions ago (level K); the low comes back to ≤ K + 0.25σ and the close holds ≥ K (mirror short)

Role · trigger

Measure · ⚪ v2.8: long +0.52R (77 trades) but +0.89 / −0.16 depending on the half; short 11 trades

A25Fib extension cluster (Léo's red box)❌

Rule · ≥ 3 different legs have a Léo extension (−0.234 … −2.616, log) within ±0.25σ of the day high (SPY 06/10/2026: 7-10 anchors between 780.6 and 782.6)

Role · location + target

Measure · ❌ (guide 35): 82% of 10 d highs sit in a cluster → sorts nothing; exit zone, not a short trigger

A26Price meets time (Léo's doctrine)⚪

Rule · the correction (from the impulse top to the correction low) lasts at least as many bars as the impulse (from the starting low to the top), extended-session bars (pre and post-market included); correction duration / impulse duration ratio ≥ 1

Role · location

Measure · ⚪ (20/10, case C047: SPY 760.64 Fri 15:00 → 770.46 Mon 11:30 = 23 bars of 30 min; correction → 763.71 Tue 08:00 = 23 bars, ratio 1.00; 1.04 in 15 min; 1.55 in regular-session minutes only); then rebound +1.1% in 45 min

A27Volume above the 90th percentile (BTC checklist)❌

Rule · volume > 252-session 90th percentile at the 10-day low

Role · trigger

Measure · ❌ 21/10/2026 : −0.37R vs −0.22R

A283 or more reds + accelerating ATR (BTC checklist)❌

Rule · ≥ 3 red closes, ATR14 > 1.2 × ATR14 of 3 sessions earlier

Role · trigger

Measure · ❌ 21/10/2026: spots the true low slightly better, not enough for 🟡

A29Drawdown from the high ≥ 25% (15% ETF)⚪

Rule · close ≥ 25% below the highest available high

Role · location

Measure · ⚪ = the base (stocks); ❌ ETF version (21/10/2026)

A30Liquidation proxy (gap + volume)❌

Rule · gap ≥ 1σ + volume ≥ 2× normal

Role · trigger

Measure · ❌ 21/10/2026: −1.02R

A31Wave 1 in 5 swings (Elliott) then buy wave 2❌

Rule · leg from a 60-day low to the next pivot (causal zigzag k = 4σ) containing a valid lower-degree impulse (k = 2σ, 4 Elliott rules); buy the 0.618-0.786 (or 0.5-0.618) pullback, stop below the origin, target 1.618 × wave 1

Role · trigger

Measure · ❌ 21/10/2026 , 538 assets 2018-2026): +0.12R vs +0.39R for legs without 5 swings (166 / 4,391 trades), 2nd half −0.53R; 0.5-0.618 zone same (❌) — the 5-wave structure adds nothing in these test conditions (daily, coarse sub-wave detection)

A32Macro hidden order block (Léo's HOB indicator)⚪

Rule · 1W → 12M: 3-candle FVG covering ≥ 50% of the body of the last opposite order block (not mitigated at mid-body); zone = overlap; dies if a wick crosses its midpoint, survives a body crossing only if a new FVG forms → count x (x2+ = « double hidden »); zones from several timeframes merged

Role · location

Measure · ⚪ 21/10/2026, spec strategies/S009-hob-macro/spec.md (H1 reaction vs placebo zone, H2 x2+ > x1, H3 1W ∩ 1M, H4 + lab context, H5 targets); AMZN case 1W ≈ 267.8 (short toward 165.7, earnings on 29/10 → K9 conflict); no public evidence of an OB / FVG edge

A33Wyckoff LPS after SOS (phase D)🟡

Rule · accumulation range (SC = 60-day low, volume ≥ 2× median, range ≥ 1.5 ATR, drop ≥ 15%; AR = high of the next 10 bars = T); SOS = close > T on volume ≥ 1.5× median; LPS = 1st pullback within 15 sessions that touches T and closes above; entry at close

Role · trigger

Measure · 🟡 21/10/2026 : +0.06R vs −0.11R for the same pullback after any break of the 60 d high (385 trades, 288 dates, halves +0.13 / +0.20) — real gain but low absolute R → to combine (S10b at the SC, F19 / F28 in the range)

A34Wyckoff UTAD (short)⚪

Rule · distribution range (BC = 60-day high, volume ≥ 2×, range ≥ 1.5 ATR, rise ≥ 15%); false breakout of the range top ≤ 1 ATR then close below; short at the close

Role · trigger

Measure · ⚪ 21/10/2026: −0.13R vs −0.22R for any rejection of a 10 d high (726 trades, halves +0.11 / +0.12) — better than the base but losing in absolute terms, like the other shorts outside S11 / S13

B1Sweep❌

Rule · the low of a 1-min bar goes below an A5 level, then a 3-min close comes back above

Role · trigger

Measure · ❌ alone at the moment of the touch, ⚪ combined · v2.13: version « sweep of a 0DTE node ≥ 50% of the King + 3 min reclaim » (raked entry B, hourly maps) ❌ −0.09R on 1,674 trades, blind period −0.02R, empty-strike control −0.12R (guide 37) · 21/10/2026 on liquidity levels , 10 assets, 1 year of 1 min bars, 0 credits): sweep of PDL / LOW10 / pre-market low then 1 min close above within ≤ 15 min, 09:30-11:30: +0.10R vs buying the touch (the 2 halves positive) but −0.36R in absolute terms with 0.05% costs per side (≈ 0 without costs), LOW10 ❌ → ⚪; the tight stop (0.1σ) makes costs huge in R: to retest with a stop ≥ 0.5σ and a filter (S10b, flow F29)

B2Limited sweep⚪

Rule · the overshoot below the level stays < 0.3σ (a poke, not a break)

Role · trigger

Measure · ⚪

B3Fast reclaim⚪

Rule · back above the level within ≤ 5 min after the low

Role · trigger

Measure · ⚪

B4Displacement⚪

Rule · +0.5σ within 10 min after the reclaim (otherwise break-even / exit)

Role · risk / stop

Measure · ⚪

B5Opening flush⚪

Rule · the sweep happens between 09:30 and 09:45; the IB low forms first

Role · trigger

Measure · ⚪

B6Overnight range rule⚪

Rule · session inside the overnight range → the overnight high / low serves as target

Role · target

Measure · ⚪

B7Sector flush⚪

Rule · ≥ 3 assets of the same sector (or QQQ) sweep their PDL in the same 5-min window (MRVL: AMD, MU, AVGO, TSM, NVDA, QQQ at 09:32-09:35)

Role · trigger

Measure · ⚪

B8Failed IB breakout⚪

Rule · price pierces the IB high / low toward a node then re-enters the range within ≤ 15 min

Role · trigger

Measure · ⚪

B9Acceptance, not a wick⚪

Rule · a level is broken only on a close beyond it: daily close in swing, 15-min close intraday

Role · risk / stop

Measure · ⚪

C1Significant node at the level❌

Rule · |GEX| of the 2 nearest strikes ≥ 20% of the King (Garma rule)

Role · location + risk / stop

Measure · ❌ alone (version ≥ 2% of the board) / ⚪ version 20% of the King

C2Sign of the node at the level❌

Rule · size beats sign. Positive = floor that cushions; negative = can block or accelerate, read with C16 and C17

Role · context / bias

Measure · ❌ alone (holds 47-53% whatever the sign or size) · v2.13: nodes 45% vs empty strikes 44% at the same place → the node alone does not hold better than an empty strike

C3King at the level⚪

Rule · the King strike is within ±0.25σ of the level

Role · location

Measure · ⚪ · pinned days v2.0: close within ±0.5σ in 65% of cases vs 66% without a pin → ❌ for selling credit spreads

C4Positive King just above✅

Rule · King > 0, above price, < 2% away

Role · target + context / bias

Measure · ✅ KING v1 (SPXW 01/10 and 14/10/2026: target hit, guide 34) · audit v2.13: +0.18R vs −0.02R for the same touches without a positive King (target at the same distance); the King strike as a target adds only +0.06R; zero at around 0.09% slippage per side → liquid stocks / ETFs, not options

C5Opposite King as target✅

Rule · target = largest node above (long); R ≥ 3 up to it

Role · target

Measure · ✅ (best exit tested) · v2.13: weak standalone contribution of the target (+0.06R vs a target at the same distance)

C6Air pocket toward the target🟡

Rule · all strikes between entry and target < 20% of the King

Role · target

Measure · 🟡 small n (+0.69 R, n=12) · contradicted by the community study (target hit 40% vs 54%), to retest

C7Gatekeeper on the path❌

Rule · node ≥ 25% of the King between entry and target → target cut to the gatekeeper

Role · target + veto

Measure · ❌ as an entry filter (+0.03 R); keep for placing the target

C8Node growing at the level—

Rule · |node| at the level +50% or more within 15 min (GOOGL ×3)

Role · trigger

Measure · 👁 · v2.14 (guide 37, 1,542 breaks, hourly maps): a node that grows after a break → break accepted 41% vs 62% if it melts, but at equal distance the gap disappears → not a break signal; it says the level still attracts price

C9Rising floor / falling ceiling🟡

Rule · strike of the largest positive node below price rising over 2-3 maps; variant: King rising over 3 maps

Role · context / bias

Measure · 🟡 small n (+0.15 R, n=23)

C10Tug of war⚪

Rule · Σ|exposure| above vs below price

Role · context / bias

Measure · ⚪ · community study: direction read from the map ≈ coin flip

C11Net GEX❌

Rule · sum of all strikes > 0 (dampened market) / < 0 (accelerated market)

Role · context / bias

Measure · ❌ alone · intraday: SPY net GEX < 0 at the session low → rebound +0.22σ vs +0.07σ (⚪, guide 36) · v2.14: half size in negative index gamma ❌; S10b does even better in negative gamma (+1.6 / +1.7R vs +0.3 / +0.4R, small n) → lead for an S10b filter

C12Re-mapping after the sweep—

Rule · within 30 min after the sweep, the King or the floor rises

Role · trigger

Measure · 👁

C13Reverse rug (long)⚪

Rule · positive node below price, negative node(s) just above it, price above the positive node → hold and bounce → never short. Size comes first

Role · veto

Measure · ⚪ · at 10 d low +0.79R (13 trades), outside S10b days −0.08R → S10b bonus · community study: as direction 45-55%

C14Rug (short)⚪

Rule · positive node above negative node(s) just below it, price below the positive node → rejection then acceleration to the downside

Role · trigger

Measure · ⚪ · at 10 d high +0.32R (40 trades), unstable · v2.14: « the node grows during the break = real break » does not hold (C8)

C15Map size✅

Rule · average node size vs its last 60 days: small map = wider day

Role · context / bias

Measure · ✅ (magnitude)

C16Weight of the node at the level on the weekly map⚪

Rule · on the weekly expiry (and 0-31 d): the level's strike is in the top 3, ≥ 50% of the King and ≥ 8% of total exposure, whatever its sign (MRVL 250; SPY 14/10/2026: 767 positive no. 3 at the low's strike)

Role · location + risk / stop

Measure · ⚪ (alone at the touch: −0.02 R → used for the stop and sizing)

C17What lies below the level (risk)⚪

Rule · largest node between the level and −2σ: < 25% of the King = air pocket → tight stop just below the level's node; 2nd large node within < 1σ → stop below that 2nd node

Role · risk / stop

Measure · ⚪

C18Node consumed after the sweep⚪

Rule · within 30 min after the sweep, the level's VEX node loses ≥ 50% and the GEX node becomes the weekly King (MRVL 09:35)

Role · trigger

Measure · ⚪ (hourly maps)

C19Expiry carrying the node (roll-off)🟡

Rule · share of the node coming from the nearest expiry; if ≥ 70%: the target only holds until that expiry (CRCL King 90)

Role · target + context / bias

Measure · 🟡 partial

C20Nearest floor / ceiling⚪

Rule · first positive node ≥ 15% of the King within ≤ 2.5% below / above price

Role · location + target

Measure · ⚪

C21Dormant node⚪

Rule · distance to the node > 1.5 × expected move until its expiry → not a target on size alone

Role · target + veto

Measure · ⚪

C22Level node melting after the reclaim⚪

Rule · within 30 min after the sweep, |GEX| of the node at the level drops ≥ 20% = confirmation; if it holds or grows = no-go

Role · trigger + veto

Measure · ⚪ · v2.14: delivery rule confirmed: after a touch, node that melts → price comes back 46% vs 73% if it grows (2,159 touches, stable)

C23Negative ceiling building up⚪

Rule · during the trade, negative nodes grow above price (+50% over the session) → exit at the first positive node below them

Role · risk / stop + target

Measure · ⚪

C24Skylit node labels❌

Rule · nodeType field (king, gatekeeper, pika, barney, normal) instead of in-house thresholds

Role · location + target

Measure · ❌ alone (broken gatekeeper → next node 58% vs normal strike 60%)

C25Positive King above swelling (swing)⚪

Rule · King > 0 within < 2.5% above, value ×2 or more in ≤ 4 sessions while price holds (XLU 40)

Role · target + context / bias

Measure · ⚪

C26Giant negative reclaimed on its expiry day⚪

Rule · the level's node grows ×5 vs the prior day on the day's expiry, price closes above, then the post-expiry map is a reverse rug (PLTR 112, HOOD 69, NVDA 167.5, AMZN 200; SPXW 01/10/2026: 7620 −$82M 0DTE, close above); also seen at tops → pivot marker, direction comes from the rest

Role · trigger

Measure · ⚪ · 21/10/2026 , dataset B, 0 credits, 17 cases / 16 dates): low pivot marker ⚪ (6% vs 4%, 2nd half 0%), high pivot ❌, long trigger ❌ (6 trades, −0.37R vs −0.11R) — no edge in these test conditions; the ×5 filters almost nothing on expiration day; to retest on more history

C27Giant Friday pin in the base❌

Rule · expiry Friday: positive weekly King ≥ 3× its value of 2-3 sessions earlier, close within ±0.5% of the strike

Role · context / bias

Measure · ❌ direction / 🟡 weak for magnitude

C28Positive King above flipping negative⚪

Rule · at the top, positive GEX King just above; on the retest 2-6 sessions later, the same strike is negative and the top 3 is negative (SPY 785: +568 → −$147M)

Role · trigger

Measure · ⚪: only 3 retests over 9 assets × 6 months

C29Index Kings disagreeing (friend, rule 1)❌

Rule · 0-31 d GEX Kings of SPXW, SPY and QQQ not all on the same side of their price

Role · veto

Measure · ❌ as a KING v1 veto (19/10, 213 trades): +0.22R with vs +0.10R without; most frequent state (75% of trades) → definition to clarify with him (expiration? 0DTE?)

C30GEX King and VEX King opposed (friend, rule 1 bis; = N9)❌

Rule · on the traded asset, GEX King and VEX King on either side of price

Role · veto

Measure · ❌ as a KING v1 veto: +0.42R with vs +0.14R without (45 / 116)

C31King too big (friend, rule 2)❌

Rule · |King| / Σ |strikes| of the 0-31 d board in the top third (threshold 24.7% set on Oct.-March)

Role · veto

Measure · ❌ as a KING v1 veto: +0.32R with vs +0.09R without, in all 3 periods; logic: KING v1 targets the King, a big King attracts better. To test as a veto on breaks (pin)

C32« Vanna support below a gamma accelerator » map during a fast drop⚪

Rule · negative GEX above price + positive VEX below price (Skylit map per expiry) during a fast drop

Role · context / bias

Measure · ⚪ descriptive 21/10/2026 , 43 cases, 11 assets, 6 months): V 9/10 (90%) vs 45% with neither; higher FINRA dark (z30 +1.2 vs +0.5); testable with ≈ 2,110 credits of historical maps (Léo's OK required)

C33Clear SPX and SPY maps (Léo's doctrine: institutions hedge on SPX, the SPY map differs; not clear = no trade)🟡

Rule · on the 0-31 d SPXW and SPY maps, the GEX King is ≥ 1.5× the 2nd largest node (dominant King)

Role · veto

Measure · 🟡 21/10/2026 , KING v1, 213 trades): clear maps +0.55R (35 trades, 77% winners) vs +0.12R (178); better in all 3 periods (+0.17 vs +0.11 · +1.20 vs +0.47 · +0.43 vs +0.02); indices alone +0.81R (8) vs +0.09R (64) → filter or A grade for KING v1; as a strict veto it removes 84% of trades

C34SPX and SPY Kings on opposite sides⚪

Rule · 0-31 d GEX King of SPXW and of SPY on either side of their price

Role · veto

Measure · ⚪ 21/10/2026: +0.21R with vs +0.18R without (108 / 105), 1 period out of 3 worse → side disagreement does not hurt; what matters is the absence of a dominant King (C33)

C35Gamma flip (« switch », price where market makers' cumulative gamma exposure changes sign)❌

Rule · cumulative sum of 0-31 d gamma nodes by ascending strike; flip = sign change nearest to price at 09:35; below the flip = amplifying hedging (external reading), above = dampening

Role · context / bias

Measure · 21/10/2026 , SPY / QQQ / SPXW, 10/2025 → 10/2026, 140-159 sessions): session range below the flip ❌ (ratio 0.88-1.00); direction: SPY below its flip −0.20σ vs +0.06σ, 60% down vs 42% (2nd half −0.13σ, below the threshold) ⚪; QQQ ❌; SPXW map → SPY ⚪; only 14-27% of sessions below the flip; approximate definition (Skylit's « flip » in explain_levels is not documented)

C36« Falling ceiling » short (NVDA case 28-30/10)❌

Rule · positive GEX (or VEX) King above price falling over 3 15:55 maps, + failed reclaim of a broken 10-day low or close < SMA20

Role · trigger

Measure · ❌ 02/11/2026 , dataset B April → Oct. 2026): 2 to 24 trades, −0.87 to −1.17R, below the bases (−0.59R); bullish window, not a ban; extending to 2022 ≈ 7,500 credits, not justified

D1VEX node at the level❌

Rule · |VEX| of the 2 nearby strikes ≥ 20% of the vanna King

Role · location + risk / stop

Measure · ❌ alone at the touch (−0.07 R) → used for the stop

D2VEX King in the trade's direction⚪

Rule · the largest positive vanna node of the weekly / monthly expiry is above price (long) (GME: 30)

Role · target + context / bias

Measure · ⚪ · ❌ on daily as a signal → target only

D3Stacked King⚪

Rule · GEX King and VEX King on the same strike

Role · target

Measure · ⚪

D4VEX × implied volatility❌

Rule · positive VEX above + falling IV → market makers buy

Role · context / bias

Measure · ❌ alone

D5VEX staircase (several days)❌

Rule · vanna nodes rising (low) / falling (top) day after day

Role · context / bias

Measure · v2.14 (guide 37): day-after-day version ❌ (almost never fires); the community's « staircase across expirations » version on long +0.37R vs +0.12R (132 trades) but 2nd half +0.18R and no better than momentum → 🟡 fragile

D6VEX tug of war⚪

Rule · signed sum of VEX below price vs above

Role · context / bias

Measure · ⚪

D7VEX node velocity—

Rule · 1 h change of the VEX King and of the node at the level

Role · context / bias

Measure · 👁

D8VEX barrier⚪

Rule · negative vanna node ≥ 25% of the vanna King above price (long) = resistance; target placed below it (MRVL 290, HOOD 75, AMZN 290 / 310, IREN 55, SPXW 7900 on 14/10/2026); read on the latest map only (XLU counter-example)

Role · target + veto

Measure · ⚪

D9Positive VEX King below price = magnet for a short⚪

Rule · largest positive vanna node (all expiries) below price during the top = short target (NVDA 200, IREN 30, IBM 280 / 285)

Role · target

Measure · ⚪ · ❌ as a top signal; stays a target · 21/10/2026 as a target : ❌ at 10-session highs (reached 22% vs placebo 23%, n 304); ⚪ promising on S11 days (7 cases, 86% vs 37%) and S13 (19 cases, 32% vs 15%, 2 halves positive) — recount when n ≥ 20

D10Mixed floor at the low's strike⚪

Rule · at the low, the same strike carries a no. 1-2 positive GEX node and a negative ≤ 31 d VEX King (IBM 200 on 23/07/2026)

Role · location

Measure · ⚪ (1-2 cases)

E1Capitulation🟡

Rule · signed delta tilt (trades ≥ $10k, single-leg, ATM/OTM) over the 30 min before the level ≤ −0.25; on daily: net selling premium on the day of the low

Role · trigger

Measure · 🟡 (+0.17 R; KING v2 +0.25 R)

E2Absorption⚪

Rule · during the flush: puts sold at the bid + calls bought at the ask ≥ 40% of premium traded

Role · trigger

Measure · ⚪

E3Puts sold on the low⚪

Rule · puts at the bid on strikes ≤ level, within 10 min of the low

Role · trigger

Measure · ⚪

E4Call sweeps at the ask after the reclaim⚪

Rule · ≥ 1 ATM/OTM call sweep at the ask within 15 min after the reclaim

Role · trigger

Measure · ⚪

E5Position openings❌

Rule · calls at the ask, size > OI, within the hour

Role · context / bias

Measure · ❌ too rare

E6Flow momentum⚪

Rule · flow z-score 5-30 min vs the same hour over 20 days ≤ −2

Role · trigger

Measure · ⚪

E7Defended strike⚪

Rule · top-3 strikes by premium at or just below the level

Role · location

Measure · ⚪

E8Hedge filter—

Rule · ignore long-dated puts far from price

Role · —

Measure · cleaning rule

E9Wait for the flow to turn❌

Rule · enter when the flow flips back to buying

Role · —

Measure · ❌ (−0.15 R: too late)

E10Price / flow divergence (FL8)🟡

Rule · price makes a low ≤ that of [t−60, t−20 min] but the cumulative signed delta since t−60 is higher

Role · trigger

Measure · 🟡 (+0.20 R, n=47)

E11Dominant contracts⚪

Rule · top 5 contracts by premium: calls bought at the ask and/or puts sold at the bid ≥ 60% = buying bias

Role · context / bias

Measure · ⚪

E12Daily capitulation (swing)⚪

Rule · net selling premium ≥ 3 sessions into the low + RVOL ≥ 1.5, then flip to buying within ≤ 3 sessions (PLTR; SPXW 30/09-01/10/2026)

Role · trigger

Measure · ⚪ · ❌ on daily

E13Flow euphoria at the top⚪

Rule · on the day of the high: net buying premium and signed delta in the top 2 of the last 10 sessions, then selling the next day (NVDA 14/05, IBM 01/06, SPXW 13/10/2026)

Role · trigger

Measure · ⚪ · ❌ on daily; + close in the bottom 30% ❌ (−0.15R) · 22/10/2026 without the crossed trades : + close in the bottom 30% −0.04R vs base −0.12R (8 trades, GOOGL 18/09 drops out) → ⚪ (was ❌), still losing in absolute terms

E14Distribution at the top⚪

Rule · negative signed delta on the 2 sessions of the high while price rises (AMZN 04-05/05)

Role · trigger

Measure · ⚪

E15Abnormal option openings before the pivot🟡

Rule · opening premium of the last 5 sessions ≥ 2× the median of the 30 sessions before

Role · context / bias

Measure · 🟡 small n; flags activity, not direction

E16Large opening trade against the move⚪

Rule · opening trade ≥ $10M in the direction of the reversal, 1-3 sessions around the pivot

Role · trigger

Measure · ⚪ (2 cases)

E17Buying flow on the day of the low🟡

Rule · net bullish premium (trades ≥ $25k) > 0 on the day of the reversal signal — known at the close

Role · trigger

Measure · 🟡 entry at T+1 open +0.34R vs 0.00R; outside S10b days +0.14R → S10b bonus · 22/10/2026 without the crossed trades : T+1 entry +0.40R vs −0.05R (🟡 unchanged); separate agent with close entry +0.42R vs base +0.25R → 🟡 (was ⚪), outside S10b days +0.27R → ⚪ (was ❌); flip caused by 2 trades out of 53 (crossed trades that reversed the sign of the day's premium: MSFT 18/08 comes in, NVDA 21/08 drops out), margin +0.17R for a +0.15R threshold → fragile

E18Next-day OI on the bullish side⚪

Rule · on the morning of T+1, ≥ 60% of OI increases of contracts traded on T are on the bullish side

Role · context / bias

Measure · ⚪ inverse (−0.38R vs +0.23R, n 18) → veto lead

E19Options sold at the low🟡

Rule · at the 10-day low, share of premium traded at the ask (vs bid) in the bottom third of its 20 sessions (single-leg trades ≥ floor, before 16:00)

Role · trigger

Measure · 🟡 small n (19/10, dataset B, 9 assets, 6 months): +0.83R vs base +0.36R (48 trades, 31 dates, 2 halves +0.65 / +0.98); S10b + E19 +0.85R vs S10b +0.37R (22 trades); to confirm on 2023-2026 (chain_ratio)

E20Panic put buying at the low❌

Rule · same rule, share of puts bought at the ask in the top third

Role · veto

Measure · ❌ −0.26R vs +0.36R (45 trades); buying options at the ask (E19 high) −0.13R

E21Very negative spot CVD (Atlas signed volume)❌

Rule · 60-session robust z of bought − sold delta ≤ −2 at the 10-day low

Role · trigger

Measure · ❌ 21/10/2026 , 2024-2026)

E22Heavily sold option premium at the low (perp CVD proxy)⚪

Rule · strongly net selling premium at the 10-day low

Role · trigger

Measure · ⚪ descriptive 21/10/2026: +1.06R vs +0.25R (17 trades) — consistent with E19

E23Retail fear spike at the low (CBOE equity put/call)⚪

Rule · day's equity put/call ratio in its top 5% of 252 sessions and SPY at a 10-session low; measured close → +20 sessions

Role · trigger

Measure · ⚪ 23/10/2026 (CBOE 2006-2019, 0 credits): +1.41% at 20 d vs +0.67% at a 10 d low without a spike, halves +0.80 / +1.75, n 45 · the spike alone (≥ 0.86 / 0.90 / 1.00) ≈ normal day at 5-10 d (« every spike = a low » ❌) · same family as 5 d fear at the low , +1.25 pt) · out of sample 2020-2026 ❌ , CBOE daily files): −0.69% at 20 d vs +1.17% at a 10 d low without a spike (−1.86 pt, n 21); 2020-22 −3.14 pt (2020 crash, 2022 bear market: fear at the low kept going) / 2023-26 +0.62 pt → retest only with a pre-registered regime filter · caution: StockCharts' $CPCE (0.86 on 22/10) is not the CBOE series (0.69 the same day)

E24Options positioning at an extreme → long / short regime (post @TailThatWagsDog 05/11, inspired by Druckenmiller: blue / red bars on /NQ)❌

Rule · proxy: ln(call ÷ put premiums) of the day, rank over W sessions; rank ≤ q (extreme puts) → long regime, ≥ 1 − q (extreme calls) → short regime, held until the opposite extreme; position at the close of t+1

Role · context / bias

Measure · ❌ / ⚪ 06/11/2026 , pre-registered, Skylit 2023-03 → 2026-11, 12 variants QQQ / SPY / SPXW × W 63-126 × q 5-10%, Holm): 0 🟡, 8 ⚪, 4 ❌; long-short from −8.9 to +4.7% per year against +20 to +24% buy-and-hold; QQQ positive in 2023-24 then negative in 2025-26; SPXW flow the worst (4 ❌ out of 4 variants → consistent with « SPX = hedging »); limit: our proxy is not the near-the-money front expiration nor the buyer / seller side from the post

E25ES futures options flow by delta bucket (post @TailThatWagsDog 06/11: ES more directional than SPX, less hedging and 0DTE)⚪

Rule · call / put premiums bought vs sold on near ES expiries, by delta bucket

Role · context / bias

Measure · ⚪ not testable: CME futures options are not in Skylit (stocks / indices only). « Directional vs hedging » proxy = SPY vs SPXW in E24: neither works

F1DP level❌

Rule · one of the 20 largest prints (≥ 2.5% ADV, 90 days) within ±0.5σ of the level

Role · location + risk / stop

Measure · ❌ alone / ⚪ in a cluster

F2Freshness⚪

Rule · ≤ 3 revisits, or VEX present at the DP strike (Léo's rule)

Role · location

Measure · ⚪

F3Largest print of the quarter⚪

Rule · the level carries the no. 1 print of the 90 days (MRVL 251.90; SPY $3.41bn at 758.80 on 18/09/2026 = low of 01/10 to the cent)

Role · location

Measure · ⚪

F4Accumulation⚪

Rule · ≥ 3 prints at the same price (±0.1%) over ≤ 5 sessions

Role · location

Measure · ⚪

F5DP ladder above⚪

Rule · prints ≥ 1.5% ADV between entry and target, grouped by price = intermediate targets

Role · target

Measure · ⚪

F6Massive DP in the base⚪

Rule · ≥ 3 prints ≥ 10% of ADV in the base over ≤ 5 sessions (XLU)

Role · location

Measure · ⚪

F7Dark pool surge🟡

Rule · DP notional of the 6 sessions ≥ 3× the median of the 30 sessions before, prints ≥ 5% ADV

Role · location

Measure · 🟡 weak

F8Potential accumulation (FINRA DPI)⚪

Rule · 5-day return < 0, 5-day DPI rising, in the top 20% of its 60 sessions, off-exchange volume ≥ its 21-day average

Role · trigger + context / bias

Measure · ⚪ (consistent direction, carried by the 2nd half)

F9Potential distribution (FINRA DPI)⚪

Rule · 5-day return > 0, 5-day DPI falling, bottom 20%, volume ≥ average (IREN, SPY 07-13/10/2026)

Role · context / bias

Measure · ⚪ unstable

F10Market absorption (FINRA DPI)⚪

Rule · SPY 5-day < 0 and 5-day market DPI in the top 20%

Role · context / bias

Measure · ⚪ adds nothing (the real DIX: F13 / F14)

F11Market DPI breadth❌

Rule · share of liquid symbols with DPI > 50%, top 20%

Role · context / bias

Measure · ❌

F12Largest DP print on the day of the top⚪

Rule · the no. 1 print of the 90 days lands on the day of a rejected high (IREN 13/08)

Role · context / bias

Measure · ⚪ (1 case)

F13DIX spike after a low (SqueezeMetrics)🟡

Rule · SPX close at a 10-session low within the last 3 sessions and DIX in the top 5% of its 60 sessions (02/10/2026: rank 98%)

Role · trigger

Measure · 🟡 t+1 entry (06/11, : +1.60% at 10 sessions (+1.84% at t), 53 episodes, halves +0.95 / +1.28 pt, p 0.001 → v2 ✅ (no costs, no trial registry) → ✅ candidate (Léo review) · before 06/11: 🟡 (SPX 2011 → 2026): +1.84% at 10 sessions, 81% up (n 94) vs +0.91% / 62%; stable before / after 2019; weaker since 2022: +0.89% over 10 d, 61% up (18 signals) vs +1.99%, 88% before · regimes 21/10/2026: positive gain with SPX above and below its MA200, VIX 15-20 and 20-30 (+0.70 to +1.34 pt); too few episodes to judge

F14Low DIX at the high (SqueezeMetrics)⚪

Rule · SPX within ≤ 1% of its 20-day max and DIX at rank ≤ 30% (12/10/2026: rank 25%, GEX −43%)

Role · context / bias

Measure · ⚪: less upside, not a decline → reduce, do not short

F15Low SqueezeMetrics GEX❌

Rule · SqueezeMetrics GEX < 0 or in its lowest 10% (60 sessions)

Role · context / bias

Measure · ❌ alone; 🟡 with H13 (agent 5)

F16Long-term money at the low⚪

Rule · VOO volume / SPY volume in the top 20% of its 60 sessions at the 10-day low

Role · location

Measure · ⚪: true pivot 42% vs 27%, but R per episode = base → locates the low

F17Dark volume anomaly (FINRA) at the low⚪

Rule · 10-session low and the day's off-exchange volume ≥ +2 robust deviations (log, median / MAD) over 30 and 90 sessions; entry the next day (FINRA file published in the evening)

Role · location

Measure · ⚪ (113 assets 2022-2026: −0.00R vs −0.10R base, 1,744 trades, the 2 halves above; + high DPI the same day: ❌ −0.22R)

F181-week DPI ranking (Signal Sigma columns)⚪

Rule · each day, top decile of the universe's 5-session DPI

Role · context / bias

Measure · ⚪ (+0.65% over 10 d vs bottom decile, t = 1.0, not significant; 1, 10, 20, 60 sessions flat or negative; « accumulating » at the low ❌)

F19Dark accumulation by location (Léo's rule)on PANDORA🟡

Rule · dark volume anomaly (FINRA, z30 and z90 ≥ 2) in a buy location: 10-day low closed in the top of the range, discount 0.5-0.786, trend pullback (> MA200 + discount); entry the next day

Role · location + trigger

Measure · 🟡 (113 assets 2022-2026, vs the same location without anomaly: absorbed low +0.29R vs −0.05R, 439 trades; discount +0.11R vs −0.12R, 648; trend pullback +0.52R vs −0.04R, 78; positive gain on the 2 halves; 6 🟡 out of 26 tests, 1-2 expected by chance). 2026 check (19 Tempest assets): 50% of F19 days are S10b days; F19 outside S10b −0.34R (n 11) → not a standalone agent, S10b size bonus: S10b with F19 +1.46R vs +0.29R without (n 11 / 57, to confirm) · Confirmed on 600 assets (19/10, : discount +0.18R, deep discount +0.19R, absorbed low +0.19R, trend pullback +0.40R, the 2 halves positive. Daily radar: liquid universe, score min(z30, z90) of total dark volume · regimes 21/10/2026 : positive gain in every SPX trend, every VIX bucket and every episode (+0.10R to +0.74R) → holds everywhere · F19 = stress signal (21/10/2026, : stressed market (SPX < MA200 or VIX ≥ 20) → +0.33R absolute (1,695 trades, 307 dates, gain +0.36R vs the location without anomaly), on the 113 original assets (+0.44R) and the 424 never-seen ones (+0.29R), 4 years out of 5 (2026 −0.49R, n 115); calm market → −0.04R absolute (gain +0.11R) → trade it alone only under stress; stress rule chosen after the regime audit → to confirm live

F20Dark distribution by location❌

Rule · same anomaly at the 10-day high, in premium, above weekly / monthly +1σ, rejected high, failed bounce

Role · context / bias

Measure · ❌ (no gain over the location alone: the dark anomaly at the high does not signal a decline)

F21SqueezeMetrics GEX regime = size of moves✅

Rule · rank of the prior day's GEX within its 252 sessions: high (≥ 2/3), low (≤ 1/3), negative

Role · context / bias

Measure · ✅ descriptive (SPX 2011-2026: absolute return 0.42% in high GEX vs 0.86% low vs 1.71% negative; SPY range 0.95% vs 1.56% vs 3.02%; same after 2022); no direction

F22Raw DIX ≥ 50% (SqueezeMetrics)❌

Rule · day's DIX ≥ 50%

Role · context / bias

Measure · ❌ since 2022 (+0.25% over 10 d vs +0.47% all days; before 2022: +4.0%, 14 days): genuine carry short-selling, no longer a buy signal

F23Breakout in low GEX⚪

Rule · close at the 20-day high when the prior day's GEX is low or negative

Role · trigger

Measure · ⚪ (500 stocks 2018-2026: +0.14R better than in high GEX, on the 2 halves, just below the 0.15R threshold; absolute R ≈ 0)

F24Very negative GEX + heavy dark buying (Signal Sigma / TailThatWagsDog, 18/10)on PANDORA⚪

Rule · SqueezeMetrics GEX z-score over 126 sessions ≤ −2 and DIX rank among its last 60 sessions ≥ 0.95 (top 5 %), the same day; a new mark only when F24 was off the 5 prior sessions; known at close t (pandora/strategies/f24_extreme_accumulation.py)

Role · trigger

Measure · Definition corrected 19/11/2026 to match the code (old text: « GEX ≤ −2σ from its 126-session mean and the ETF's « bought » dark volume (FINRA short volume) ≥ 2× its 63-session median; their version: total GEX of QQQ / SMH; ours: the index's SqueezeMetrics GEX (proxy) ») · ⚪ promising, small n (proxy 2022 → 15/10/2026, measured from the next day's close): QQQ 3 episodes +6.6% at 10 d, 3 out of 3 up (vs +0.65%); SMH 6 episodes +4.5%, 83% (vs +0.96%); SPY 4 episodes +1.6% / +5.3% at 20 d; extreme GEX without dark buying ≈ base; widened version (−1.5σ, 1.5×) diluted → the extremes matter. Fired on 16/10/2026: index GEX −2.19σ, QQQ 2.84×, SMH 3.13× (SPY 1.69×). F19 family (dark accumulation at a buy location). Generalized (engine , any asset with data): native SPX (SqueezeMetrics GEX + SPY FINRA dark) 4 episodes +1.5% / +5.3% at 20 d; SPX with the DIX as dark buying (2011 →) 13 episodes +1.6% vs +0.5%; 538 assets with index GEX as proxy (2022 →): +1.44% at 10 d vs +0.48%, 61% up, but clustered on 52 dates: the first signals of a decline fail (Feb-Mar 2026), the later ones pay → to diagnose. SMH case 16/10/2026 (registry C041): dark buying 3.13×, dark prints ≥ $5M stacked on the 0.618 / 0.66 / 0.706 fibs of the discount ($592M), tracking open. Live (19/10): SPX in the PANDORA Telegram brief (state IDLE / ARMED / CONFIRMED); per ticker, never mix Skylit GEX and open-interest GEX in the same z-score (guide 44) · Live 16/10/2026: failed at +10 sessions (30/10: QQQ −2.7%, SMH −3.2% below its invalidation 529.53, SOXX −4.4%); +20 on 13/11

F25Multi-horizon dark buying index (DPAI, Léo's idea 19/10)❌

Rule · day's FINRA dark buying ≥ 2× its 5, 21, 63 and 252-session median, and ≥ 2× the market's median pop that day (option: 2 days in a row)

Role · context / bias

Measure · ❌ as an improvement (600 assets 2022-2026, : 0 🟡 out of 7 locations, always below F17 / F19; total dark volume and the robust z beat « bought » volume and a multiple. With GEX ≤ −2σ: +3.1% at 20 sessions vs +1.1% (F24 lead)

F26Gamma-implied vol (GIV proxy, SqueezeMetrics)⚪

Rule · VIX − past median of 21-day realized vol in the same GEX bucket, ranked over 5 years

Role · context / bias

Measure · ⚪ (19/10, 2013-2026): adds nothing to the VIX alone; without index VEX, the proxy restates the GEX

F27Large dark block at the low (AMZN 16/10)⚪

Rule · a dark print ≥ 0.5% of ADV within ±0.1% of the session low, that low being a buy location (10-day low, discount 0.5-0.786, −1σ); opening / closing auctions excluded; known intraday (Skylit prints)

Role · location + trigger

Measure · ⚪ (1 case: AMZN 16/10, $95.7M at 248.605 = low to the cent, normal total dark volume → F19 silent)

F28Unusual dark blocks (Skylit, prints ≥ $1M) in a buy locationon PANDORA🟡

Rule · day's block notional ≥ 3× its 30-session median, robust z30 ≥ 2, ≥ 2 blocks; entry at the next day's close

Role · location

Measure · 🟡 fragile (20/10, 519 tickers, 05 → 10/2026): +0.21 to +0.29R vs the location alone, but alone −0.36R in the 2nd half; with F19: +0.20R, 2 halves positive; same-day entry ⚪ / ❌ · example XLF 29-30/09/2026 (C049): blocks 7.1× and 11.1× the median at the 10 d low below −1σ month, low 52.82 on 01/10 on a double discount, +4.4% by 21/10

F29Dark block shelf touched in a discount, then flow flip (C047)🟡

Rule · 3 largest buckets of blocks ≥ $1M of the previous 5 sessions; touched from above in a 0.5-0.886 discount; 5-min net option premium > 0 after a candle < 0 within 20 min

Role · trigger

Measure · 🟡 filter (20/10, 9 assets, 04 → 10/2026, 239 touches): +0.17R vs the touch alone (2 halves), but trade ≈ 0 (+0.05R); SPY / QQQ −0.57R (n 20)

F30Persistent dark activity (« stochastic »)⚪

Rule · total FINRA dark volume above its 60-session median on ≥ 14 of the last 20 sessions, 20-session return ≤ 0, in a buy location

Role · location

Measure · ⚪ 21/10/2026 , 113 assets 2022-2026, 0 credits): +0.12R / +0.09R vs the location alone, below F19; version on the dark share ⚪ alone, ❌ with F19

F31F19 + persistence (F30)⚪

Rule · F19 anomaly (z30 and z90 ≥ 2) on the same day as F30, on discount / absorbed low / trend pullback

Role · location + trigger

Measure · ⚪ (downgraded, see end of line) · 1st measurement 21/10/2026: +0.53R vs +0.19R for F19 alone (a2/a6/a7, 263 trades, 163 dates, gain +0.41 / +0.29 on the 2 halves); +0.25R vs +0.06R on a1-a7 (561 trades); 2026 ≈ 0 (−0.04R) → to confirm on 600 assets and on 2026 before any use · downgraded ⚪ 21/10/2026 : on 538 assets +0.20R vs +0.10R for F19 (1,213 trades, halves +0.23 / +0.03); on the 424 never-seen assets +0.03R / 0.00R vs F19 → the edge over F19 does not replicate; gain came mostly from VIX ≥ 20 regimes (2022), hypothesis to pre-register · regimes 21/10/2026: vs F19, loses when VIX < 15 (−0.06R) and in 2025 (tariffs −0.14R, recovery −0.33R) → regime-dependent

F32Bullish dark divergence⚪

Rule · close at a 20-session low below the previous one, 5-session average of dark z30 higher than at that previous low

Role · trigger

Measure · ⚪ alone · ❌ vs F19 (21/10/2026)

F33Effort / result absorption❌

Rule · dark z30 ≥ 2, day range ≤ 0.8 ATR20, close in the top of the range

Role · trigger

Measure · ❌ on long: no edge in these test conditions (3R, 10 sessions, stop below the extreme, very tight on a narrow candle; yet the 10-session return is +2.5%) → retest with a stop of at least 0.5σ · short mirror ⚪

F34Dark divergence at highs (short)⚪

Rule · higher 20-session high with lower dark activity (or higher, variant)

Role · trigger

Measure · ⚪ / ❌ (21/10/2026); F20 unchanged ❌: dark activity at highs does not signal a decline

F353σ drop + burst in dark shareon PANDORA🟡

Rule · drop ≥ 3σ (20-day realized vol) over 3, 5 or 10 sessions + at least one day with z30 ≥ 3 of the dark share (FINRA / total volume) over the last 3 or 5 sessions; entry at the next day's close

Role · trigger

Measure · 🟡 candidate 21/10/2026 : 6 neighbouring cells on 113 assets (+0.37 to +0.78R vs −0.17R for the drop alone, 25-48 trades); replicates on 424 never-seen assets (+0.12 to +0.31R vs −0.03R; F19 +0.07R), 2 halves positive; small samples, 0-3 trades in 2026 → live tracking · regimes 21/10/2026: zero gain when VIX 15-20 (−0.01R, n 94) and during the 2022-2024 recovery (0.00R, n 109), +0.41 to +0.63R elsewhere → regime-dependent

F3610% drop in 10 sessions + burst in total dark volume (5 sessions)⚪

Rule · same as F35 in % and on total volume

Role · trigger

Measure · ⚪ 21/10/2026: +0.14 to +0.17R stable on 3 panels, out-of-universe gain +0.13 (below the threshold), 2023 and 2026 negative

F37z velocity + burst + F19 location⚪

Rule · 5-session return z ≤ −2 + dark burst + discount / absorbed low / trend pullback

Role · trigger

Measure · ⚪ 21/10/2026: good on 113, ≈ F19 out of universe

F38Fast rise + dark burst (distribution)❌

Rule · rise ≥ 3σ / 15% over 3-10 sessions + dark burst at the 20-day high → short

Role · trigger

Measure · ❌ / ⚪ 21/10/2026: comes from 2025, nothing out of universe; confirms F20

F39Abnormal Mag 7 dark activity at the QQQ low (communicating vessels, Léo)on PANDORA⚪

Rule · average of 30-session robust z of total FINRA off-exchange volume of AAPL, MSFT, NVDA, GOOGL, AMZN, META, TSLA ≥ 2, QQQ closed at a 10-session low within the last 3 sessions; measured from the close of D+1 to D+11

Role · trigger

Measure · ⚪ promising 21/10/2026 , 2021-2026): +4.61% at 10 d vs +0.70% at the same location without a signal, gain +3.9 pt, 6 episodes out of 6 up, 2 halves positive, p 0.02; 6 episodes < 8 required, all with VIX ≥ 20 (consistent with F19 under stress) → live tracking, to recount · fired on 30/10/2026 with VIX at 19.4, below the 20 of every measured episode; D+1 to D+11 result to recount

F40Abnormal SPY + VOO + IVV dark activity at the QQQ low⚪

Rule · same rule, S&P basket (average z of SPY, VOO, IVV)

Role · context / bias

Measure · ⚪ 21/10/2026: +1.86% vs +0.68% (20 episodes), 1st half −0.12 pt, p 0.15 · reverse direction QQQ → SPY ❌ (−0.79 pt) · distribution at the high (strong S&P basket, or activity drying up over 5 d) ❌; strong Mag 7 at the QQQ high ⚪ (7 episodes)

F41ETF pullback + off-exchange burst in negative GEX (X post SMH, 28/10)on PANDORA🟡

Rule · close ≤ 0.92 × 63-session high, 30-day robust z of FINRA off-exchange volume ≥ 2 on D to D−2 (since 03/11: bursts on expiry Fridays, their eve and month-ends excluded; cleaned rule +1.48 / +2.71 pt, 133 cases, does not fire on 30/10), SPX GEX (SqueezeMetrics, proxy) < 0 or bottom 10% over 252 d; variant + 5-day short ratio ≥ 80th percentile

Role · context / bias

Measure · 🟡 t+1 entry (06/11, : r20 +1.2 / +3.7 pt (original rule; the « cleaned » version of 03/11 has no script); v2 ✅ but 12 pooled ETFs → p and episodes optimistic, and rule tweaked after the signal → 🟡 · before 06/11: ✅ 12-ETF panel 28/10/2026 , 2021-26): r20 +1.1 / +3.4 pt on 2 halves (88 / 69 cases), placebo +0.4 / +2.1; r5 near zero in the 1st half (+0.1); 60 d vol after +6.7 pt (« more volatile » claim confirmed) · short ratio variant ⚪ (r20 +2.5 / +4.8, negative placebo, 22 / 18 cases) · SMH alone ⚪ (5 / 10 cases, 1st half negative) · SMH's own GEX: after recharge · Fired on 30/10/2026 on QQQ and SMH (QQQ 0.905 of the 63 d high, off-exchange z30 +3.3, index GEX rank 6%; SMH 0.817, +2.5); F24 proxy narrowly missed (GEX z −1.76, QQQ dark buying ×3.17, SMH ×2.21); GEX OI convention (calls +, puts −, 1-120 d) QQQ −$6.5bn / 1% vs −7.3 on 16/10, SMH −0.33 vs −0.85

F42Stretched market breadth: 2-year z-score of the IWM / SPY ratio < −1.5 (X post 28/10 · same rule as H29, keep both results)⚪

Rule · IWM / SPY ratio, z over 504 sessions; episode = 1st day then 63 sessions of silence

Role · context / bias

Measure · ⚪ 28/10/2026 , 2002-26): 31 episodes, SPY +2.6% at 1 month (+1.7 pt vs normal, placebo +0.3), +8.5% at 6 months (+2.8 pt); but 2nd half 2014-26 +0.6 pt at 1 month < placebo +1.1; small caps catch up +1.2 pt at 1 month; all days counted (the post's method): +1.6% at 1 month (not +3%), 6 months positive 80% vs 77% (not 91%); 2007-08: 4 signals before the crash

F43Small caps / value decorrelation: 60-day corr. IWM-IVE < 0.65 (X post 28/10)⚪

Rule · correlation of daily returns over 60 sessions

Role · context / bias

Measure · ⚪ 28/10/2026: rare (2000, 2017-18, 2026: 5 episodes); IWM lags IVE by −4.6 pt at 3 months afterwards (consistent with the post: indebted small caps penalized); SPY without direction; active since 15/10 (0.61)

F44Intraday dark block at the day's low (reading of the 26-30/10 exports)⚪

Rule · Skylit block ≥ max($20M, 95th percentile 30 sessions), 09:30-15:50, in the bottom 20% of the day range, down day; cleaned of prints at the closing price, outside the range (late reports) and duplicates; entry at the close

Role · location

Measure · ⚪ 31/10/2026 (9 stocks, 04 → 10/2026, 72 episodes): +5 d = down days without a block (+0.01 pt, placebo 51%); +10 d +0.61 pt (2 halves +0.79 / +0.43) to track; blocks at the high −0.29 pt at 5 d (not pre-registered)

F45Persistent dark activity near a Mag 7's high⚪

Rule · close ≥ 97% of the 63-session high and FINRA off-exchange volume above its 60-session median on ≥ 14 of the last 20 sessions

Role · context / bias

Measure · ⚪ 31/10/2026 : drop ≥ 10% at 20 d in 24% of cases vs 13% (63 episodes, p 0.01), 18% vs 13% on 2018-21 (45, p 0.34), lower return both times; does not pass the multiple-testing correction, +2 pts on the other assets

F46Mag 7 top rotation❌

Rule · at least 3 Mag 7 10% below their 63-session high, QQQ within 3% of its own; variant + dark burst on one of them

Role · context / bias

Measure · ❌ price alone 31/10/2026: 33% vs 18% of 5% drops on 2021-26 (18 episodes) but 11% vs 16% on 2012-21 (18); dark variant ⚪ 45% vs 18% (11), holdout of only 3 episodes

F47FINRA patterns near the high, one by one and stacked⚪

Rule · 8 other single-stock patterns and 11 patterns between Mag 7 and ETFs (same study); stack ≥ 3 within 5 sessions

Role · context / bias

Measure · ⚪ / ❌ 31/10/2026: none passes the correction, combined zero (14% vs 13%, 103 episodes)

F48Dark + price model, trained on the past only⚪

Rule · 32 comparable FINRA variables + price, boosting, target −2σ at 20 sessions, 112 assets 2019-26

Role · context / bias

Measure · ⚪ 31/10/2026 : +0.028 AUC over prices, 5 / 6 years, above the placebo; but −0.038 on the Mag 7 and only +0.010 once market context is added; 2026 replay unusable

F49Quiet top: below-normal dark activity before a top followed by a decline⚪

Rule · −2σ tops vs highs that keep going, within each stock: persistent activity −0.36 standard deviation, z30 −0.31 at D−20/−11, distribution days −0.31

Role · context / bias

Measure · ⚪ 31/10/2026: same sign 2018-22, 2023-26, Mag 7; possible contamination (low vol makes −2σ easier); day by day no slope on the Mag 7

F50Thin liquidity near an index high❌

Rule · 20-session price impact per dollar ≥ 80th percentile of 252, index within 3% of its high

Role · context / bias

Measure · ❌ 31/10/2026 : 1 → 10% by quintile on 2019-26, but 5% vs 7% on 2000-18 (64 episodes), 7% vs 7% on 7 other ETFs

F51Low return concentration among leaders⚪

Rule · share of the 1st component over 60 sessions in the bottom 20% of its 252 values (Mag 7 → QQQ, sectors → SPY)

Role · context / bias

Measure · ⚪ 31/10/2026 : Mag 7 2013-18 24% vs 16% (21 episodes, p 0.32), 2019-26 24% vs 26%; sectors 2000-26 31% vs 35% (210); not a decline signal

F52Dominant down volume in a range after a rise⚪

Rule · 40-session range after a rise, ln(volume of up days / down days) ≤ −0.20 → range exit to the downside first, 40 sessions

Role · context / bias

Measure · ⚪ 31/10/2026 : at equal position +3 to +5 pts on 2000-18, +3 to +7 on FINRA 2019-26, p ≤ 0.09; small, and the 20 d return is not lower

F53Wyckoff: heavier volume at the top of the range, false upside breakout, stacked❌

Rule · same study, top third of the range

Role · context / bias

Measure · ❌ / ⚪ 31/10/2026: volume at the top −1 pt, false breakout −3 pts, combined −5 pts on 2000-18

F54The big buyer at the bottom selling at the top❌

Rule · block ≥ $10M at the bottom of the range, size within ±1% at the top 5-60 sessions later, 9 stocks 04 → 09/10/2026

Role · context / bias

Measure · ❌ 31/10/2026: 184 low → high pairs vs 202 at random, 187 high → low vs 213; repeated sizes look like algorithm slices (hypothesis)

F55Dark pool ranked alone = short-term weakness⚪

Rule · cross-sectional ranking by off-exchange z30, by change in 5-day short share, or « dark + location » composite

Role · context / bias

Measure · ⚪ 02/11/2026 : negative spreads IS and OOS (rising short share → underperformance at 5 d, t −4.3 / −5.1); reverse direction not pre-registered → validate live before a short; does not contradict L4 / F19, which filter on location

F56Very low DIX (SqueezeMetrics)⚪

Rule · DIX in the bottom 2% of its 252 sessions (02/11/2026: 39.97%, rank 1.2%)

Role · context / bias

Measure · ⚪ 03/11/2026 , SPX 2011-26, 63 cases): no direction (+0.82% at 20 d vs +1.05% all days, 57% up); 10 d realized vol higher only since 2019 (23.9% vs 20.5% after a decline, 16.1% all days; before 2019: lower) → confirms SqueezeMetrics « more vol, no bias », on the recent period only

F57GEX at a low then DIX at a low (SqueezeMetrics)on PANDORA🟡

Rule · GEX in the bottom 10% of 252 sessions on one day from t−5 to t−1, GEX back up since, and DIX in the bottom 5% at t, below its level on the day of the GEX low (02/11/2026: GEX 6% on 30/10, DIX 1.2% → present)

Role · trigger

Measure · 🟡 t+1 entry (leading since 06/11): +3.4 / +3.5 pt at 20 d (+4.35% at t), 19 episodes (< 30: ✅ impossible), p < 0.001 · before 06/11: 🟡 fragile 03/11/2026 , SPX 2011-26, n 20): +4.35% at 20 d, 85% up vs +1.31% / 68% after a decline; halves +3.88% / +4.83%; excluding 2020 +3.50%; 36 settings all above; low GEX alone already does +2.11% (70%) → the DIX mostly adds consistency; high vol (27%) · v2 04/11: 🟡 (19 episodes, p < 0.001, t+1 entry: +3.4 / +3.5 pt at 20 d instead of +4.35%)

F58Classic top profile (SqueezeMetrics + vol + breadth)❌

Rule · S&P within ≤ 1% of its 252-day high, GEX ≥ 80%, VIX ≤ 30%, 20-day RSP/SPY breadth ≤ 20% (252-day ranks); 06/10/2026: present

Role · context / bias

Measure · ❌ 03/11/2026 , 2011-26): it is the average signature of the 11 major tops, but ex ante n 49, −10% crash within 60 d 4% vs 7% base; risk rises only below −5% from the high (defensive rotation 31%, inverted VIX 32%, GEX < 0 29% vs 10%)

F59SPY: puts dominate options flow (Skylit)⚪

Rule · (call premiums − put premiums) / total SPY premiums in the bottom 5% of its 120 sessions (30/10/2026: rank 2%, present)

Role · context / bias

Measure · ⚪ since 07/11/2026: PANDORA computes it on flow without packages (single-leg trades ≥ $50k, since 04/2026), not re-measured; total flow counted multi-leg packages (06/11: a $283M SPY hedge roll wrongly fired F59 and F60) · old measurement: 🟡 fragile 03/11/2026 , 2023-26, 5 credits): effect at 20 d vs all days +2.04 / +0.59 pt (n 14 / 18), 2nd half 50% up; short history

F60SPY: option premium spike (Skylit)⚪

Rule · total SPY premiums, robust z ≥ 2 vs 30 sessions (30/10/2026: z +2.1, present)

Role · context / bias

Measure · ⚪ since 07/11/2026: PANDORA computes it on flow without packages (single-leg trades ≥ $50k, since 04/2026), not re-measured; total flow counted multi-leg packages (06/11: a $283M SPY hedge roll wrongly fired F59 and F60) · old measurement: 🟡 fragile 03/11/2026 : +0.82 / +0.14 pt at 20 d (n 26 / 27); puts z ≥ 2 near the high ❌ (not bearish)

F61SPY: unusual off-exchange volume (FINRA)on PANDORA⚪

Rule · total SPY off-exchange volume, robust z ≥ 2 vs 30 sessions, anywhere / at the low / near the high; variant excluding expiry and month-end

Role · context / bias

Measure · ⚪ 03/11/2026 , 2018-26): no direction (negative 2018-22, positive 2022-26); 29% of spikes fall on an expiration day (8% of sessions) → mechanical effect; high short share at the low ❌

F62High DIX while GEX is negative (dark pool accumulation, SqueezeMetrics)on PANDORA🟡

Rule · DIX in the top 10% of its 252 sessions and GEX < 0 on the same day (rule-based version of the cyan dots on the @TailThatWagsDog GEX-DIX chart, placed by hand)

Role · trigger

Measure · 🟡 03/11/2026 , SPX 2011-26, 32 episodes): +20 d +2.55% / +2.57% vs base +0.75 / +1.24% on the 2 halves, next-day entry same; +60 d 91% / 67% up. Their orange dots (low DIX near the high): no effect, not kept

F63The stock's largest dark pool prints (size ranking, Skylit)🟡

Rule · the day's largest print is in the top 5 of the stock's history (≥ $25M, ≥ 40 sessions of history); at the top of the 60-day range (D) or at the bottom (A), cluster = 2nd within 30 sessions

Role · location + context / bias

Measure · 03/11/2026 , 522 tickers, 04 → 10/2026 only): at the bottom 🟡 (+2.76 / +0.36 pt vs SPY at 20 d; clusters +3.27 / +1.10); at the top ⚪ (−1.25 then +0.68 pt; clusters −3.06 then +1.10): « distribution » holds in only one half. AMD June 2026 (cluster $505-550): −16 to −18% vs SPY at 20 d, then $650

F64Record off-exchange days (FINRA, dollars)❌

Rule · same rule as F63 with FINRA days (not individual prints), excluding expiry / month-end, 2018 → 2026

Role · context / bias

Measure · 03/11/2026 , 519 surviving tickers, 2019-26, 3,000+ cases): at the top ❌ as distribution (the stock beats SPY: +1.27 to +1.54 pt at 20 d) but more breakage (drop ≥ 10% within 40 d 33-36% vs 24-27%) → volatility flag, not a short; at the bottom ⚪

F65Record dark pool activity on index and sector ETFs (F63 / F64 applied to SPY, QQQ, VOO, IVV, IWM, DIA, SMH, sectors…)⚪

Rule · F63 (Skylit) and F64 (FINRA) rules, absolute returns, S&P 500 counted once per date

Role · context / bias

Measure · 03/11/2026 : Skylit at the top ⚪ low n but right direction (29 cases, −3.38 / −1.32 pt at 20 d vs base, drop ≥ 10% within 40 d 40% / 17% vs 9%); FINRA 2019-26 at the top ❌ as distribution (153 cases, +0.79 / +0.49 pt) but drops ≥ 10% more frequent (18% / 14% vs 14% / 7%); at the bottom ⚪. Recent signals: VOO 22/09 and XLK 01/10 at the top; XLP, XLU, XLF, XLY, XLI at the bottom (Sept.-Oct.)

F66Leveraged ETF confluence: dark pool record at the top + activity on short ETFs, or at the bottom + activity on 2X / 3X bull ETFs (Léo's idea)❌

Rule · FINRA, unusual off-exchange volume (log robust z ≥ 2) on SQQQ SH PSQ SDS SPXU SPXS QID SOXS TZA (short) or TQQQ QLD SSO UPRO SPXL SOXL TNA (bull) within the 5 sessions

Role · context / bias

Measure · ❌ / ⚪ 03/11/2026 , 2019-26): adds nothing. At the top + active short: +3.87 then −1.00 pt vs without; at the bottom + active bull: −14.98 then +3.31 pt; variant SPY near the high + active short: +0.28 / −0.04 pt; SPY at the low + active bull: worse (−0.85 / −0.46 pt, ❌). Short and bull ETFs activate on the same days (choppy days, internalized retail flow): the activity has no directional meaning

F67Rotation read in the dark pool: defensives (XLP + XLU) vs QQQ🟡

Rule · FINRA record at the top on SPY / VOO / IVV / QQQ / XLK / SMH or at the bottom on XLP / XLU / XLV within the 20 sessions → defensives − QQQ pair

Role · context / bias

Measure · 🟡 fragile 03/11/2026 , 2019-26, 21 cases): +2.04 / +0.70 pt at 20 d, but negative at 40 d in the 2 halves; RSP − QQQ ⚪; inactive on FINRA as of 30/10, active on Skylit (7 months, inconclusive)

F68Price returning to a cluster of record dark pool prints (level)❌

Rule · ≥ 2 top-5 days within 30 sessions, zone = dark pool VWAP of those days ± 0.5%, return from above or below, rejection ≥ 3% before a break ≥ 3% within 10 sessions

Role · location

Measure · ❌ 03/11/2026 , 526 returns, 04 → 10/2026): rejection 31% vs 34% on random zones, break 28% vs 24%: clusters do not hold as levels

F69Net imbalance short ETFs − leveraged bull ETFs (FINRA)⚪

Rule · average robust z (30 valid sessions) of off-exchange volume of S&P / Nasdaq short ETFs minus that of bull ETFs; ≥ +1.5 or ≤ −1.5, near the high (≤ 2%) or at the 10-session low

Role · context / bias

Measure · ⚪ 03/11/2026 , 2019-26): too rare (2 to 11 cases per half); only clear result: bulls more active near the high on QQQ = not a top (QQQ does better, +3.21 / +0.60 pt, ❌ as a bearish read). As of 29/10: NET ≈ 0 (SPY −0.2, QQQ +0.5)

F70Leveraged ETF flows (Bloomberg post « US Leveraged ETF Fund Flows »)on PANDORA⚪

Rule · ProShares, 18 leveraged index ETFs (long + inverse), 1-month flow ÷ AUM, 2-year rank; HIGH ≥ 90%, LOW ≤ 10%

Role · context / bias

Measure · 03/11/2026 , 2008-26): strong inflows ⚪, not bullish (+2.13 pt then −1.13 pt at 60 d; since 2018 drop ≥ 10% within 60 d 30% vs 14%); strong outflows = bullish (+2.77 / +1.85 pt, after a pullback +5.65 / +2.89 pt, breakage 10% / 0% vs 26% / 24%) → 🟡 contrarian. As of 02/11: inflows at rank 93% (HIGH episode on 22/10), net longs − inverses 96%

F71Record volume on a bullish leveraged ETF, index ETFs calm (VolumeLeaders post « SSO biggest trade since 2014, SPXL, TQQQ » before the 03/11 open)🟡

Rule · Day's dollars of SSO QLD UPRO SPXL TQQQ ≥ 1-year high, SPY QQQ IVV VOO neither at a record nor unusual (robust z < 2), mechanical days excluded

Role · context / bias

Measure · 04/11/2026 : total volume 2008-26 🟡 modestly bullish (24 cases: +1.1 / +1.4 pt at 20 d vs same location) but drop ≥ 10% within 60 d more frequent (25% / 25% vs 18% / 15%) and next day negative on average; FINRA off-exchange version 2019-26 ⚪ (10 cases). On 03/11/2026: SSO + QLD at a record (SSO also off-exchange), SPY did +1.65% the next day · check 05/11: on 03/11 SPY / VOO / IVV calm (volume z −0.8 / −0.1 / −0.8, off-exchange z −0.4 / 1.4 / 0.3), anomaly limited to SSO (z 16.7, off-exchange 30.1), SPXL (off-exchange 9.6) and QLD; broad activity only on 04/11 (SPY off-exchange z 2.8, UPRO, SDS, TQQQ). No PANDORA layer (next day negative on average), proposed as switched-off context, awaiting Léo's go

F72Massive share redemption on bullish leveraged ETFs (one day)⚪

Rule · ProShares, day's flow ÷ prior day's AUM ≤ 1st percentile of the previous 2 years (LONG = 10 funds, CORE = SSO + QLD); mirror: inflow ≥ 99th percentile

Role · context / bias

Measure · 04/11/2026 , 2008-26): extreme outflows and inflows both slightly bullish (+0.5 to +1.1 pt at 20 d) → no directional information (⚪ in substance); LONG outflows: breakage less frequent (12% / 6% vs 16% / 14%). On 03/11/2026: SSO −4.0 M shares (≈ −$270M), QLD −7.55 M (≈ −$614M), LONG −1.6% of assets, and on the bearish side QID −25% / SQQQ −6%: deleveraging on both sides (election eve). The side of the big block itself remains unknown (day's net flow ≠ side of a print)

F73Large leveraged ETF dark print = creation / redemption plumbing⚪

Rule · Skylit print ≥ $25M on SSO QLD TQQQ UPRO SPXL QID; size vs the day's ProShares share change, price vs net asset value

Role · context / bias

Measure · 04/11/2026 , prints 04 → 11/2026): 7 of the 8 largest prints = 0.83 to 1.07× the day's creation / redemption; on 03/11 all below NAV (SSO −46 bp, QLD −12, TQQQ −19) and not before the open (14:58 → 16:02). SPY next day +0.19% vs +0.11% (n 53) → the print gives no direction; read shares created / redeemed ⚪

F74Crowded put hedges on the index (2 days)on PANDORA🟡

Rule · Put share of SPXW premium (Skylit, daily) in its top 10% of 120 sessions, two days in a row; variants net premium, SPXW + QQQ the same day

Role · trigger

Measure · 04/11/2026 , 2023-26): SPY +5 d +1.27% / +1.01% vs +0.38 / +0.40% (11 + 11); next day ≈ base → 🟡 lead (single period, 3 variants); on 03/11 rank 86%: did not fire · v2 04/11: 🟡 (19 episodes, p 0.003, t+1 entry

F75Event premium in 1-day volon PANDORA⚪

Rule · Tempest 1-day vol ÷ 30-day vol ≥ 1.25 at the close (or top 10% over 120 d)

Role · context / bias

Measure · ⚪ t+1 entry (06/11, : +0.53 / −0.10 pt (2nd half flips sign at the open), 2 years → v2 ⚪ · before 06/11: 04/11/2026 , 2024-26): Nasdaq-100: next day +0.76% / +0.24% vs +0.09%, up 70% / 61% (23 + 23) → 🟡 lead; S&P ⚪. 03/11: 1.31 (S&P) and 1.56 (NDX, rank 99%) · v2 04/11: 🟡 (35 episodes, p 0.019, only 2 years → not ✅)

F76Bottom setup when the protection regime says « cash »⚪

Rule · S10b without Tempest (10-day low + VIX +20% / 5 d or VIX / VIX3M top 20%) × PANDORA v_joint regime; exit 3R / stop / 10 d

Role · context / bias

Measure · 04/11/2026 , SPY + QQQ 2008-26): in cash +0.07R / +0.27R (208 trades), 3R reached 16-17% vs 3% in long regime; no better than a 10 d low on 2017-26 → ⚪, but the cash regime is not a veto of the setup

F77Oil collapse (≤ −7% in one session)⚪

Rule · WTI or Brent, 2001-26

Role · context / bias

Measure · 04/11/2026 : SPY next day +1.35% / −0.22%, +20 d −1.7 / +3.2 pt → ⚪

F78Day after a US election⚪

Rule · S&P, 1950-2026

Role · context / bias

Measure · 04/11/2026: +0.16% on average, 62% up (midterms 73%, n 15), average move 1.36% vs 0.77% in November → more movement, unreliable direction ⚪

F79Session option walls (call wall / put wall)⚪

Rule · Strike with the highest call (above) / put (below) gamma relative to price, open interest convention, prior day's chain, 5 expiries + monthly (SpotGamma definition)

Role · context / bias

Measure · ⚪ 05/11/2026 , SPY + QQQ, 22 sessions): call wall held 64% vs 57% for a level at the same distance (both tickers), put wall no better (59% vs 59-62%); walls at 0.5-0.7% from price; lead, shown in the brief since 05/11

F80DIX collapsed near the highs (SqueezeMetrics posts « Sagging » 03/11, « Utterly limp » 05/11)⚪

Rule · DIX in the bottom 5% of its 252 sessions at least 2 times in 3 sessions, S&P within 4% of its 1-year high; first day of the episode (10 sessions apart)

Role · context / bias

Measure · ⚪ 05/11/2026 (descriptive, thresholds set after the 05/11 drop, SPX 2011-26, 25 episodes): +20 d +0.64% vs +0.99% all sessions, a session ≤ −2% within 5 d 12% vs 13%, ≤ −3% 0% vs 4%; halves −0.2 / +1.4 pt → does not warn of a drop (only 2018-01 and 2020-09 followed by a clear decline). Active on 02-04/11 (DIX 40.0 / 42.2 / 40.8%, 1-year rank 1 / 8 / 4%) before the −4% of 05/11, caused by news (elections, AI). Variant 4 sessions in a row in the bottom 10% (06/11/2026, SPX 2011-26, 20 cases, ⚪ descriptive): +20 d +0.31% (55% up) vs +1.49% (67%), +60 d +0.81% vs +3.67%, back below the signal price within 20 sessions 18 / 20; only strong rebound = 03/2020. Active on 05/11 (4th session, DIX 37.9% = lowest since 24/01/2022)

F81Strong DIX jump from a low level (Léo's idea 08/11 « bottom forming »)⚪

Rule · prior day's DIX in the bottom 10% (or 20%) of its 252 sessions and DIX +10% relative in 1 session (variants: +10% in 3 sessions from the 5-day min; +3 / +4 / +5 pt; 60-day robust z ≥ 2 / ≥ 3); 1st day of the episode (10 sessions)

Role · trigger

Measure · ⚪ 08/11/2026 , SPX 2011-26, t+1 close entry, spread vs sessions at the same distance from the high): bottom 10% +10% 1 d: 31 episodes, +21 d +1.02% = −0.30 pt vs context, halves −1.27 / +0.16, p 0.66; bottom 20%: 40 ep., −0.28 pt, halves −0.54 / −0.12 → ❌; 9 variants, no 🟡 (Holm 1.00); sensitivity 5 / 25 settings > 0; 21 d bottom 42% (+2 pt); GEX afterwards no faster (−4 pt of rank). Lead +63 d (horizon chosen afterwards): +3 pt from bottom 10% +2.17 pt (53 ep., halves +1.2 / +2.9, p 0.006, Holm 0.06), z ≥ 2 +2.74 pt (p 0.013) → to confirm. Overlaps P3 44%, F57 34%. Active on 06/11 (37.86 → 42.36%, +11.9%, z +3.2, prior-day rank 0.4%); on the S&P chart since 19/11/2026 (Léo's go « the DIX pump of 18/11 is not on the chart »): every qualifying day drawn (37 since 2011, 32 episodes with this DIX file vs 31 in the lab report), 18/11/2026 qualifies (prior rank 0.4 %, +13.6 %) as the 2nd jump of the episode of 06/11; not in the brief

F82DIX jump without a price rebound⚪

Rule · F81 (bottom 10%, +10% relative) and SPX ≤ +0.5% over the same window (1 d or 3 d): the 06/11 case (SPX +0.3%)

Role · trigger

Measure · ⚪ 08/11/2026 : 1 d 22 episodes, +21 d +1.06% = −0.18 pt vs context, halves −2.16 / +0.75, p 0.58; 3 d 34 ep., −0.20 pt; the 5 d low holds 21 d in 27% / 12% of cases (−12 / −28 pt vs context, partly mechanical: price stuck at the low) → not a bottom sign

F83DIX jump with GEX at a low⚪

Rule · DIX 60-day robust z ≥ 2 in 1 session and GEX in the bottom 10% of its 252 sessions (DIX at any level); 1st day of the episode

Role · trigger

Measure · ⚪ 08/11/2026 , best pattern chosen on the 1st half): 18 episodes, +21 d +3.12% (83% up) = +0.95 pt vs context, halves +1.45 / +0.46, p 0.16; low GEX alone +2.24% (375 sessions) → DIX contribution +0.9 pt; overlaps P3 79%, P13 63%, F57 42%; sensitivity 13 / 25 > 0. Active on 06/11 (GEX rank 6%)

F84Unusual off-exchange activity (record 3 → 30 months or ≥ × 2 the 6-month average), read long or shorton PANDORA⚪

Rule · (Léo 09/11: « biggest dp activity » over 3, 6, 12, 18, 24, 30 months, long or short, with the day's GEX, all PANDORA assets) FINRA off-exchange volume × close at the high of the previous 63 / 126 / 252 / 378 / 504 / 630 sessions; the longest window beaten is shown; read by location: LONG = down day and ≥ 5% below the 1-year high (discount), SHORT = within 2% of the high (premium), otherwise context; day's GEX = rank of the S&P GEX over 252 sessions (low ≤ 20% or < 0, high ≥ 80%); 1st day of the episode (10 sessions)

Role · context / bias

Measure · ⚪ 09/11/2026 , 16 tickers SPY QQQ VOO IVV SMH IWM DIA XLK XLF + Mag 7, 2018-26, t+1 open entry, Holm over 24 tests): 0 🟡. LONG + low GEX: +5 d +1.2 to +2.1 pt vs context, 62-70% up (6 months: 37 ep., +2.52%, 68%) but 1st half negative everywhere; SHORT + high GEX: +5 d −1.0 to −2.3 pt vs context (18 months: 24 ep., −1.49%, halves in the right direction) but price still rises in more than one case out of two, p 0.34-0.39; 6-month record alone , 183 ep.): +0.15 pt; a down day ≥ 1%: +5 d +1.12% (61% up) → a read to watch, not a trade. 05/11/2026: QQQ 30-month record (LONG + low GEX), SPY 6 months, VOO 30 months, XLK 30 months · multiple , 10/11): ≥ × 2 the 6-month average at a discount + low GEX 114 ep., +5 d +1.47% (+1.01 pt vs context, 58%); ≥ × 3 at a premium + high GEX 34 ep., +21 d −2.50%, halves in the right direction, p 0.35; 12 tests, 0 🟡

G1Daily σ❌

Rule · move since the close ≤ −1σ (strong: ≤ −2σ)

Role · trigger

Measure · ❌ alone (location, not trigger)

G2Weekly σ⚪

Rule · price below the week's −1σ band

Role · location

Measure · ⚪ (location) · 21/10/2026 , 23 assets 10/2024 → 10/2026, 0 credits): close below Tempest −1σ week, entry at the close +0.22R vs base +0.01R (173 trades, 85 dates) but 1st half −0.15R → ⚪; next-day entry 🟡 +0.41R vs +0.18R (140 trades, halves +0.35 / +0.48) — candidate to confirm out of sample

G3Monthly σ −1σ (Tempest)⚪

Rule · anchor = close of the last monthly OPEX; live: /v1/vol/cones month level; history: anchor close × 30-day IV × √(days / 365)

Role · location

Measure · ⚪ (location) · 21/10/2026 : close below Tempest −1σ month, 48 episodes −0.64R vs base +0.01R (next day −0.61R vs +0.18R), stays below the base without the April 2025 gaps, 56% of episodes = S10b days → ❌ no edge in these test conditions; to retest with a longer history

G4Band edge + node❌

Rule · daily ±1σ edge on a node

Role · location

Measure · ❌

G5High skew🟡

Rule · expensive puts vs the last 60 days

Role · context / bias

Measure · 🟡 (2 periods out of 3); ❌ for pinpointing a low

G6Upcoming earnings—

Rule · release within the trade window (em_1d_event, /v1/vol/events)

Role · veto

Measure · v2.14: earnings during the holding period (stocks, S10b / S11 / S13) −0.43R vs +0.68R, stop 73% vs 52% (n 26); within the first 5 sessions −0.83R → swing veto adopted (Léo 15/10) = K9

G7Remaining σ budget⚪

Rule · the target fits within the day's remaining expected move

Role · target

Measure · ⚪

G8Sweep of a σ band⚪

Rule · the low pierces a σ band (day, week, month) then the 3-min close comes back above

Role · trigger

Measure · ⚪

G9IV rising in the base🟡

Rule · 30-day IV +20% or more in 5 sessions at the low

Role · context / bias

Measure · 🟡 measured: pivot 43% vs 24% (lift 1.81, stable) = S10b trigger

G10Front IV spike then crush⚪

Rule · at the low: front IV +15% and inverted curve; the next day −20%

Role · trigger

Measure · ⚪ unstable

G11Monthly σ +1σ / +2σ as target⚪

Rule · hi1 / hi2 of the month level = swing targets; +2σ = exhaustion zone

Role · target

Measure · ⚪ · ❌ as a top (price keeps going)

G12Near-term vol deflating in the base⚪

Rule · term slope ≤ −5 while price holds

Role · context / bias

Measure · ⚪ · ❌ (lift 1.02)

G13Skew flipped at the top⚪

Rule · rr25 ≤ 0 at the high while it was ≥ +5 at the last low

Role · context / bias

Measure · ⚪

G14Stacked day + week + month σ🟡

Rule · long: ≤ −1σ day + ≤ −0.5σ week + ≤ −0.5σ month; mirror short

Role · location

Measure · 🟡 long (location); ❌ short (continuation)

G15Call skew at the low⚪

Rule · rr25 < 0 at the low (threshold measured in research/pivot_detector.py:211; the « ≤ −5 » of the old text was never tested — audit 04/11)

Role · context / bias

Measure · ⚪ · ❌

G161-day vol panic (Tempest)🟡

Rule · svx1d / svx30 in the top 20% of its 60 sessions on a 10-session low

Role · trigger

Measure · 🟡 pivot 36% vs 24% (lift 1.51) = S10b trigger

G17IV above its usual range⚪

Rule · svx30 > p80 or iv_pct ≥ 90 on a low

Role · context / bias

Measure · ⚪

G18Term inversion at the high (short)🟡

Rule · 9-day vol > 30-day vol (term_slope > 0) on a 10-session high

Role · trigger

Measure · 🟡 t+1 entry (06/11, : S13 +0.25R vs base −0.15R (+0.52R at t); v2 🟡; 9 months of history · before 06/11: 🟡 with G1r = Setup 13 (+0.52R on 65 trades, fragile)

G1rRejection at +1σ (mirror of G1)—

Rule · day high ≥ prior close + 1σ AND close < prior close + 1σ (σ = prior day's em_1d_pct × prior close); code scripts/morning_scan.py:165-166, research/pivot_detector.py:237

Role · trigger

Measure · — row added on 06/11 (used by Setup 13 since 19/10 without a catalogue definition); no standalone measurement

G19Asymmetric σ bands (live)—

Rule · /v1/vol/cones: dn1_pct / up1_pct

Role · location + target

Measure · 👁

G20Remaining move budget (live)—

Rule · /v1/vol/sigma running_sigma

Role · target

Measure · 👁

G21Session pace and re-pricing (live)—

Rule · /v1/vol/screener day_pace, day_reprice

Role · trigger + context / bias

Measure · 👁

G22Earnings: implied vs past move⚪

Rule · /v1/vol/events: implied_move_pct vs median_past_move_pct

Role · context / bias

Measure · ⚪

G23Close below the daily lower Bollinger (Léo's ALYA indicator)⚪

Rule · close < SMA 20 − 2 × 20 standard deviation

Role · trigger

Measure · ⚪ alone; 🟡 with S10b: +0.81R (78 trades) vs +0.33R → best S10b filter · 21/10/2026 alone at a 10 d low, 114 assets 2022-2026 : ❌

G24ALYA indicator SDevBands❌

Rule · SMA 20 − 1.618σ / + 1.35σ (the dynamic multiplier cancels out)

Role · location

Measure · ❌; upper bands against the short (continuation)

G25Post-earnings low⚪

Rule · 30-day IV −15%, RVOL ≥ 1.5, sweep of a major low, close in the top 30% (IBM 23/07/2026)

Role · trigger

Measure · ⚪ (1 case)

G26High VVIX (vol of vol)⚪

Rule · VVIX in the top 20% of its 60 sessions (Tempest /v1/vol/market history.vvix)

Role · context / bias

Measure · ⚪ alone; 🟡 with H13 (agent 5)

G27Complacency at the high—

Rule · at the 10-day high: VIX9D/VIX3M or COR1M at the bottom of their range, or SqueezeMetrics GEX at its high

Role · veto

Measure · price keeps going (SPX 2011 → 2026) → short veto (K8)

G28VIX9D rising at the high⚪

Rule · SPY at the 10-day high and VIX9D ≥ 1.1 × its 10-day low

Role · context / bias

Measure · ⚪: more true tops (lift 1.51) but short loses → exit for a long

G29Astrological cycles❌

Rule · moon, retrogrades, stations, eclipses, aspects, Bradley siderograph (offline ephemerides; e.g. Venus retrograde 03/10 → 14/11/2026)

Role · context / bias

Measure · ❌ (guide 38): 112 tests + 5 published rules, nothing passes the multiple-testing correction or replicates on SPX 2011-2026

G3030-day skew in its top 10% (BTC checklist)❌

Rule · Tempest rr25_30 in the top decile of 250 sessions at the 10-day low

Role · trigger

Measure · ❌ 21/10/2026

G311-day IV at z > 2 (BTC checklist)❌

Rule · 60-session robust z of svx1d > 2

Role · trigger

Measure · ❌ 21/10/2026: 98% of its days are already S10b days

G32Usually overpriced earnings (« sell the earnings »)⚪

Rule · over the 4 previous releases, median actual move / priced move (Tempest 1σ) ≤ 0.8 and < 1 on ≥ 3 of the 4; straddle-selling proxy u = 0.8 − |actual| / priced

Role · context / bias

Measure · ⚪ 21/10/2026 , 202 stocks, 1,616 reports 2024-2026, 64 credits): options overprice 72% of reports (median realized = 0.58 × priced), but selling the straddle loses on average (u −0.11: 9.5% of reports move > 2× the priced move, worst −6.6); selection by history does +0.05 vs all (+0.08 / +0.02 depending on the half, 500 reports) and persistence is weak (rank +0.11) → « overpriced most of the time » ≠ profitable to sell; confirms K9 for stock swings

G33APMEM volatility regime (nextSignals, PANDORA copy)on PANDORA🟡

Rule · 22-day vol forecast from the APMEM model (Engle-Gallo, Parkinson variance); 12-day exponential average crossing above the 40-day (2 d of confirmation) = SHORT « volatility picking up again »; below = LONG; spike z ≥ 2.5 = LONG « recovery after a crash »; minimum 50 sessions between 2 signals; two variants on the PANDORA S&P chart (docs/SIGNAUX.md): G33 (a) apLong / apShort = rolling refit 756 / 21, γ fixed 0.332, horizon 21 (pandora/chart/layers.py); G33 (b) volRegLong / volRegShort = yearly walk-forward fit since 1990, horizon 22, the one measured here and the one in the brief (pandora/reports/market_stats.py)

Role · context / bias

Measure · 🟡 28/10/2026 , parameters re-estimated each year on past data, 2000-26, 39 LONG / 40 SHORT): S&P at 3 months +1.3 pt after LONG vs −0.8 pt after SHORT, spread in the right direction on the 2 halves; as an exit rule ❌: Sharpe 0.32 vs 0.41 buy-and-hold, CAGR 5.8% vs 8.3% (worst drawdown −48% vs −55%); matches 5 / 7 dates from the site; SHORT since 21/10/2026

G34Inverted VIX futures curve (McMillan VX3 − VX1 < 0)on PANDORA⚪

Rule · settlement of the 3rd monthly VIX contract − 1st (Cboe) < 0 at the close; « negative for equities » state until back to ≥ 0; long proxy VIX3M − VIX

Role · risk / stop + context / bias

Measure · ⚪ t+1 entry (06/11, : Sharpe 0.69 vs 0.75 buy-and-hold, worst drawdown −24.5%; as a 21 d signal: opposite sign on the 2 halves (❌); as protection ⚪ · before 06/11: 28/10/2026 , futures 2013-26 + proxy 2008-26): as a bearish signal ❌ — S&P +1.9 pt at 21 d vs normal on inverted days (2 halves, 2 datasets), median episodes of 2 sessions, S&P down during the episode only 32% of the time; as protection 🟡 — stepping out during inversion: same Sharpe (0.76 vs 0.75) with a worst drawdown of −22% vs −34%, but −3 pt of return per year; 3 halves out of 4. Curve positive as of 28/10

G35@TailThatWagsDog QQQ APMEM signals (replicated)🟡

Rule · APMEM-style vol on QQQ fitted to his published values; SHORT extreme vol: session ≤ −2.5% and vol ≥ 1.6 × the prior day; SHORT exhaustion: within 2% of the 252-day high, vol below the 80th percentile, down or VXN rising; LONG: from the SHORT state, ≥ 18 sessions, session ≥ +1%, 5 d ≥ +1%, vol ≤ 0.7 × its peak

Role · risk / stop + context / bias

Measure · 🟡 31/10/2026 : 12 / 19 dates from his chart within ±3 sessions for 40 signals emitted (extreme vol 4 / 4 to the day, exhaustion not found without his options data); outside the chart 2001-22 long / cash Sharpe 0.39 / 0.98 vs 0.24 / 0.96, worst drawdown −41 / −22% vs −54 / −33%, but much lower return in H2 (9.8 vs 18.9%); short selling ❌ (Sharpe 0.32 / 0.02)

G36@TailThatWagsDog SPY APMEM system (replicated from his public page)on PANDORA🟡

Rule · (Léo 10/11: « why does he have a signal and you don't ») his page marvelous-meringue-47da32.netlify.app embeds 752 sessions (11/2023 → 11/2026: APMEM vol, regime, signals, flags); S&P APMEM vol fitted to his series (correlation 0.93, frozen scale); LONG « vol rollover » = down day, ≥ 4% below the 1-year high, vol receding, 1-year vol rank ≥ 60%, vol ≤ 93% of its 20-day peak, from cash; SHORT « momentum exhaustion » = ≤ 2.5% from the high, vol rank ≤ 30%, 20 d ≤ +1%, 5 d < 0; 1-day / 1-week / 1-month forecasts = reversion toward 13.75% at 0.922 / day (his 3 published sets reproduced to the hundredth)

Role · risk / stop

Measure · 🟡 10/11/2026 , grid fitted on his window only): his dates found to the day 04/04/2024 short, 30/07/2024 long, 05/11/2026 long; 3 / 9 within ±3 sessions, same regime 67% of days; outside his window long / cash vs buy-and-hold: 2000-11 Sharpe 0.09 vs 0.14, worst drawdown −55% vs −55%; 2012-23 Sharpe 0.66 vs 0.82, worst drawdown −27% vs −34% → cuts the worst drawdown, not the return. His top / bottom watch flags (Signal Sigma options) not reproducible; Cboe put / call substitute ⚪

G37@TailThatWagsDog SPX APMEM system (5 types replicated)❌

Rule · frozen S&P APMEM vol (G36); SHORT extreme vol: session ≤ −2% and vol ≥ 1.8 × its 20-day low; shock near the top: ≤ 2.5% from the 252-day high, session ≤ −0.5%, vol ≥ 1.25 × its 20-day low → « exhaustion » if 20-day gain ≤ +1%, otherwise « ignition »; LONG from cash, ≥ 5 sessions after the short, 5 d ≥ +2.5%, vol ≤ 0.7 × its peak since the short → « recovery after a crash » if 1-year vol rank ≥ 70%, otherwise « confirmed recovery »; 20 sessions between 2 signals of the same type

Role · risk / stop

Measure · ❌ 10/11/2026 , grid fitted on his 09/2021 → 08/2026 window only): 9 / 13 of his dates with the right type within ±3 sessions (exhaustion 4 / 4 to the day, extreme vol 1 / 1, confirmed recovery 2 / 3, crash 1 / 2, ignition 1 / 3) but 40 signals emitted, same regime 52% of days; his 13 dates give +79.9% / −6.0% (he claims +91.5% / −5.9%), our rules +26.7% / −23.9%; outside the window 2000-10 / 2011-21 Sharpe −0.14 / 0.78 vs 0.05 / 0.79, worst drawdown −66 / −15% vs −57 / −34%. Ignition and the 2 longs cannot be separated without his short-term GEX

H1Trinity 2 out of 3⚪

Rule · at least 2 of SPX / SPY / QQQ tell the same story

Role · context / bias

Measure · ⚪

H2Alignment🟡

Rule · ≥ 3 other assets of the universe in discount within the hour

Role · context / bias

Measure · 🟡 (strong in 2026)

H3Laggard⚪

Rule · the asset in discount while its sector / index holds

Role · context / bias

Measure · ⚪

H4Market Tide❌

Rule · lean of the day

Role · context / bias

Measure · ❌ alone

H5VIX divergence❌

Rule · at the sweep, the VIX does not make a new high

Role · trigger

Measure · ❌; daily ⚪ unstable

H6Indices all at their low🟡

Rule · SPY, QQQ and SPX within < 0.25% of their session low

Role · context / bias

Measure · 🟡 (+0.14 R, n=167)

H7Divergence vs the macro driver⚪

Rule · the asset stops making lows while its driver still does (XLU ↔ TLT)

Role · context / bias

Measure · ⚪

H8Relative strength at the low⚪

Rule · the asset makes its low before the indices and does not make a new low when SPY / QQQ make theirs

Role · context / bias

Measure · ⚪ · ❌ on daily

H9Low aligned with the market (beta)🟡

Rule · asset and QQQ at the 10-day low on the same day

Role · context / bias

Measure · 🟡 measured: pivot 33% vs 5% (lift 6.5)

H10Top before the indices⚪

Rule · asset's high 2-3 sessions before that of SPY / QQQ

Role · context / bias

Measure · ⚪ · ❌

H11Tempest market regime⚪

Rule · /v1/vol/market: regime, VIX curve, VVIX, dispersion, Fear & Greed (120 sessions of history)

Role · context / bias

Measure · ⚪

H12Regime from moving averages—

Rule · BULLISH = EMA21 > EMA55 > EMA144 and close > SMA200; BEARISH = the reverse

Role · context / bias

Measure · S11: −0.19R in bullish vs +1.20R in range → S11 veto in a bullish regime

H13Unconfirmed index low🟡

Rule · SPY / VOO / QQQ at a 10-session low while IWM or SMH do not make their 10-day low

Role · trigger

Measure · 🟡 (guide 36): with stress (G26 or F15) pairs validated on 2023-2026: +0.61 to +0.78R vs +0.16R = agent 5

H14Everyone confirms❌

Rule · IWM, RSP, QQQ and HYG at the 10-day low on the same day as SPY

Role · veto

Measure · ❌ −0.27R (broad selling, not a low) → veto on the index long

H15Leader / follower⚪

Rule · the sector ETF made its 10-day low between D−5 and D−1 and closes above that day's high, while the stock makes its 10-day low on D

Role · context / bias

Measure · ⚪ (19/10, 423 S&P stocks, 2021-2026): −0.07R vs base −0.17R, beats H16 in the 2 halves, below the bar; reversed on the 19 Tempest assets in 2026

H16Sector at the low on the same day❌

Rule · the stock and its sector ETF at the 10-day low on the same day

Role · veto

Measure · ❌ −0.21R vs −0.17R, worse in the 2 halves; semis −0.31R. Not an S10b veto: with the vol panic, S10b + sector at the low = +0.79R (77 trades, 2026, descriptive)

H17Fast drop in negative index GEX = slide, not a V⚪

Rule · SPX ≥ 2-4σ below its 20-day high in ≤ 3-10 sessions with SqueezeMetrics GEX < 0

Role · context / bias

Measure · ⚪ 21/10/2026 , 2011-2026): V 31% vs 80% (n 16, p < 0.001, 2 halves), 20 d trough −6 to −11% vs −2% but +20 d no worse → do not buy the V, wait

H18Fast rise with index GEX in its top 10% = the rally holds⚪

Rule · SPX fast rise with GEX in the top decile of 252 sessions

Role · veto

Measure · ⚪ 21/10/2026: gives back 50% of the rally 11% vs 48% (n 19, p 0.002, 2 halves)

H19Fast rise with low DIX (index distribution)⚪

Rule · SPX fast rise, low DIX

Role · trigger

Measure · ⚪ 21/10/2026: gives back 58% vs 40%, +10 d −0.5% vs +0.7%; 🟡 in a single cell (k2 Y10), ❌ at k3

H20FINRA short volume spike on VXX → S&P low⚪

Rule · robust z60 of log ShortVolume VXX ≥ 3, 10-session episodes

Role · context / bias

Measure · ⚪ 21/10/2026 : 9 spikes (none before 2024), +1.54% at 10 d vs base, 56% near a low vs 20%; the signal comes from volume, not the ratio; spike in progress on 16/10 (z +5.0) → review at n ≥ 15

H21MTUM / USMV deeply oversold❌

Rule · EMA12 − EMA26 oscillator (log ratio) ≤ −2 robust z

Role · context / bias

Measure · ❌ 21/10/2026: 13 episodes, −1.0% at 20 d vs base (3 variants)

H22FINRA short volume spike on QQQ⚪

Rule · robust z60 of log ShortVolume QQQ ≥ 3

Role · context / bias

Measure · ⚪ 21/10/2026: 4 episodes; overlaps the F19 anomaly (X1) at 75%; 16/10: z +2.8, 2-year high but below the threshold

H23Quality outperforming (QUAL / SPHB proxy)🟡

Rule · 4-week outperformance of QUAL / SPHB in its top 10%

Role · context / bias

Measure · 🟡 fragile 21/10/2026: 15 episodes, +2.21% at 8 weeks vs base, 2 halves positive (+1.73 / +2.82%); random p 20%; ETF proxy (not the Dow Jones ratio); active the week of 20/10

H24QLDY / QQQ rising (vol-selling regime)—

Rule · QLDY / QQQ ratio, oscillator

Role · context / bias

Measure · not measurable 21/10/2026: QLDY missing from Atlas

H25ES Dec. 2027 / Dec. 2028 (leading growth)—

Rule · ratio of two far-dated ES futures

Role · context / bias

Measure · not testable 21/10/2026: CME futures not available; slow carry read, not a timing signal

H26Correlation skew rising fast (COR10D / COR90D)🟡

Rule · 4-week change of the ratio in its top 5%

Role · context / bias

Measure · 🟡 t+1 entry (06/11, : −2.4 / −1.8 pt (worse); v2 🟡 · before 06/11: 🟡 to qualify 21/10/2026 : 8-week drawdown −5.1% vs −2.9% (n 26, 2 halves); mostly vol already high (high VIX: −6.8% vs −4.0%) · the ratio « topping out » as a buy signal: ❌ (n 16)

H27Low beta vs high beta spike (SPLV / SPHB proxy)❌

Rule · ratio oscillator in its top 5%

Role · context / bias

Measure · ❌ 21/10/2026 as timing: −0.6% at 20 d vs +1.2% (n 15); marks locations (20% within ±3 sessions of a low vs 11%) but the rebound comes later; active as of 20/10

H28VIX / VIX3M ≥ 1 with the oscillator at a spike❌

Rule · backwardation + oscillator top 5%

Role · context / bias

Measure · ❌ 21/10/2026: −0.3% at 10 d vs +0.4% (n 40); ratio ≥ 1 alone ⚪ (+0.3 pt, n 108); PANDORA's B2 fires within the 10 sessions after each case (28 / 28)

H29Small caps stretched vs S&P → ratio reversion (IWM / SPY pair) · same rule as F42⚪

Rule · 504-session Z of the IWM / SPY ratio < −1.5; 63-session episodes; target ratio 63 / 126 d

Role · context / bias

Measure · ⚪ 28/10/2026: ratio +0.85 / +0.67 pt, n 31 (10 / 21), weak 2nd half; SPY +1.3 pt at 63 d; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H30Equal-weight stretched vs S&P → S&P lagging (RSP / SPY)⚪

Rule · Z 504 of the RSP / SPY ratio < −1.5; target SPY 21 / 63 / 126 d

Role · context / bias

Measure · ⚪ 28/10/2026: SPY −0.4 / −1.8 / −2.6 pt, n 34, mostly 2002-13; the ratio does not come back; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H31International stretched vs US → weaker S&P (EFA / SPY)⚪

Rule · Z 504 of the EFA / SPY ratio < −1.5; target SPY 63 / 126 d

Role · context / bias

Measure · ⚪ 28/10/2026: SPY −1.6 pt at 63 d (2 halves), n 40; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H32Consumer discretionary stretched vs staples → weaker S&P (XLY / XLP)⚪

Rule · Z 504 of the XLY / XLP ratio < −1.5; target SPY 63 d

Role · context / bias

Measure · ⚪ 28/10/2026: SPY −3.3 pt at 63 d (2 halves), n 17 including 6 in 2014-26; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H33Semis stretched vs Nasdaq → semis reversion (SMH / QQQ)⚪

Rule · Z 504 of the SMH / QQQ ratio < −1.5; target ratio 63 d

Role · context / bias

Measure · ⚪ 28/10/2026: ratio +0.8 pt at 63 d (2 halves), n 28 including 6 in 2014-26; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H34Semis / Nasdaq correlation break⚪

Rule · 60-day corr. of SMH vs QQQ returns < 0.70 (5th historical percentile); targets ratio and SPY 63 d

Role · context / bias

Measure · ⚪ 28/10/2026: ratio −0.9 pt, SPY −0.7 pt at 63 d, n 16; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H35Credit / Treasury decoupling⚪

Rule · 60-day corr. HYG vs IEF < −0.55 (5th percentile); target SPY realized vol over the next 21 sessions

Role · context / bias

Measure · ⚪ 28/10/2026: vol +11.9 pt, n 13, but mostly vol persistence (already 23 vs 15) and 2008 / 2020; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H36Financials stretched to the upside → reversion (XLF / SPY)⚪

Rule · Z 504 of the XLF / SPY ratio > +1.5; target ratio 21 d

Role · context / bias

Measure · ⚪ 28/10/2026: ratio −0.95 pt at 21 d, n 26; grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H37Very strong credit vs Treasuries → weaker S&P (HYG / IEF)⚪

Rule · Z 504 of the HYG / IEF ratio > +1.5; targets SPY 63 / 126 d

Role · context / bias

Measure · ⚪ 28/10/2026: SPY −0.9 / −1.3 pt, n 23 (4 in 2002-13); active as of 28/10 (Z +1.61); grid at chance level: 16 ✅ out of 195 tests vs 22 for the placebo

H38Cross-sectional ranking: distance from the 252-day low (relative strength)🟡

Rule · each day, liquid stocks ranked by close / 252-day low; long the top decile, 20 sessions

Role · context / bias

Measure · 🟡 fragile 02/11/2026 : OOS 2024-26 spread +4.13% per rebalance (t 2.4), long leg +3.11% (t 2.5), but IS 2022-23 +0.54 (t 0.5), current S&P 500 members (survivorship bias) and an AI momentum period; 126 d momentum, 5 d reversal, 20 d strength ⚪ / ❌; no short passes

I1Session window⚪

Rule · entries 09:30-11:30 and 13:30-15:30; no new options entry after 14:30

Role · veto

Measure · ⚪ · v2.14: KING v1 09:30-11:30 +0.15R (107), 11:30-13:30 −0.05R (23), 13:30-15:30 +0.13R (18), pre-market +0.39R (60) → too few to change anything

I2Calendar⚪

Rule · FOMC / CPI / NFP / OPEX: veto or reduced size

Role · veto

Measure · ⚪

I3Post-market✅

Rule · no entry in post-market

Role · veto

Measure · ✅ (−0.49 R)

I4Monthly OPEX week❌

Rule · nodes carry less weight during monthly OPEX week

Role · veto

Measure · ❌

I5Top before monthly OPEX and earnings⚪

Rule · high ≤ 1 session before OPEX and ≤ 5 sessions before earnings

Role · context / bias

Measure · ⚪

I6Top after an earnings gap⚪

Rule · high 1-4 sessions after an earnings gap

Role · context / bias

Measure · ⚪

I7Don't wait for the next day⚪

Rule · enter at the close of the signal day rather than at the T+1 open

Role · veto

Measure · ⚪ 06/11: replaced — signal data is published after the close → entry at the next open for S10b / S10 / agent 5 / S11 / S13 (lab and PANDORA PR #82); the +0.32R at the close has no source found, the +0.12R at the open comes from · before 06/11: 🟡: close T +0.32R vs open T+1 +0.12R

I8EMA13 / EMA21 cross against the trend⚪

Rule · loss of momentum, not a reversal

Role · context / bias

Measure · ⚪

J1Structural stop❌

Rule · stop = min(swept low, node strike, DP price) − 0.1σ; low taken on the 24 h price

Role · risk / stop

Measure · ❌ too tight without a buffer → always with 0.1σ and J3 (stop ≥ 0.5σ in our tests)

J2Target in R⚪

Rule · (target − entry) / (entry − stop)

Role · target

Measure · v2.14 (275 trades S10b / S11 / S13): 1R +0.10R · 2R +0.36R · 3R +0.55R · 5R +0.79R (all 3 agents, the 2 halves) → Léo decision 15/10: 5R target with a 1st step at 3R for S10b / S10, S11, S13; KING v1 keeps the King, agent 5 keeps 3R (not tested at 5R) · 06/11: pre-registered comparison of the 4 S10b exits , t+1 open entry, 113 episodes) → ⚪; S10b / S10 go back to 3R (original V1 rule) in the scan, the lab journal and PANDORA (PR #82)

J3Reasonable stop⚪

Rule · entry − stop ≤ 1σ

Role · risk / stop

Measure · ⚪

J4Break-even❌

Rule · stop to entry after +1R or after B4

Role · risk / stop

Measure · ❌

J5Size by magnitude✅

Rule · reduced risk if the expected magnitude is large (C15, rising VIX, big gap)

Role · size

Measure · ✅ (predictable magnitude)

J6Targets = clusters⚪

Rule · each target is a node ≥ 25% of the King and if possible a σ band and / or a DP group at the same price ±0.5σ; extension clusters (A25) = exit zones

Role · target

Measure · ⚪

J7One theme = one trade⚪

Rule · several correlated assets with the same signal on the same day = a single position (NVDA / SOXX / SMH; SPY / SPX / VOO)

Role · risk / stop

Measure · ⚪

J8Slippage❌

Rule · KING v1 drops to zero at around 0.09% lost per side (average risk 1.3% of price)

Role · risk / stop

Measure · v2.13: liquid stocks / ETFs only; options = ❌ for tight-stop intraday setups

J9Limit add near the low (C051 NVDA 30/10)🟡

Rule · after 1/3 at the close of a capitulation low (10-day low, ≤ −2σ day, discount 0.5-0.80), the remaining 2/3 as a limit order at low + 0.25σ for 10 sessions, same stop (low − 0.75σ)

Role · size

Measure · 🟡 02/11/2026 : 1,100 events 2012-26, PnL / risk +0.23 vs −0.04 all at D0 (+0.39 / +0.11 per half); adds on confirmation (reclaim of the broken level, D0 high, D1 follow-through) ≈ 0 since 2019; the location alone loses since 2019 (−0.10)

J10Exit by family (exit study 02/11)🟡

Rule · agent 5 / H13: time exit at 10 sessions or trailing after +1R (stop = low of the last 2 sessions), structural stop kept · L4-type dark accumulation: time exit at 20 sessions · never take half at 1R nor move the stop to entry after 1R

Role · target

Measure · 🟡 02/11/2026 , exits compared on the same entries): agent 5 (239 trades, SPY / QQQ) +0.76R with a 10 d time exit and +0.73R with the trailing stop vs +0.29R for 3R / 10 d, 2 halves; L4 (172, 2021-26) +0.29R with a 20 d time exit vs −0.01R; capitulation alone: no exit rescues the entry since 2019; half at 1R and stop to entry ❌ in all 3 families · S10b (02/11, , 36 tickers with cached Tempest, 231 trades 2024-26): 10 d time exit +0.91R vs +0.38R for 3R / 10 d (halves +0.95 / +0.87), trailing stop after 1R +0.83R; stocks only +0.73R (100 trades, 58 dates), without VOO / SPX duplicates +0.80R; with G23 +1.33R (87); « any 10 d low » base +0.09R → the edge comes from the trigger, the exit doubles it

K1Pinned market—

Rule · dense positive gamma within ±0.5σ on both sides

Role · veto

K2Indices against the trade—

Rule · at least 2 of the 3 indices (SPX, SPY, QQQ) make a new session low or break their floor (long; mirror for a short)

Role · veto

K3Target already delivered—

Rule · the targeted node has already been touched today

Role · veto

K4Move already done—

Rule · the asset has already moved > 1σ in the trade's direction

Role · veto

K5Rainbow Road—

Rule · ≥ 6 large nodes of both colors within ±1.5%, no dominant King

Role · veto

K6Fib 0.786 broken on a close—

Rule · 3-min close below 0.786 before the trigger (for a fib discount)

Role · veto

K7Never long above the prior-day high, never short below the prior-day low (Léo's rule)—

Rule · applies to all setups, no exceptions

Role · veto

K8No index short in complacency—

Rule · G27: at the 10-day high with VIX9D/VIX3M, COR1M at the bottom of their range or SqueezeMetrics GEX at its high

Role · veto

K9No swing through earnings (Léo 15/10)—

Rule · no S10b / S10, S11, S13 if the company's earnings fall within the 10 holding sessions (≈ 14 calendar days) — G6: −0.43R vs +0.68R

Role · veto

L1Index low Aon PANDORA✅

Rule · (close < MA200 or ≥ 10% below the 252-day high) AND SKEW in its bottom 10% over 252 d AND SMH − QQQ over 5 d ≥ +2 pt

Role · trigger

Measure · ✅ 22/10/2026 , 13 ETFs, test 2020-26): +2.9 pt at 20 d vs normal

L2Index low Bon PANDORA✅

Rule · 10-session low AND RSP / SPY over 20 d in its bottom 10% AND monthly OPEX week

Role · trigger

Measure · ✅ 22/10/2026 : +5.0 pt at 20 d in the test

L3Index distribution Con PANDORA✅

Rule · close ≥ 98% of the 252-day high AND RSP / SPY over 20 d in its bottom 10%

Role · context / bias

Measure · ✅ 22/10/2026 : −2.0 pt at 20 d, −5.1 pt at 60 d, 13 / 13 assets; on SPY alone since 2021: holds at 60 d only

L4Stock low with off-exchange buying🟡

Rule · FINRA off-exchange share 30-day robust z ≥ 2 AND close ≤ 1.05 × 252-day low AND 20-day return ≤ −10 pt vs SPY (R3) or ≤ −8 pt vs its sector ETF (R4)

Role · trigger

Measure · 🟡 t+1 entry (06/11, : +3.18 pt (vs +2.96 at t); v2 🟡 (4.997 years of history, just below the 5-year ✅ threshold; pooled stocks → optimistic p) · before 06/11: ✅ 23/10/2026 : +3.0 pt at 20 d out of universe (398 S&P 400 stocks, n 617, t 5.6); K9 veto if earnings within 21 sessions

L5Clustered insider buying✅

Rule · ≥ 2 distinct insiders buy, total ≥ $100k, over 10 sessions (SEC Form 4)

Role · context / bias

Measure · ✅ weak 23/10/2026

L6Clustered insider buying near the lows✅

Rule · L5 AND (close ≤ 1.10 × 252-day low or ≤ 0.80 × MA200)

Role · trigger

Measure · ✅ 23/10/2026 : +2.2 pt at 20 d vs SPY net of placebo (test 2023-26, t 5.3)

L7Insiders + off-exchange buying✅

Rule · L6 AND L4

Role · trigger

Measure · ✅ fragile 23/10/2026: +3.8 pt at 20 d (45 episodes)

L8New AMF-declared short near the high✅

Rule · a holder goes from < 0.5% to ≥ 0.5% AND close ≥ 0.97 × 252-day high

Role · context / bias

Measure · ✅ 23/10/2026 : −1.4 / −3.3 / −4.4 pt; daily scan scripts/eu_shorts_scan.py

L9Crowded AMF short near the low✅

Rule · total declared shorts ≥ 3% AND close ≤ 1.10 × 252-day low

Role · context / bias

Measure · ✅ cautious 23/10/2026: bias from delisted companies missing from the test

L10Nasdaq turbulence (≥ 2 alerts out of 5)on PANDORA🟡

Rule · semis lagging QQQ · VIX9D > VIX · VVIX in its top 20% · HYG / IEF in its bottom 20% · SPX GEX < 0 or bottom 20%

Role · context / bias

Measure · 🟡 t+1 entry (06/11, : QQQ ≤ −2% open → t+1 close 6.6% vs 3.4%; v2 ✅ (47 episodes 21 sessions apart) → ✅ candidate (Léo review); signal of big sessions in both directions · before 06/11: 🟡 24/10/2026 : next session ≤ −2% 10.3% and ≥ +2% 11.2%; 47% near the lows → big sessions in both directions, not a top signal

L11Credit rotation🟡

Rule · ON: HYG / IEF over 20 d in its bottom 10% AND SPY < MA50; OFF: credit relaxed AND SPY > MA50

Role · risk / stop

Measure · 🟡 26/10/2026 : 2008-26 worst drawdown −36% vs −51.5%, Sharpe 0.77 vs 0.66; protection, not a predictor

L12Cboe put/call fear / complacencyon PANDORA⚪

Rule · 10-day average of total put/call ≥ 0.99 = fear; ≤ 0.82 = complacency

Role · context / bias

Measure · ⚪ 24/10/2026 : fear → S&P +1.2 / +1.6%; complacency +0.4 / −0.4%

L13Confluence: VIX curve capitulation (P14) + buy confluence (P19)on PANDORA🟡

Rule · (Léo 10/11: « if we have this signal, this signal and this signal, what does it give ») P14 and P19 each lit within the last 5 sessions, one of the two on the same day; entry at the t+1 open; 10 sessions between episodes

Role · trigger

Measure · 🟡 10/11/2026 , PANDORA chart 2021-09 → 2026-11, 225 combinations tested, Benjamini-Hochberg): 18 episodes, +21 d +2.59 pt above the base, 89% up, q 0.04; only pair that beats its members (P19 alone +2.11 pt); short history (5 years), gap vs P19 alone not tested; whole families (2 or 3 bullish signals within 3 or 5 sessions) ⚪

P1Gamma×Curve « release » (buy)on PANDORA⚪

Rule · SqueezeMetrics aggregate GEX proxy: exit from an extreme state, VIX curve normalizing

Role · trigger

Measure · ⚪ weak · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 31 cases, +1.33%, 58% up

P2Gamma×Curve « process » (sell)on PANDORA❌

Rule · same proxy, each qualified sell observation

Role · context / bias

Measure · ❌ as a sell · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 44 cases, the S&P rose 1.1% afterwards

P3Daily DPI divergence ▲on PANDORA🟡

Rule · S&P ≤ −0.5% in the session and DPI (DIX) ≥ +1 pt

Role · trigger

Measure · 🟡 · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 93 cases, +1.46%, 70% up

P4Daily DPI divergence ▼on PANDORA❌

Rule · S&P up and DPI down

Role · context / bias

Measure · ❌ as a sell · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 109 cases, S&P +1.2% afterwards

P5DPI absorption (accumulation line)on PANDORA⚪

Rule · new price low, off-exchange accumulation line (cumulative DIX − 252-day median) above its low

Role · context / bias

Measure · ⚪ weak · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 28 cases, +1.15%, 61%

P6DPI distribution (accumulation line)on PANDORA❌

Rule · new price high, accumulation line stalling

Role · context / bias

Measure · ❌ as a sell · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 69 cases, S&P up afterwards

P7Accumulation regimeon PANDORA⚪

Rule · DIX in its top quintile and GEX falling over 5 d (start of episode)

Role · context / bias

Measure · ⚪ weak · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 49 cases, +0.87%, 63%

P8GEX Crossover ▲on PANDORA🟡

Rule · very negative GEX then a sharp rebound, or 5-day average crossing back above zero

Role · trigger

Measure · 🟡 · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 25 cases, +3.12%, 68%

P9GEX Crossover ▼on PANDORA❌

Rule · first day of negative GEX after a positive phase

Role · context / bias

Measure · ❌ as a sell · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 40 cases, S&P +2.6% afterwards

P10GEX Flip (5-day average changing sign)⚪

Rule · total GEX EMA5: negative → positive = ▲, positive → negative = ▼

Role · context / bias

Measure · ⚪ (low n) · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 7 + 7 cases

P11Price·GEX (pivot divergence)on PANDORA⚪

Rule · lower low on the S&P and higher low on GEX (▲) or the reverse at the top (▼), confirmed 20 sessions later

Role · context / bias

Measure · ⚪ (low n) · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 6 + 6 cases

P12GEX quiet dropon PANDORA❌

Rule · fast GEX drop while the VIX does not move

Role · risk / stop

Measure · ❌ as an alert · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 17 cases, no decline afterwards

P13DIX × GEX capitulation (PANDORA « B1 », ≠ the lab's B1)on PANDORA🟡

Rule · ≥ 3 red days out of 10 with DIX ≥ 85th percentile 63 d and GEX SMI < −20

Role · trigger

Measure · 🟡 (1st day of episode, 06/11) · before: ⚪ weak · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 9 cases, +2.82%, 78% · 06/11 (PANDORA , SPX 2011-26, D+1 close entry, 21 sessions, 10 bp): every marked day (209 d, 28 episodes) +0.61 pt vs base, p 0.21 → ⚪; 1st day of episode +1.05 pt, halves +0.78 / +1.34, p 0.070 → 🟡 (28 episodes < 30) → chart: 1st day of episode only (PR #85)

P14VIX curve capitulation (PANDORA « B2 »)on PANDORA🟡

Rule · VIX9D / VIX3M > 1.05, signal on the first easing below 95% of the episode peak

Role · trigger

Measure · 🟡 · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 30 cases, +3.35%, 77% · re-measured on 10/11/2026 over the full history (PANDORA , entry at the next open, 21 sessions): 2011-26 153 cases, +2.01% against +0.81% all sessions (circular permutation p 0.013, 90% CI +1.3 / +2.7); 2011 → 2021-09: 115 cases, +1.61% against +1.04% (p 0.18); 2021-10 → 2026-10: 38 cases (not 30), +3.23%; 108 intervals out of 153 are shorter than 21 sessions (clustered signals); conditional « recent stress » benchmark (ratio > 1.05 in the last 10 sessions): +1.84% on 2011-26, so the trigger adds ≈ +0.2 pt to the stress state: to test against this benchmark before any ✅

P15Washouton PANDORA⚪

Rule · ≥ 4 of the 6 capitulation channels at the same time (DIX, GEX, VIX curve, breadth, credit, drawdown)

Role · trigger

Measure · ⚪ (low n) · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 5 cases, +6.83%, all up

P16VIX renormalizationon PANDORA⚪

Rule · exit from a lasting VIX curve inversion

Role · trigger

Measure · ⚪ (low n) · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 5 cases, +4.81%

P17Credit vetoon PANDORA⚪

Rule · S&P at a 6-month high while HYG / LQD lose > 2% over 63 d

Role · risk / stop

Measure · ⚪ (low n), contradicted · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): 7 cases, S&P +2.7% afterwards

P18CD13 (DeMark 13) ▲ / ▼, triggered and filtered version⚪

Rule · countdown 13 completed, validated or blocked by the PANDORA engine

Role · context / bias

Measure · ⚪ · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): buy 2 cases (low n), sell 9 cases weak (+0.1%)

P19Audited BUY confluenceon PANDORA🟡

Rule · ≥ 2 independent buy families on the same day or within the 2 prior sessions (capitulation, off-exchange, dealers, exhaustion)

Role · trigger

Measure · 🟡 · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): capitulation + off-exchange 23 cases, +4.45%, 87% (best combination, PASS)

P20Accumulation under stress (WATCH) and DPI WATCH ▲⚪

Rule · GEX proxy ≤ −2σ + participation rebound on a red session; S&P ≤ −0.30% and DIX ≥ +0.75 pt

Role · context / bias

Measure · ⚪ · DPI WATCH ▲ tested 06/11 (PANDORA , SPX 2011-26): 65 episodes, −0.31 pt vs base at 21 sessions, halves +0.25 / −0.89, p 0.75 → no edge → removed from the chart (PR #85, too noisy per Léo); accumulation under stress (WATCH) stays displayed, not tested

P21Small caps decoupled from value⚪

Rule · 60-day correlation of IWM / IVE returns < 0.65 (the « stretched small caps » version z IWM / SPY < −1.5 = F42 / H29)

Role · context / bias

Measure · ⚪ 28/10/2026 : rare (2000, 2017-18, 2026), small caps lag value afterwards, no clear direction for the S&P

P22v_joint protection regime (green / red bands)🟡

Rule · SMA200 trend + F13 gate (flow / gamma) + f5 gate (vol) on SPY; long 60% or cash

Role · risk / stop

Measure · 🟡 protection · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): Sharpe 0.93 vs 0.75 buy-and-hold, lower return; does not veto setups (F76)

P23VIX curve ribbon (contango)❌

Rule · VIX3M > VIX = contango

Role · context / bias

Measure · ❌ as a filter · PANDORA 5-year audit (PANDORA , 2021-10 → 2026-10, S&P at 21 sessions, sells sign-flipped): Sharpe 0.60 vs 0.75 buy-and-hold

P24Brief trades (« Brief trades » layer)🟡

Rule · S10b, agent 5, F35 setups sent in the brief, log closed every morning (stop, 3R, 10th session) + S10b backtest 2024-26

Role · trigger

Measure · 🟡 · live journal since 22/10/2026; S10b backtest SPY 22 trades +1.15R, QQQ 19 trades +0.93R

P25Extreme negative GEX (PANDORA GEX×DPI page) (post « SPY negative GEX at crisis levels », 06/11)⚪

Rule · SqueezeMetrics GEX < mean − 1.1σ; page: mean and σ of the window shown since 01/2025 (descriptive, uses the whole window); SPY dark pool panel = FINRA short ÷ total

Role · context / bias

Measure · ⚪ descriptive page (PANDORA PR #86, /gex-dpi): 63 sessions marked since 01/2025 (March-April 2025, March 2026, 05/11/2026). Causal version (σ and mean over the prior 252 sessions), quick measurement not pre-registered 06/11, SPX 2011-26, D+1 close → 21 sessions: all marked days +2.17% vs +1.04% (462 days, 45 episodes, halves +0.95 / +1.31 pt, p 0.007); 1st day of episode +1.81% (halves +1.51 / +0.03 pt, p 0.08); also marks 05/11 (causal z −1.23; correction 06/11: a 1st read said the opposite, the last days were excluded for lack of a known forward return) · with SPY dark pool (quick measurement 06/11, FINRA since 2018): day's DPI ≥ its 1-year median → +3.14% at 21 sessions (18 episodes, halves +2.06 / +1.96 pt, p 0.007); DPI < median → +1.46% (21 episodes, halves +0.10 / +0.25 pt, p 0.28); as of 05/11 SPY DPI 1-year rank 0.59 (favourable case) · GEX recovering above −1.1σ after < −2σ over 10 d: +2.49% (22 episodes, halves +1.14 / +1.78 pt, p 0.046); after < −2.7σ: 6 episodes, too few; close to F15 / F21 / F57

P4mDPI divergence ▼ over 2 to 5 sessions (lead R1)❌

Rule · S&P ≥ +0.5% and DIX ≤ −0.010 over k = 2..5 sessions (production function, lookback = k), 1st day of the episode, entry at the t+1 open

Role · context / bias

Measure · ❌ 13/11/2026 (PANDORA , SPX 2011-26, ~400 episodes per k): no significant spread at t+1/t+5/t+21, neither IS nor OOS (Bonferroni 15 trials)

P4nFiltered DPI divergence ▼: high DIX and/or price near the high (lead R1b)❌

Rule · P4m (k = 1..5) with A: DIX at t−k in the top 10% of 252 sessions; B: S&P within < 2% of the 252-session high; AB: both

Role · context / bias

Measure · ❌ 13/11/2026 (PANDORA : 60 trials with R1, none below p < 0.00083; A at t+1 ≈ −0.2 pt in both periods (p 0.03-0.13), unstable sign at 21 d; AB n < 30

P26GEX collapse from a top, DIX falling, S&P flat (lead R2)❌

Rule · on the evening of t: GEX of t−1 in the top 10% (or 20%) of 252 sessions, GEX(t) ≤ −20/30/40% vs t−1, DIX(t) < DIX(t−1), |S&P(t)| ≤ 0.5%

Role · trigger

Measure · ❌ 13/11/2026 (PANDORA , SPX 2011-26): 26-58 cases; next day −0.01 to −0.09% vs +0.05%, p ≥ 0.13 (threshold 0.0028); days ≤ −1% less frequent (0-8% vs 9-12%); fired on 12/11/2026

G41VIX / SPY opening gap alignment and VIX expanding or crushing (educational post, lead R4)⚪

Rule · H1: VIX gap ≥ ±5% opposite to the SPY gap ≥ ±0.3% (aligned) or in the same direction (misaligned) → SPY open→close; H2: after a VIX gap ≥ +5%, VIX below the prior close (crush) or above the open (expansion); H3: VIX open→11:00 ≥ ±3%

Role · trigger

Measure · ⚪ 13/11/2026 (PANDORA : 9 trials, none below p < 0.0056; alignment = continuation in 2004-15 (−0.13 / +0.19%), gone since; H2 n < 30 (2 years of intraday); H3b VIX ≥ +3% → short 11:30→close +0.09 / +0.07 pt (p 0.17 / 0.10) = lead

G42Negative VRP: VIX below 20-day realized vol (forensic 13/11, lead F1)⚪

Rule · VIX(t) − 20-day S&P realized vol in its bottom 10% of 252 sessions, 1st day of the episode, entry at the t+1 open

Role · context / bias

Measure · ⚪ 13/11/2026 (PANDORA , SPX 2011-26): t+1 −0.28 pt (IS, n 55, p 0.015) and −0.14 pt (OOS, n 30, p 0.18), threshold 0.0033 (15 trials); lit on 12/11 (Tempest: rank 1%)

H39Long rates and dollar rising fast (forensic 13/11, lead F2)⚪

Rule · 5-session change of the 10-year (^TNX) AND the dollar (DX-Y.NYB) both in their top 10% of 252 sessions, 1st day of the episode, entry at the t+1 open

Role · context / bias

Measure · ⚪ insufficient sample 13/11/2026 (PANDORA : t+1 −0.55 pt (IS, n 22, p 0.005) and −0.20 pt (OOS, n 29, p 0.10), t+5 negative in the 2 periods; not significant after correction; lit on 12/11

H40Semis stretched vs Nasdaq over 5 sessions (forensic 13/11, lead F3)❌

Rule · SMH − QQQ over 5 sessions in its top 10% of 252 sessions, 1st day of the episode, entry at the t+1 open (link H33, 63-day horizon)

Role · context / bias

Measure · ❌ 13/11/2026 (PANDORA : t+1 +0.04 / +0.01 pt, t+5 +0.23 / −0.16 pt (IS n 119 / OOS n 66), p ≥ 0.27; lit on 12/11

E26High Cboe equity put/call near the high (forensic 13/11, lead F4)⚪

Rule · Cboe equity put/call in its top 10% of 252 sessions AND S&P within ≤ 3% of its 252-day high, 1st day of the episode, entry at the t+1 open (≠ E23, at the low)

Role · context / bias

Measure · ⚪ 13/11/2026 (PANDORA , data since 10/2019): IS 2020-22 n 16 (+0.09 pt), OOS 2023-26 n 35 −0.23 pt at t+1 (p 0.04, threshold 0.0033); lit on 12/11

E27Big option unload: single-leg print ≥ $5M sold and closed (OI T+1) on SPY / SPXW / QQQ (Léo 19/11 « a guy unloading a big series of puts can make a bottom on his own »)⚪

Rule · one trade ≥ $5M premium 09:30-16:00 ET, crosses removed, not a package leg (isMultiLeg or another ≥ $1M trade in the same second), price ≤ 20 % of the bid-ask spread (sold), and the contract's open interest the next session down by ≥ 50 % of that day's big-print contracts (closing); put → PUT_UNLOAD, call → CALL_UNLOAD; dated on the print day, known the next morning

Role · context / bias

Measure · ⚪ 19/11/2026 (skylit-lab , 2026-07-01 → 11-17, 848 credits): 4,073 prints ≥ $5M, 95 % package legs; 4 single-leg put unloads (QQQ 3, SPY 1), none at an 11-session low (base 22-38 %); v2 , 2026-03-02 → 11-17, D+1 open entry, context-matched baseline, 788 credits): 6 put / 7 call single-leg unload days (n too small), big put closings incl. package legs 21-25 per ticker: no cell 🟡 of 36, halves of opposite signs, put closings at a low less often than context (SPY 29 % vs 41 %); history seeded from 02/03/2026; phase 2 , 2024-06-03 → 2026-11-17, ≈ 3,690 credits, same rules): 29,853 prints, big put closings with package legs 68-78 per ticker → +0.70 / +0.36 / +0.85 pt vs context at 20 s. (SPY / QQQ / SPXW), SPXW p 0.029 with both halves positive = 🟡 fragile (1 cell of 36, Bonferroni 0.0014 not met; 2024-06 → 2026-02 alone +0.70, p 0.10, halves −0.05 / +1.45); closings mark the low less often than context; call closings followed by MORE rise (no top); single-leg unloads still n < 10; recorded forward by → data/big_prints.csv; on the SPX / QQQ charts since 19/11/2026 (Léo's go), not in the brief

E28Big SPXW put closing: put print ≥ $5M (package legs included) whose OI falls ≥ 50 % of its size the next session (R10 phase 2)🟡

Rule · at least one SPXW put print ≥ $5M (09:30-16:00 ET, crosses removed, any leg type) with the contract's next-session oiChange ≤ −50 % of that day's big-print contracts on it; at most one mark every 5 sessions (gap from the last kept day); dated on the print day, known the next morning; entry next open

Role · trigger

Measure · 🟡 fragile 19/11/2026 (skylit-lab , 2024-06-03 → 2026-11-17, ≈ 3,690 credits): 68 cases, SPY open D+1 → close +20 sessions +0.85 pt vs sessions with the same distance-from-high quintile and 5-session sign, halves +1.06 / +0.64, p 0.029 one-sided (1 cell of 36, Bonferroni 0.0014 not met); 2024-06 → 2026-02 alone +0.70 pt, p 0.10, halves −0.05 / +1.45; not a bottom marker (31 % within 1 % of the 11-session low vs 37 % context); SPY +0.70 (p 0.07), QQQ +0.36 (p 0.31) ⚪; on the S&P charts since 19/11/2026 (Léo's go), not in the brief

E29Big position loaded / unloaded (SPY+SPXW ≥ $50M, QQQ ≥ $25M per contract-day, OI T+1 confirmed)⚪

Rule · contract-day whose prints ≥ $5M (any leg type) add up to ≥ $100M on SPY / SPXW (was $50M until 19/11/2026 noon) or ≥ $25M on QQQ, read « opening » (load) or « closing » (unload) from the next-session open interest (±50 % of the printed size); hover lists contract, premium and OI change; thresholds chosen for ≈ 1-2 marks a week, not by results

Role · context / bias

Measure · ⚪ display only, not tested as a signal (Léo 19/11/2026: « a big boy loading or unloading always has a reason »); 06/2024 → 11/2026: S&P 128 load days / 31 unload days at $100M (215 / 77 at $50M); in the Essential view since 19/11/2026; on the SPX / VOO / IVV / QQQ charts since 19/11/2026, not in the brief; related tests R10 (E27 / E28)

F85Low FINRA composite DIX (forensic 13/11, lead F5)❌

Rule · dix_composite from data/finra_dp.csv (24 stocks) in its bottom 10% of 252 sessions, 1st day of the episode, entry at the t+1 open (link F9 / F11)

Role · context / bias

Measure · ❌ 13/11/2026 (PANDORA : IS 2020-22 next days ≤ −1% 24% vs 11% (p 0.03), OOS 2023-26 3% vs 10%, t+5 +0.29 pt: opposite directions; lit since 09/11

P27DPI pre-signal ▲ (« pre shot », hollow yellow dot)⚪

Rule · S&P down 0.4% to 0.5% in one session and DIX ≥ +0.010 (just below the confirmed divergence P3, mutually exclusive with it); display only

Role · context / bias

Measure · ⚪ 14/11/2026 (SPX 2011-26 measurement, t+1 open entry, 54 cases): 2011-20 (33 cases) next day −0.15% vs +0.02%, 21 sessions +1.33% vs +0.97% (p 0.32); 2021-26 (20 cases) next day +0.19% vs +0.02%, 21 sessions +2.39% vs +1.07% (p 0.07); small sample, not proven. Fired on 13/11/2026

P28DPI pre-signal ▼ (mirror of P27), not displayed❌

Rule · S&P up 0.4% to 0.5% in one session and DIX ≤ −0.010 (just below the ▼ divergence P4)

Role · context / bias

Measure · ❌ 14/11/2026 (SPX 2011-26, t+1 open entry, 54 cases): 2011-20 (36) next day −0.07% vs +0.02%, 21 sessions +0.18% vs +0.97% (p 0.13); 2021-26 (18) reverse direction: 21 sessions +2.03% vs +1.07%, 17% down → not a sell signal; not displayed (quant's recommendation, Léo agrees)

P29Location base score: P(swing low within 10 sessions) from the S&P's location alone (confluence audit step 4)🟡

Rule · cell = quintile of the distance to the 252-session high (edges on the training sessions) × sign of the 5-session return; P = frequency of « swing low (lowest close ±10 sessions, then +5 % within 20) dated t … t+10 » in that cell over the sessions ≤ last − 31, shrunk toward the base rate (k = 20)

Role · context / bias

Measure · 🟡 19/11/2026 (skylit-lab , walk-forward 2015-2026, each year from the past only): AUC 0.671, Brier skill +7.8 % vs the base rate; top decile over-states (46 % predicted, 36 % realised), bottom decile under-states (6 % / 16.5 %); P(top) tested and rejected (AUC 0.614); 18/11/2026: 38 % vs 19 %; untouched test 1990-2010 (skylit-lab , model frozen on 2011-22): AUC 0.691 but high readings over-state (deciles 8-9: 38-43 % predicted, 24-29 % realised) → read as a ranking, not an exact probability; entering on P29 ≥ 30 % alone +0.2 pt vs any day at 20 sessions (p 0.29), no price confirmation helped (−0.6 to −1.0 pt on 2011-22); shown under the regime at the top of the S&P charts and one brief line since 19/11/2026 (Léo's go)

H41CTA weekly selling ≥ $22.5bn (Goldman) → low is in? (lead R8)❌

Rule · Goldman one-week change in CTA SPX positioning ≤ −$22.5bn (5 dates read off the chart) and a price-only CTA proxy (trend 20/60/120/250 × vol target 15 %, weekly change ≤ 2nd percentile), plus Goldman's flat-tape buying projection

Role · context / bias

Measure · ❌ 18/11/2026 (PANDORA , SPX 1928-2026, 0 credits): 5 Goldman cases up 2 of 5 at 5/10/20 sessions, mean 0.0 % at 20 vs +1.4 %; proxy 203 events no better than random (p(up) ≥ 0.42); projected CTA buying 20 sessions −0.54 % vs +0.63 %; 4 of 5 dipped 3-6 % within 20 sessions; fired 16/11/2026

A35S&P closes under a level it held 5 sessions → more likely lower (lead R6)❌

Rule · close(t) < L ≤ close(t−1), closes of the 5 prior sessions ≥ L; L = multiple of 100 or 250 points, or the lowest close of the prior 20 / 50 sessions; 10-session cooldown

Role · context / bias

Measure · ❌ 18/11/2026 (PANDORA , SPX 1950-2026, 0 credits): round-level breaks followed by better than any day (round250 next day +0.65 % vs +0.04 %, p(down) 1.0); 20/50-session-low breaks next day −0.07/−0.09 % before 2021 (p 0.009/0.028 > 0.0042), gone since; fired 16/11 (7,500) and 17/11/2026

L14Round-level break absorbed by dark buying (R6 × F13/F62, lead R7)on PANDORA🟡

Rule · S&P closes under a multiple of 100 it held 5 sessions (R6 round100) and DIX ≥ 0.45 the same evening; long close t → t+20; 10-session cooldown

Role · trigger

Measure · 🟡 18/11/2026 (PANDORA , SPX 2011-26, 0 credits): 43 cases, 20 sessions +2.91 % vs +0.98 %, up 77 % vs 67 %, p 0.001 two-sided (Bonferroni 0.0025 over 20 crosses); 2011-20 +4.68 % (n 7) and 2021-26 +2.56 % (n 36); 15 variants DIX 0.43-0.47 × held 4-6 all +1.9 to +3.4 %; not fired 16/11/2026 (DIX 0.409); on the S&P chart since 18/11/2026 (Léo's go), not in the brief

H42Index / QQQ at its high while most sectors are under their 20-day average → dump (Skylit post, lead R9)⚪

Rule · count of the 9 SPDR sectors above their 20-session average ≤ 2 / 3 / 4 while SPY or QQQ closes at a 52-week or all-time high; 10-session cooldown

Role · context / bias

Measure · ⚪ 18/11/2026 (PANDORA , 1999-2026, 0 credits): rare and recent (SPY 0 cases before 2021, 11 since); no rule reaches 30 cases; strictest versions lean weaker (SPY ≤ 3/9: 20 s. +0.62 % vs +1.04 %, dips ≥ 3 % 40 % vs 31 %; QQQ ≤ 3/9 52w −3.01 % / −0.42 %), loosest show nothing; not met on 17/11/2026 (3/9 but indices below their highs)

E30Aggregated large options activity: fills ≥ $1M summed per contract and session (SPY+SPXW ≥ $50M, QQQ ≥ $25M shown)⚪

Rule · per contract and session, regular-session non-cross trades ≥ $1M summed (a parent order split into many fills counts as one), stored from $5M (data/big_aggregates.csv), shown from $50M on SPY / SPXW and $25M on QQQ; hover: contract, combined premium, fills, contracts, share at the ask / at the bid / mid (apparent aggressiveness, not trader identity or intent), next-session OI change and its read (±50 %); grey ring, no direction; dated on the print day, known and actionable from the next session's open; thresholds set for frequency, not by results

Role · context / bias

Measure · ⚪ context only (Léo 19/11/2026, Analysis view, never Decision): lab skylit-lab , 2024-06 → 2026-11 — aggregated SPXW put closings +0.75 pt at 20 sessions 🟡 (same as E28), the days aggregation adds are noise (2 🟡 of 18 trials ≈ chance, none Bonferroni); catches split orders such as 18/11/2026 SPXW Nov 20 7,470P ($52.6M seen in ≥ $1M fills); not in the brief

Personal decision-support tool, not investment advice. Past results do not guarantee future results; a trade can lose more than the planned risk on an opening gap. Glossary