INVERSION v1.6.0
generated 2026-11-19 05:13Z
9,186 observed exits · post-gate

Inversion — start from the winners

Feature-first analysis asks "does X predict Y" and drowns in multiple testing — 3,839 combos were tested on 260726 and exactly one survived. Inversion starts from the outcome and works backwards. This page unifies both inversion lineages on this box: the on-chain pre-pump signature lift (accum_explosion_forensic, case-control vs look-alike duds) and the realised-shape inversion (chart_archetypes).

retrospective timing finding 1–4 h highest observed P(peak≥50%): 37% · n=817; descriptive, not an entry rule
the book's biggest bucket 37% of measured timing rows peak inside 5 minutes · expectancy −$9/trade
proof boundaryRESEARCH ONLY retrospective shapes and prospective archetypes are visibly separated; decision effect: NONE
top-decile cutoff +72% peak needed to enter the winners cohort (n=917)

P−1 Decision inversion — taken vs eligible skipped

M-20 is now measured inside the canonical universal paper path: point-in-time TDE evidence rides the same position and realised exit, while skipped selection quality uses the same fixed 4h outcome lens. Hard-gate rejects are excluded from hesitation arms.

eligible taken · fixed 4h -0.05% resolved 1,557/23,567 · coverage 6.6%
eligible skipped · fixed 4h -0.10% resolved 262/7,231 · coverage 3.6%
measurement-law diagnosis unknown_low_coverage WITHHELD — either cohort is below coverage/N gate
decision effectNONE shadow challenger · no admission, size, exit or live execution authority

Do not read the descriptive EV gap as hesitation skill yet. Coverage is 6.6% taken versus 3.6% skipped; the comparison remains selection-biased until both clear the 80% gate. Canonical source: paper_scorecard.json → decision_quality.m20_attention.

P0 Formal post-pump shelf classifier — point-in-time

A plateau is not automatically bearish. Low volume while price holds a shallow, compressed range above anchored VWAP can be constructive supply dry-up; low-volume breakout is weak; high volume with no progress and negative flow is distribution. The classifier uses only bars visible at classification time and stays shadow-only until forward proof passes.

CONSTRUCTIVE ABSORPTION
tight hold · volume dries · higher lows
BULL-FLAG BREAKOUT
compressed shelf · accepted breakout
NO-DEMAND SHELF
flat + quiet · direction unresolved
DISTRIBUTION PLATEAU
high effort · no result · negative flow
FAILED SHELF
breakout rejected · support lost
UNSTABLE CONSOLIDATION
pause too deep/wide/loud · directionally unresolved
ORDINARY CHOP
no qualifying pole · not a post-pump setup at all

green = continuation structure · amber = unresolved · red = reversion/failure risk. Thin bars are normalized volume; red bars illustrate sell-dominant flow. Schematics—not historical outcomes.

shapemeasurable structurereadconfirmation / invalidation
CONSTRUCTIVE_ABSORPTIONstrong pole · shallow retrace · range+volume contract · AVWAP hold · higher lowswait / continuation candidateclose above shelf high with renewed positive volume/flow
BULL_FLAG_BREAKOUTconstructive shelf + close above resistancecontinuation candidateparticipation required; fail if close returns inside shelf
NO_DEMAND_SHELFflat price + dry volume, weak support evidencedirection unknownwait for directional close + flow
DISTRIBUTION_PLATEAUhigh effort/no result · negative signed volume · flat/lower highsreversion risksupport break confirms; positive-flow upside break negates
FAILED_SHELFclose below support or deep surrender of polereversion risk confirmedmust reclaim shelf low/AVWAP before bullish reuse
UNSTABLE_CONSOLIDATIONpause too deep, too wide, too loud, or directionally unresolved (catch-all after a qualifying pole)waittighten above VWAP for continuation, or close below support for failure
ORDINARY_CHOPno qualifying impulse — pole_return below the pump-floor priorno edgenot a post-pump setup; wait for a new impulse or an independently validated catalyst

Proof: HYPOTHESIS_SHADOW_ONLY. Starting thresholds are priors, not facts. Required: timestamped OHLCV, matched controls, sealed temporal OOS, cost/latency stress, chain×age×liquidity×regime strata, BH-FDR, neighboring-parameter stability, forward paper challenger, then Chris approval.

P0b Broader chart-pattern taxonomy — chart_pattern_intel (12-class, prospective)

P0 covers the post-pump SHELF specifically. chart_pattern_intel.py (v1.0.0, wired into ta_capture.py/token_ta.py/trading_decision_engine.py since 260731) classifies the broader visible chart into 12 additional shapes — acceptance/rejection of prior value, liquidity sweeps, parabolic exhaustion, stair-steps, balance rotation — using the same point-in-time discipline. Until this section existed these lived only as invisible fields (chart_archetype/_bias/_tags) inside token_ta_cache.json; nothing rendered them anywhere.

VALUE ACCEPTANCE UP
CONTINUATION_CANDIDATE · breaks & holds outside prior balance, volume confirms
VALUE ACCEPTANCE DOWN
DOWNTREND_CONTINUATION · breaks & holds below prior balance
FAILED BREAKOUT UP
REVERSION_RISK · closed outside value, then closed back inside
FAILED BREAKDOWN RECLAIM
BOUNCE_CANDIDATE · closed below value, then reclaimed it
UPSIDE LIQUIDITY SWEEP
REVERSION_RISK · wick above prior high, close rejects back down
DOWNSIDE LIQUIDITY SWEEP
BOUNCE_CANDIDATE · wick below prior low, close rejects back up
PARABOLIC EXHAUSTION
NO_CHASE_TRAIL · accelerating return, upper wick or volume spike
VOLATILITY CONTRACTION
WAIT_FOR_EXPANSION · range and volume both shrinking vs prior window
STAIR STEP UP
TREND_CONTINUATION · higher lows, steady positive slope
STAIR STEP DOWN
AVOID_OR_EXIT · lower highs, steady negative slope
BALANCE ROTATION
MEAN_REVERSION_ONLY · flat slope, range inside prior balance
UNCLASSIFIED TRANSITION
WAIT · no rule matched — genuinely ambiguous

green = continuation/bounce bias · amber = wait / no-chase · red = reversion risk. Schematics—not historical outcomes, same as P0. Score is a rule-match strength, not a win probability (score_semantics: RULE_MATCH_NOT_WIN_PROBABILITY) — the classifier has no admission, sizing, or exit authority; proof status is HYPOTHESIS_SHADOW_ONLY pending the same 9 gates as P0.

P0c POINT-IN-TIME PROBABILITY LAB

Frozen chart state at T joined to forward outcomes by explicit horizon. Raw strategy arms collapse to unique economic decisions. RESEARCH ONLY · decision effect: NONE. A rule-match score is never relabelled as a win probability.

rows → decisions3,369 → 1,581duplicate arms removed: 1,788
PIT eligible349UNKNOWN archetypes: 1,049
outcome provenance quarantine0legacy exact-zero labels withheld as ambiguous, never treated as flat returns
promotionBLOCKED6 open proof gates · locked-forward false

Outcome provenance is a hard gate. Required next: canonical measured-return receipts, one-for-one matched controls, costs/fill stress, declared trials, and locked-forward evidence. Source: chart_archetype_probability_report.json; schema chart_archetype_probability_lab/1.0.0.

00 At a glance

Slices are share of trades (always a positive count, so the geometry is honest). Colour and the legend show whether that cohort makes or loses money — 🟢 profitable, 🔴 losing. A pie cannot draw a negative slice, so signed magnitudes are shown as diverging bars underneath instead of being faked into wedges.

Book by chart archetype

FLATLINE: 4,505 trades (50.3%) · −$4/tradeCHOP: 1,098 trades (12.3%) · $1/tradePOP_FADED: 981 trades (10.9%) · −$4/tradeSPIKE_FADE: 509 trades (5.7%) · −$16/tradeDOA: 476 trades (5.3%) · −$31/tradeGRIND_DOWN: 420 trades (4.7%) · −$21/tradeRUNNER_BANKED: 417 trades (4.7%) · $52/tradePOP_BANKED: 333 trades (3.7%) · $18/tradeRUNNER_SCRATCH: 161 trades (1.8%) · $2/tradeRUNNER_GAVE_BACK: 49 trades (0.5%) · −$18/tradeSUSPECT_PRICE: 12 trades (0.1%) · $433/trade8,961trades
  • FLATLINE 50% · 4,505 🔴 −$4
  • CHOP 12% · 1,098 🟢 $1
  • POP_FADED 11% · 981 🔴 −$4
  • SPIKE_FADE 6% · 509 🔴 −$16
  • DOA 5% · 476 🔴 −$31
  • GRIND_DOWN 5% · 420 🔴 −$21
  • RUNNER_BANKED 5% · 417 🟢 $52
  • POP_BANKED 4% · 333 🟢 $18
  • RUNNER_SCRATCH 2% · 161 🟢 $2
  • RUNNER_GAVE_BACK 1% · 49 🔴 −$18
  • SUSPECT_PRICE 0% · 12 🟢 $433

Book by feed

volume_path: 2,426 trades (31.2%) · −$3/tradeknown_conv: 1,254 trades (16.1%) · −$2/tradewhalemover_rh: 1,110 trades (14.3%) · −$2/tradesource_strategy: 824 trades (10.6%) · −$2/tradewallet_intel: 603 trades (7.8%) · −$4/tradesa_accum: 485 trades (6.2%) · −$4/tradewhalemover: 431 trades (5.5%) · −$2/traderanker: 325 trades (4.2%) · −$5/tradeclu:fomo: 316 trades (4.1%) · −$7/trade7,774trades
  • volume_path 31% · 2,426 🔴 −$3
  • known_conv 16% · 1,254 🔴 −$2
  • whalemover_rh 14% · 1,110 🔴 −$2
  • source_strategy 11% · 824 🔴 −$2
  • wallet_intel 8% · 603 🔴 −$4
  • sa_accum 6% · 485 🔴 −$4
  • whalemover 6% · 431 🔴 −$2
  • ranker 4% · 325 🔴 −$5
  • clu:fomo 4% · 316 🔴 −$7

Book by shape timing

<5 min: 3,430 trades (37.4%) · −$9/trade>4 h: 1,573 trades (17.1%) · $2/trade15–30 min: 1,371 trades (14.9%) · $0/trade5–15 min: 1,150 trades (12.5%) · −$4/trade30–60 min: 833 trades (9.1%) · $1/trade1–4 h: 817 trades (8.9%) · $2/trade9,174trades
  • <5 min 37% · 3,430 🔴 −$9
  • >4 h 17% · 1,573 🟢 $2
  • 15–30 min 15% · 1,371 🟢 $0
  • 5–15 min 13% · 1,150 🔴 −$4
  • 30–60 min 9% · 833 🟢 $1
  • 1–4 h 9% · 817 🟢 $2

Expectancy per trade — signed, so bars not wedges

SUSPECT_PRICE$433
RUNNER_BANKED$52
POP_BANKED$18
RUNNER_SCRATCH$2
CHOP$1
POP_FADED−$4
FLATLINE−$4
SPIKE_FADE−$16
RUNNER_GAVE_BACK−$18
GRIND_DOWN−$21
DOA−$31

01 Retrospective timing — when did the eventual peak arrive?

time_to_peak contains future information. It describes outcomes but does not prove the chart was flat at minute 15 and cannot authorize a live delayed-entry rule by itself. Keep the Wilson intervals and ex-top-1 check as descriptive evidence; use P0 plus prospective snapshots for prediction.

peak arrivesnP(peak ≥ +50%) 95% confidence intervalexpectancyex-top-1verdict
<5 min3,4301% [1–1%]−$9−$9loses money
5–15 min1,1509% [8–11%]−$4−$5loses money
15–30 min1,37110% [9–12%]$0−$0loses money
30–60 min83316% [13–18%]$1$1robust — survives ex-top-1
1–4 h81737% [33–40%]$2$1robust — survives ex-top-1
>4 h1,57336% [33–38%]$2$2robust — survives ex-top-1

◆ Strong separation, but retrospective

The intervals for <5min [22–25%] and 15–30min [41–53%] do not overlap — that is a clean statistical separation, not noise. And the positive bands stay positive with their best trade removed, which almost nothing else in this book does.

Safe reading: trades whose eventual peak occurred immediately performed worse in this book. Unsafe reading: assuming every late peak had visibly plateaued early. P0 now measures the actual early geometry and volume state without that hindsight substitution.

02 Inversion by configuration — what do winners have in common?

Top decile by peak (n=917, peak ≥ +72%) versus everything else (n=8257). Lift is percentage-point over-representation among winners. A lift above ~6pp is worth a look; below that, treat as noise at this sample size.

exit profile

valuein winnersin restliftskew
SWING_ARM2143.5%18.4%+25.1pp
SWING_SL2529.8%16.5%+13.2pp
SWING10.3%2.9%+7.4pp
TIGHT56.7%1.8%+4.8pp
SCALP-RATCHET0.2%0.9%-0.7pp
SCALP0.4%3.8%-3.4pp

source / feed

valuein winnersin restliftskew
ranker10.7%2.7%+7.9pp
whalemover_rh15.9%11.6%+4.3pp
wallet_intel9.5%6.2%+3.3pp
scout4.1%1.5%+2.7pp
sa_accum7.4%5.1%+2.4pp
alpha_brief4.4%3.1%+1.3pp

chain

valuein winnersin restliftskew
robinhood28.4%18.9%+9.5pp
sol26.7%24.9%+1.8pp
bsc24.3%23.0%+1.3pp
eth15.5%15.0%+0.5pp
avalanche0.0%0.0%-0.0pp
base5.1%18.2%-13.1pp

regime at entry

valuein winnersin restliftskew
poor57.9%49.1%+8.8pp
extreme_poor8.8%10.5%-1.7pp
ground33.3%40.4%-7.1pp

03 Realised shape archetypes

Every closed position classified by what it actually did. Post-hoc by construction — you cannot filter for RUNNER_BANKED at entry. What this does is size each problem: which shapes carry the book and which bleed it.

archetypenexpectancy ex-top-1win ratearmed median peakrobust?
SUSPECT_PRICE12$433$33367%100%0%robust +
RUNNER_BANKED417$52$50100%99%195%robust +
POP_BANKED333$18$1899%100%68%robust +
RUNNER_SCRATCH161$2$260%100%154%robust +
CHOP1,098$1$157%14%8%robust +
POP_FADED981−$4−$413%100%44%negative
FLATLINE4,505−$4−$41%0%-1%negative
SPIKE_FADE509−$16−$168%10%8%negative
RUNNER_GAVE_BACK49−$18−$180%100%143%negative
GRIND_DOWN420−$21−$210%11%12%negative
DOA476−$31−$310%0%-1%negative

◆ The book splits in two

04 Pre-pump on-chain signature lift (case-control)

From accum_explosion_forensic: exploded tokens versus look-alike duds, so each pattern gets a real lift rather than a survivorship artifact. This answers a different question from the shape work above — not "what shape won" but "what was visible on-chain before it moved".

signaturein explodedin dudsliftn explodednovel?
REACCUM4.2%0.3%16.1×13known
RANGE3.5%0.4%9.8×11known
CLUSTERED-SUPPLY2.9%0.5%6.2×9known
FRESH-SWARM4.5%1.0%4.6×14known
GMGN-CLUSTER5.2%1.1%4.5×16known
VOL-ANOM3.2%0.7%4.4×10known
REPEAT7.1%1.7%4.1×22known
HL3.5%0.9%4.0×11known
*SA-REPEAT≥1011.9%9.2%1.3×37NOVEL
*SA-CONVERGENCE≥35.2%4.0%1.3×16NOVEL
*SA-CONVERGENCE≥210.3%10.9%0.9×32NOVEL

Read these as leads, not rules. Several have n_exploded under 30, so the lift is directional. Note the standing finding from that lineage: features that look like drags on win-rate (REACCUM, HOLDER-SURGE, VOL-ANOM) are among the top explosion-lift tails — so do not down-weight a feature on win-rate alone.

05 Who's behind the winners — most-profitable wallets

The operators, not the tokens. Graded S/A wallets by realised 30-day P&L, reused from the ALPHA leaderboard (not recomputed). This is the money to copy-time into the setups above — the same wallets whose buys now name every ACCUM card.

S: 1 wallets (100.0%) · $20/trade1wallets
  • S 100% · 1 🟢 $20
walletgrade30d realisedroi 30dwin
kien28.base.ethNguyeEarlyWhaleS$394,371+20%56%
0x50dcb3f5?$299,912+16%44%
0xcf5a3392?$234,941+13%51%
cDF2-7k-AFeng阿峰_Dailyflipper?$226,728+15%68%
0x05a94a8d?$207,590+11%54%
BaseWhale枯坐p小将3-10X5FigManyFol?$156,147+46%73%
0x1057c855?$126,543+6%45%
Paingelz?$105,660+37%47%
AntPositions?$103,581+5%37%
yukaz?$85,409+40%59%
西瓜🔶BNB?$76,093+8%44%
blockdao🇨🇳 下次钻石手?$60,792+9%43%

06 The playbook — how to trade what the inversion shows

Every line is a direct read of the sections above, not generic advice.

ENTRY · timing hypothesis observe, do not gate Timing is retrospective. P0c must prove matched delay×horizon lift before any entry rule.
EXIT · profile SWING for runners, not TIGHT5 SWING is +30pp in top-decile winners; TIGHT5 is −28.6pp (Section 02). A tight trail books the fast-faders but amputates the tail that carries 75% of profit.
AVOID · admission DOA + FLATLINE ≈ 32% of the book These never tick up (−$60 / −$24). No exit rule saves them — only not entering. This is the admission layer, not a trade.
SIGNAL · pre-pump tails size small, let them run REACCUM 27×, VOL-ANOM 31×, HOLDER-SURGE 21× lift (Section 04): rare but hugely over-represented in explosions. Low win-rate is expected — asymmetric. Never down-weight on win-rate.

07 The winners themselves — top 3+ per chain, entry→peak→exit

The actual trades we learn from, guaranteeing every chain is represented. ● entry · ● peak · ● exit — see how much of each runner we captured (captured) versus what the peak offered (peak). The gap is the exit problem. Robinhood carries Monaco/Pow cabal alpha — the whole point of inversion.

SOL · 8 runners

tokenpeakcapturedsourceentry → peak → exit
$CLIP+1917%+1271%scoutentrypeak +1917%exit
$EEpng7+1794%+798%volume_pathentrypeak +1794%exit
$FOOD+1788%+1770%source_strategyentrypeak +1788%exit
$HOOKED+1480%+986%wallet_intelentrypeak +1480%exit
$HOOKED+1480%+571%wallet_intelentrypeak +1480%exit
$OP+1315%+589%scoutentrypeak +1315%exit
$MetaMask+1230%+48%scoutentrypeak +1230%exit
$MetaMask+1230%+48%scoutentrypeak +1230%exit

BSC · 3 runners

tokenpeakcapturedsourceentry → peak → exit
$0xb825+1477%+1461%alpha_briefentrypeak +1477%exit
$0xc31a+1327%+566%known_conventrypeak +1327%exit
$年年+909%+388%wallet_intelentrypeak +909%exit

BASE · 3 runners

tokenpeakcapturedsourceentry → peak → exit
$0xb200+798%+343%known_conventrypeak +798%exit
$BLOHARD+312%+55%known_conventrypeak +312%exit
$0x8fe5+266%+77%rankerentrypeak +266%exit

ETH · 3 runners

tokenpeakcapturedsourceentry → peak → exit
$DACITY+846%+146%rankerentrypeak +846%exit
$VERIDIA+836%+523%sa_accumentrypeak +836%exit
$0x6434+813%+351%rankerentrypeak +813%exit

ROBINHOOD · 3 runners

tokenpeakcapturedsourceentry → peak → exit
$Kimchi+1612%+324%whalemover_rhentrypeak +1612%exit
$?+1601%+7%rankerentrypeak +1601%exit
$PUNKS+1248%+413%wallet_intelentrypeak +1248%exit

How to use this page

inversion_gen.py v1.6.0 · shape + config inversion from chart_archetypes.json · pre-pump lift from accum_explosion_forensic.json · 9,186 observed exits (DEAD/UNRESOLVED excluded) · graded on post-gate cohort (entry ≥ 2026-09-21 UTC): 9,186 legs; 18,123 pre-gate legacy legs excluded from grading (kept in ledger)
Claude → docker exec <gw> python3 /opt/data/scripts/inversion_gen.py · Hermes → cron 6 */4, thread 37479