What the desk passed on.

Holdoween printed first at score 38.7 / MC 233133 on solana and was held to watch because the tape read 46.6 across three families (liquidity, wallet, attention) yet never cleared the email-line threshold. Early narrative and social-tape warmth were present but lacked an independent attention spike or social/news tag at detection, so the desk treated the print as incomplete. Flow/accel only appeared later; the token still advanced sharply without the high-confidence gate ever opening. The repeatable miss was requiring full email-line confirmation after an already multi-family tape instead of weighting early attention continuity more heavily. Pattern: multi-family tape < email line + delayed independent attention = systematic watch hold

Should have caught: First eval scored 46.6 with liquidity, wallet, attention — a lower confirmation floor or earlier attention weight would have held it.

PEPENOM printed first at score 41.2 / MC ~68k with only liquidity and wallet families active and the email-line threshold uncleared, so the desk correctly tagged watch. Buy-side pressure was already visible yet attention never independently spiked and no social/news tag appeared at detection, leaving the modest first-touch score unconfirmed. Flow/accel later registered the same day while the name moved ~209% without ever satisfying the high-confidence gate. The process error was treating a two-family tape print as insufficient rather than lowering the confirmation floor or weighting early smart-money pressure more heavily when narrative was already animal-meme. Pattern: two-family first print + uncleared email line + absent attention spike = systematic under-weight of early smart-money pressure on animal-meme tapes

Should have caught: First eval scored 40.9 with liquidity, wallet — a lower confirmation floor or earlier attention weight would have held it.

ARCT printed first at score 43 / MC ~120k with three families already live (liquidity, wallet, attention) and social tape warming, yet was held at watch solely because the email-line confirmation threshold was uncleared. Attention never independently spiked in later snapshots; flow/accel only appeared hours afterward. The name still delivered a large move without ever satisfying the high-confidence gate. Process error was treating a static email-line rule as non-negotiable even when multi-family tape and modest score already co-occurred. Pattern: multi-family tape + modest score held solely by uncleared email line

Should have caught: attention spike while score still modest

AVLT first printed 2026-09-16T23:06:54Z at score 65.7 / MC ~24k with four independent families already live (narrative, volume, liquidity, wallet), AI narrative match, early-vs-size/age, and buy-side pressure, yet stayed watch-only because absolute score looked modest and the high-confidence email line was not cleared. Attention never independently spiked in snapshots; flow/accel only printed ~28 minutes later at 23:35:18Z. The name still advanced ~85% off first touch without ever satisfying the email gate, so the desk treated a multi-family early tape as incomplete rather than escalating when score and email thresholds lagged the tape. Pattern: Clustered independent families + narrative match at first touch held under email/score gate until after the move—escalate on family count at touch, not absolute score or email clear.

Should have caught: cross-source confirmation

STARC hit an internal alert at firstScore 74 (rising to claimed 88.4) with six families listed and early-size framing, yet never cleared the high-confidence email line despite later 1858% move. Narrative field was Meme while decisionWhy forced an AI match, and attention never independently spiked in any snapshot even after flow/accel printed ~37 min later. The process therefore separated alert decision from outbound notification on missing independent attention plus narrative mismatch. Pattern: alert scored high but email gate blocked by absent independent attention spike and narrative/strategy mismatch

Should have caught: cross-source confirmation

FOMOPAD printed on Robinhood at score 44 / MC ~25.8k with volume, liquidity, and wallet families already accelerating together, yet the desk held at watch because the email confidence line stayed uncleared and no social/news tag appeared at first touch. Attention never independently spiked in later snapshots, so the modest initial score was treated as non-actionable even while flow/accel continued. The move (reported ~541%) therefore occurred outside the high-confidence gate, exposing a process gap where multi-family tape acceleration was discounted solely for missing cross-source confirmation. Pattern: multi-family accel (vol/liq/wallet) discounted when score modest and social/news tag absent at detection

Should have caught: cross-source confirmation

VPAD first-touched at score 46.7 / MC ~12k with four families (volume, liquidity, wallet, attention) already accelerating and early-vs-size/age plus buy-side pressure on the tape, yet stayed watch-only because the composite sat under the email line and no social/news tag was attached. Attention never printed an independent spike in snapshots, so elevation waited on a tag and a higher absolute score that never arrived before the move. Flow/accel confirmed ~46 minutes later while the desk was still gated on score magnitude and tag presence. The repeatable process error was requiring email-line clearance and a social/news tag before treating concurrent multi-family acceleration as actionable, instead of treating the four-family cluster itself as the primary confirmation while size/age remained early. Pattern: Four-family acceleration + early size/age held at watch solely for sub-threshold absolute score and missing social/news tag

Should have caught: cross-source confirmation

FOMOPAD printed first touch at score 53.5 / MC 69k on Robinhood-chain narrative with three families (volume, liquidity, attention) already co-accelerating and social tape warming, yet the desk held at watch because the absolute score looked modest and no discrete social/news tag had fired, leaving the email line uncleared. Flow/accel confirmation arrived ~27 minutes later while attention never independently spiked in snapshots; the name subsequently extended ~189% without ever clearing the high-confidence threshold. Process error was over-weighting a static score gate and single-tag requirement instead of treating multi-family acceleration plus early-vs-size as sufficient cross-source confirmation to escalate. Pattern: multi-family accel + early size ignored when absolute score modest and no single social/news tag

Should have caught: cross-source confirmation

WOMS hit score 90 with six families confirmed and an explicit alert decision, yet never cleared the high-confidence email threshold despite later 81% movement. Attention failed to print an independent spike in any snapshot while flow/accel only confirmed hours later at 15:31. The process treated narrative-meme match plus multi-family acceleration as sufficient without requiring the attention or cross-source gates that actually gate the email line. This left a high-score early signal unalerted downstream even though the internal decision already read 'alert'. Pattern: multi-family score-90 alert decision without independent attention spike or email-threshold cross-source confirmation

Should have caught: cross-source confirmation

ArcHouse hit score 90 at MC 9753 with an explicit alert decision and six confirmed families, yet never cleared the high-confidence email line. Attention remained non-independent across snapshots while narrative/volume/liquidity/wallet/arc accelerated and flow printed ~25 min later. The 530% move therefore occurred outside the email gate despite the internal high-confidence flag. Process failed at the final notification threshold rather than at scoring or family confirmation. Pattern: alert-decision set but email-line blocked when attention lacks independent spike

Should have caught: cross-source confirmation

FOMOPAD hit the tape at score 51.8 / MC 162k with six families (narrative, volume, liquidity, wallet, attention, alpha) already accelerating and a clean Robinhood-chain narrative match, yet the desk held at watch because the email clearance line stayed uncleared and the raw score looked only modest. Attention never printed an independent spike in the early snapshots, so the multi-family tape was treated as incomplete confirmation. Flow/accel only registered ~27 minutes later; by then the move was already underway and the position was never upgraded. The process error was requiring a single high-confidence gate (email line + attention spike) even after six other families had already co-accelerated on a matching narrative. Pattern: six-family acceleration + narrative match rejected solely for missing email-line / independent-attention spike

Should have caught: cross-source confirmation

CHUSD hit firstScore 79.4 at MC 23173 with an alert decision citing score 88.4, early size/age, and six accelerating families, yet never cleared the high-confidence email line despite a later 94% move. Attention stayed non-independent across snapshots while flow/accel only printed after the initial window, so the multi-family confirmation was treated as sufficient without the missing attention spike. The process therefore issued an internal alert label that did not translate into the actual high-confidence delivery path. This left a confirmed early multi-family setup un-escalated even though cross-source signals continued. Pattern: multi-family accel accepted without independent attention spike or email-threshold gate

Should have caught: cross-source confirmation

FAZE printed at score 52.6 / MC 833k and was rejected solely for mc-high / past size window despite five families (narrative, volume, wallet, attention, arc) already accelerating together and the name only beginning to trend. The desk treated the modest first-touch score and lack of an independent attention spike as decisive, then flow/accel confirmed ~30 minutes later; the token subsequently ran ~491% without ever clearing the high-confidence alert line. Process error was hard size-window veto overriding multi-family confluence and early trend state. Pattern: modest-score + multi-family accel + MC just over window → automatic size reject without confluence override check

Should have caught: cross-source confirmation

BARC hit score 90 at ~20k MC with six families already confirming (narrative/volume/holders/liquidity/wallet/arc) and explicit early-size plus buy-side pressure, yet the alert path still treated missing social/news tag as a soft veto and never escalated past the high-confidence email gate. Flow/accel printed roughly 55 minutes later while attention snapshots stayed flat, so the desk held an already-validated multi-family signal waiting for an independent social spike that never arrived. The process error was requiring an extra orthogonal tag after risk was already marked contained instead of letting the six-family cluster and still-early MC stand alone. Pattern: multi-family confirm + early MC blocked by absent social tag

Should have caught: cross-source confirmation

FEG printed at score 37.2 / sub-6k MC with four families (volume, holders, liquidity, wallet) already accelerating and buy-side pressure visible, yet stayed on watch solely because the email line and social/news tag were uncleared. Flow/accel confirmed within ~30 minutes while attention never independently spiked. The desk treated modest first-touch score plus missing cross-source tags as sufficient to withhold elevation, allowing the later 463% move to occur unalerted. Core process miss was rigid gating on social/email confirmation even when tape, size/age, and multi-family accel were already aligned. Pattern: Four-family accel + smart-money pressure at low MC/age overrides missing social/email tags for escalation

Should have caught: cross-source confirmation

SHOEDOG printed first at score 49.3 / MC 45.8k on Robinhood narrative with three families (volume, liquidity, wallet) already accelerating and size/age still early; desk held at watch because the email/high-confidence line was uncleared and no social/news tag was present. Tape sat at 46.2 with modest first-touch score, so the multi-family accel was treated as insufficient without independent attention spike. Flow/accel confirmed ~35 minutes later and the name moved ~86% while the email gate never opened. Process error was gating solely on the email line and social tag instead of treating concurrent family acceleration plus early-vs-size as a standalone escalation trigger. Pattern: multi-family accel + early size ignored when email/social gates uncleared

Should have caught: cross-source confirmation