Ewan Li

Market scan

Jul 28, 2026

Research as of published 27-day delay

schemav2.0calls3
RegimeCHOP
breadth71.0% green5d-0.99%10d-1.46%Directional tradable
Vol state
  • FULLY_NEGATIVE dealer gamma -- SPY total GEX -$947.1M, QQQ -$1,076.6M -> vol-amplifying, not suppressing. VRP FAIR both sides (SPY +2.77%, QQQ +0.55%) = no premium edge either way. VIX 18.21, down from 18.67. GEX/DEX feed vol-state only, no directional read taken.

8 event-risk item(s) flagged.

  • MEDIUM1
  • ADVISORY1
  • DROP1
MEDIUM: 1 of 3ADVISORY: 1 of 3DROP: 1 of 3
Tier mix — data table
Tier mix: tier and call count
TierCalls
MEDIUM1
ADVISORY1
DROP1

Directional book

2 calls

Sizing is removed from every call by policy — see the disclaimer.

PLNT

MEDIUMShortlaneOI_FADE10d

OI_FADE short nearing its own h10 horizon (4 sessions left), currently losing on a short-excess basis. The mechanical exit rule (earlier of h10 or a negative call_net print) is explicit and has NOT been triggered -- tonight's record call-OI build is a re-acceleration of the very signal the lane trades, not a reason to abandon the rule early. Evidence outside the validated spec (DP block-print distribution, negative-premium call writing) corroborates rather than contradicts the short. Fundamentals CAUTION and an unexplained pop are real reasons for vigilance, not for overriding a pre-registered rule four sessions from its own resolution -- especially since the named soft-invalidation (analyst upgrade/PT raise) still has not fired. HOLD at current half size, add a discretionary risk stop ($59.50, 2-day confirm) to bound the FOMC/Mag7 event-window tail risk that a pure buy-and-hold-to-horizon posture would otherwise carry uncapped.

Structure
short equity, half-cap, unchanged size. Discretionary risk stop ADDED tonight (new, not part of the original pre-registered rule): close above $59.50 with 2-day confirm -- bounds tail risk through the FOMC/Mag7 event window without pre-empting the pre-registered horizon/call_net exit rule. This is a supplementary risk control, not a replacement for the exit rule.
Entry / trigger
Entered 2026-07-20 @ $55.20. Close 2026-07-28 $55.94, -1.51% short-excess (losing). 5d net call-minus-put OI trajectory (most recent last): +395/-4/+180/+14/+1,176 (tonight, call +1,197/put +21). persistence_ratio 0.668 (< 0.85 disqualifier). Build re-accelerated hard tonight after appearing dead on 07-27.
Invalidation
HOLD, not CUT or TRIM. Exit at the EARLIER of: (a) 2026-08-03 close [h10, 4 trading days remaining: 07-29/07-30/07-31/08-03], or (b) the first genuinely negative call_net print (neither triggered tonight). Supplementary discretionary stop: close > $59.50, 2-day confirm (added tonight for FOMC/Mag7 event-window risk, does not override (a)/(b)). Would flip to CUT before the horizon if: a 2nd/3rd analyst PT raise or fresh Buy upgrade lands (the still-unfired soft-invalidation), OR dark-pool buy_ratio flips decisively positive (>0.5) with volume confirming accumulation (reversing tonight's 0.00 read), OR the discretionary stop is hit.
validated excess+0.70% (n=906)regime fit0.55fundamentalsCAUTION
Key risks
  • Losing position (-1.51% short-excess) with only 4 sessions to its pre-registered horizon and a dense Tier-1 macro/mega-cap event stack (FOMC + 5 Mag7 prints) running through the entire remaining window on a vol-amplifying negative-gamma tape
  • Tonight's call-OI build re-accelerated to the largest single-day print of the window (+1,176 net) -- directly contradicts the 07-27 'build is dead, low remaining edge' read that supported holding passively; the pre-registered cut condition (negative call_net) is further from firing tonight than it was two sessions ago
  • Fundamentals gate downgraded CONFIRM -> CAUTION tonight; a genuinely mixed fundamental picture (4/4 EPS beat streak + insider buying vs. guidance-cut narrative + 2.61x D/E + fraud litigation) that could break either way
  • Today's +3.17% is unexplained (no dated name-specific catalyst found) and lands on the same session as the record call build -- plausibly early positioning into the 08-06 print, which falls after this position's own exit, but this is inference, not evidence
  • The two new corroborating signals found tonight (DP block-print distribution, negative net call premium on the build) are NOT part of the validated OI_FADE spec (which trades raw oi_net_5d contract counts, not premium-signed or DP one-sidedness) -- they support the HOLD decision as evidence quality, not as an additional validated gate; do not hand-patch the lane spec off this reasoning (invariant #5). Log as a /calibration-audit proposal (premium-sign + DP corroboration overlay for OI_FADE) instead.
Gates
regime crashN/A

N/A -- OI_FADE is not S1/relative-weakness-gated, s1_standdown irrelevant to this lane. General out-of-strong-regime half-cap ceiling already respected (position at half, not above).

liquidityPASS

PASS -- scripts/liquidity.py: close $55.94, 20d $-ADV $94.4M, clears the $50M floor with room; confirms no repeat of the 07-24 $250M hand-multiplication error.

clusterPASS

PASS -- only other tonight's-book name is HOG, no shared theme (fitness services vs. motorcycles/recreational). See correlation-cluster note in watchlist_write_back re: HOG's absence from the truth-set panel for a quantitative recompute.

fundamentalsNOTE

CAUTION (-1 tier, NOT veto) -- downgraded from CONFIRM (07-27) to CAUTION tonight. Zero post-07-20 PT raises or fresh Buy upgrades (the named soft-invalidation trigger has NOT fired); Deutsche Bank CUT $61->$55 (07-22); Wells Fargo flat $65 (07-20); William Blair flat Hold (07-27) -- marginal analyst tape tilts bearish, consistent with the short, not against it. Contradicting: clean 4/4 EPS beat streak, net insider buying, rev +14.4%/EPS +31.4% YoY. Supporting: guidance-cut/membership-execution deterioration, securities-fraud class actions, D/E 2.61 securitized whole-business debt.

event riskNOTE

PLNT's own ER (2026-08-06 pre-market) falls AFTER the position's own h10 exit (2026-08-03) -- the print is correctly outside the remaining window, unchanged from 07-27's reasoning. FOMC (07-29) + MSFT/META/ARM/QCOM (07-29 AH) + AAPL/AMZN (07-30 AH) fall INSIDE the remaining 4-session hold window as broad market-wide vol risk (not company-specific catalysts) -- noted, does not independently veto a held short with an already-defined exit rule.

HOG

ADVISORYLonglaneS4_pcr_fade5d

S4 sentiment-contrarian long candidate: a 4-session post-earnings capitulation (07-21 high $28.45 -> $24.79, -12.9%) followed by today's reversal bar (open 24.53, low 24.45, close 25.29, +2.02%/+1.78% excess, closing near the high) on an elevated (84.1st pctl) but not extreme OI-PCR, clearing the new call-volume denominator floor that killed PFGC's divide-by-near-zero artifact. Advisory-only lane (+0.44% mean prior, never sized beyond watch) combined with a contradicted premium-signed mechanism, a missing fundamentals verdict, and a dense Tier-1 event stack inside the trade horizon -- orchestrator's WATCH read is CONFIRMED, not overturned. Log the premium-sign contradiction as a /calibration-audit evidence-quality proposal (a denominator-style hygiene check analogous to the PFGC fix), not as a new hard veto -- invariant #5 (pre-registered ≠ scored).

Structure
WATCH only -- no position taken.
Entry / trigger
Would-be long entry near today's close $25.29 off the reversal-bar setup, IF sized. Not triggered tonight.
Invalidation
N/A -- no position. Would reconsider for starter sizing (never above starter -- S4 is capped, see CLAUDE.md lane table) post-FOMC (earliest 07-30) if: (a) fundamentals-gate returns CONFIRM or CAUTION (not VETO) once spawned by the orchestrator, (b) premium-signed options flow flips to corroborate the contract-count PCR read (net_put_premium turns positive, i.e. puts actually being bought), and (c) the FOMC/Mag7 event stack clears without a regime break.
validated excess+0.44%regime fit0.30
Key risks
  • MECHANISM CONTRADICTED: net_put_premium -$11,791 (puts net SOLD) vs net_call_premium +$4,669 (calls net bought); bull_bear_skew +0.249 bullish -- on a premium-weighted (dollars-at-risk) basis, the put-heavy crowd S4's contract-count spec (vol_PCR 5.37, put_vol 1,649/call_vol 307) claims to fade does not exist
  • Book is ~$66K total premium across 1,956 contracts (~$0.34/contract) -- too thin to confidently call a crowd in either direction
  • OI-PCR 84.1st percentile (1.183) is elevated but NOT extreme (p90=1.446, p95=2.203); the same test would have similarly ranked PFGC's already-rejected 1.26, so this metric alone does not separate a real signal from noise -- the denominator floor (call_vol >=75, HOG clears at 307) is what separates them here, not the OI-PCR level
  • Consumer-discretionary, rate-sensitive name entering an h5-10 window that runs through FOMC and five Mag7 prints on a vol-amplifying negative-gamma tape
  • No fundamentals-gate verdict on record -- unresolved risk-filter gap
Gates
regime crashN/A

N/A -- S4 is not S1/relative-weakness-gated. General out-of-strong-regime half-cap ceiling would apply if sized, moot at watch.

liquidityPASS

PASS -- scripts/liquidity.py: close $25.29, 20d $-ADV $68.9M, clears the $50M floor.

clusterPASS

PASS -- no shared theme vs. PLNT (the only other name in tonight's book); see truth-set coverage gap note in watchlist_write_back.

fundamentalsNOTE

MISSING -- no fundamentals-gate verdict on record for HOG; analyses/scan/2026-07-28/ was empty at the start of Phase D. Flagging back to the orchestrator per gate rule 4: hold at watch until a verdict arrives, before any future starter-scale sizing. This is not independently determinative tonight since the advisory-lane cap already limits HOG to watch, but must be resolved before HOG can ever be sized.

event riskNOTE

-1 tier -- Tier-1 macro/mega-cap event stack (FOMC 07-29 + MSFT/META/ARM/QCOM 07-29 AH + AAPL/AMZN 07-30 AH) sits inside the h5-10 trade horizon on a vol-amplifying negative-gamma tape. HOG's own earnings (11-03) clear the h-window veto cleanly (see tail_cap).

Vol book

Non-directional, delta-neutral, net-of-cost — advisory only.

No vol-book entries in this scan.

Watch / stood-down

1
TickerTierLaneDirectionInvalidation
ORCLDROPMOM_SHORTShortN/A -- resolved via the pre-registered h10 horizon-expiry rule, not a stop-out. Note: ORCL made a new 52w intraday low ($114.50) today and reversed to close $119.96. On a SHORT, the $118.18 open fill therefore captured MORE than a close fill would have: +8.82% short-excess at the open vs +7.69% at the close (+10.16% vs +8.80% gross), i.e. closing at the open beat closing at the close by +1.13pp of excess. [CORRECTED by orchestrator: the sizer's original text asserted the opposite sign; watchlist_write_back had it right.]

Lane status

No lane produced a conceptual fire without a sized call tonight.

Watchlist write-back

6

Carried forward to the next session's watchlist — not calls.

ORCL (MOM_SHORT) -- CLOSED per the 07-27 decision, executed at the 07-28 open $118.18. Entry 07-13 $131.54; first half trimmed 07-24 @ $114.99 (+11.21% short-excess); second half closed 07-28 @ $118.18 (+8.82% short-excess). Blended realized short-excess across both halves ~+10.02%. Position is flat. ORCL printed a new 52w intraday low ($114.50) today and reversed to close $119.96 -- closing-at-the-open beat closing-at-the-close by roughly 1.5pp on the remaining exposure, a small but real cost of the horizon-driven proactive close vs. a hypothetical same-day exit; recorded for /calibration-audit as a resolved MOM_SHORT win, and as a data point for the 'close at horizon, don't try to time the print' discipline.PLNT (OI_FADE) -- HOLD at current half size. Pre-registered exit rule (earlier of h10=2026-08-03 or first genuinely negative call_net print) is NOT triggered: tonight's call/put print is +1,197/+21 = net +1,176, the largest single-day call build of the 5-session window (5d series: +395/-4/+180/+14/+1,176) -- persistence_ratio 0.668, still under the 0.85 single-block disqualifier but a clear re-acceleration, not a continuation of the 07-27 'build is dead' read. Fundamentals gate now reads CAUTION (-1 tier; downgraded from CONFIRM on 07-27) -- zero post-07-20 PT raises or fresh Buy upgrades (the named soft-invalidation), one further analyst CUT (Deutsche Bank $61->$55, 07-22), tape tilts bearish. New corroborating evidence found tonight, not part of the validated OI_FADE spec (evidence-quality only, logged for /calibration-audit, not a hand-patched gate): dark-pool block prints are overwhelmingly sell-side into today's +3.17% up day (07-27 buy_ratio 0.16/$29.4M, 07-28 buy_ratio 0.00/$10.4M on ~1.35x normal volume) and tonight's call-OI build carries NEGATIVE net call premium (-$45,693, i.e. calls being written/sold, not bought) -- both point away from a bullish accumulating crowd and toward distribution/premium-selling, which corroborates rather than undermines the short. Net effect: HOLD, not TRIM or CUT -- see decision.calls[PLNT].thesis for the full reasoning and the discretionary risk stop added for the FOMC/Mag7 event window.HOG (S4_pcr_fade) -- WATCH, not sized. Orchestrator arbitration confirmed, not overturned. Advisory-only lane (+0.44% mean prior) already caps this at watch regardless of score; independently, no fundamentals-gate verdict exists for HOG in tonight's run (analyses/scan/2026-07-28/ was empty at Phase D start) -- flagging back to the orchestrator per gate rule 4 to spawn fundamentals-gate on HOG before any future starter-scale consideration. Mechanism is contradicted on a premium-signed basis (net_put_premium -$11,791 = puts net SOLD, not bought; net_call_premium +$4,669; bull_bear_skew +0.249 bullish) against a contract-count PCR of 5.37 that read the crowd as put-heavy -- the crowd S4's contract-count spec would fade does not exist on a dollars-at-risk basis, and the whole book is ~$66K premium across 1,956 contracts (~$0.34/contract), too thin to call a crowd either way. Tier-1 macro/mega-cap event stack (FOMC + MSFT/META/ARM/QCOM 07-29, AAPL/AMZN 07-30) sits inside the h5-10 trade horizon on a vol-amplifying negative-gamma tape -- -1 tier. HOG's own earnings (11-03) clear the h-window veto cleanly. Would reconsider post-FOMC (earliest 07-30) if: (a) fundamentals-gate returns CONFIRM/CAUTION (not VETO), (b) premium-signed flow flips to corroborate the contract-count PCR read, and (c) the event stack clears without a regime break.CORRELATION CLUSTER -- HOG vs PLNT: no shared theme (motorcycles/recreational discretionary vs. fitness services), judged PASS by sector/theme inspection. Could not compute a quantitative 60-session pairwise correlation: HOG is absent from data/prices.parquet (truth-set panel), only PLNT is covered. Flagging as a truth-set coverage gap for the orchestrator to backfill -- not decision-blocking tonight since HOG is WATCH (not sized) and PLNT is an unchanged hold, so no multi-name basket is being sized together.TAIL CAP LEDGER -- MOM_SHORT new-starter cap (advisory/watch-only, PROVISIONAL) cost nothing tonight: the lane produced zero candidates worth sizing (SKHY, FRVO survived gates but the lane itself declined to size them) and today's cohort round-trip (squeeze cohort -0.40%/-1.15% mean, reversing 07-27's inversion) argues the normal factor sign has reasserted, which is exactly the environment the cap is meant to protect against being caught offside in. S4 tail cap (cut on same-day catalyst; veto if earnings inside h-window) -- no same-day catalyst identified for HOG's 07-28 reversal bar (a technical follow-through on the 07-23 print, not a fresh news event), earnings well outside window: both sub-conditions PASS, HOG stays at watch purely on the advisory-lane cap + missing fundamentals + event-risk stacking, not the tail cap itself. OI_FADE carries no lane-specific tail cap (S1-short/S2-long/vol-short only per gate 6) -- PLNT's build re-acceleration argues, if anything, for the pre-registered exit rule to keep governing rather than for a discretionary early cut.LANE PRIOR NOTE -- tonight's OI_FADE figure (+0.70% mean/n=906) and MOM_SHORT figure (-0.03% mean) are the 2026-07-24 re-baselined harness numbers per CLAUDE.md invariant #6; do not re-cite pre-07-24 figures for these lanes.
Lane glossary (3)
MOM_SHORT
Momentum, short leg — near-52w-low relative weakness. Regime-gated against momentum-crash/squeeze rebounds (h10).
OI_FADE
OI-flow fade — fades persistent multi-day net call-OI building; heavy call-OI build precedes underperformance. The most robust lane measured (short, h10).
S4_pcr_fade
Sentiment contrarian — fades a put-heavy (high put/call ratio) crowd long; the only sentiment leg that measured positive. Advisory size only (h5-10).

Personal research journal · published on a ≥ 14-day delay · position sizes removed · not investment advice.