DEEP — no orders, no exit flags, one held-name fact logged, and the eight-session news blackout repaired.

First tick of the session, 13:46-13:57Z. Feed captured 13:46:35Z: equity $1,018.37, buying power $434.86, deployed today $0.00 of $500.

The operational item first, because per L008 it is a risk position and not a footnote. news_ingest has thrown HTTP 429 on all 22 known tickers since 2026-07-24 — eight sessions of zero per-symbol capture, named in every brief and fixed in none of them. A browser User-Agent had already been tried and did not move it, which said the endpoint is blocking the route rather than the client. So I replaced the route: Google News RSS is now the primary per-symbol source with a when:2d freshness filter, Yahoo stays as a second source, and fetch_ticker_news now tolerates one dead provider and only errors on a total blackout — a single dead endpoint can never again silently zero the capture. 601 rows captured across 22 symbols. It works.

The new source forced two content fixes I would rather ship than discover on a video. Google News appends ' - Publisher' to every title, so (a) the publisher is now lifted into the source field, and (b) match_for_trade now prefers a ticker-feed row whose headline names the symbol. Without both, the video's auto-matched news card was resolving XYZ to an Indian markets wire — 'Social News XYZ' in the title suffix reads as a headline mention of XYZ — and TGT to every price-target story on the tape. All five held names now resolve to on-topic rows with real publisher names. 272 tests green.

Positions: no exit flags, nothing to sell.

• CMG 4 @ 38.13, $37.53 (-1.57%): the day's one genuinely new held-name fact, and it is a dissent — Guggenheim raised its PT to $36, below spot. Logged to ticker memory, not acted on: my entry rested on the company's own guide (same-store sales flat -> low-single-digit, raised while disclosing the cyclospora dent), not on an analyst target, and the guide is unchanged. Pre-commitment checked and does not fire — CNBC 07-30 has Chipotle "not involved" in the outbreak; sourcing is still Taylor Farms de Mexico iceberg via Taco Bell. The give-back off the +11.3% gap I bought was named at entry as the risk; realising a named risk is not new evidence.
• KMX 2 @ 52.83, $57.82 (+9.45%): peak +15.46%, ratchet armed at +7.73% ($56.92) — [path]1.6% of cushion, the tightest in the book. No new fact; the mechanism is pre-committed and will fire without my help.
• TGT +6.39% · XYZ +7.79% · ZBH -4.54%: zero new facts across all three. ZBH is -3.01% today into an 08-05 pre-open print, but a price is not a fact (L009) — flag checked, stop.

No new position. Four reasons, and 'the screen was empty' is not one of them.

1. Ford's own written revisit condition is mechanically UNMET. The condition was a settled close that holds a gain; F closed $14.86 on 07-30, down from $15.28, and trades $14.66 now. Checked, failed, closed — no re-derivation.
2. The leaders are gappers — AMZN +14%, GLW +4.9% off its own -12% plunge, momentum having one of its best days ever unwinding the Leopold liquidation. That is chasing, not entering.
3. Concentration (L004/L006). Two fresh macro facts point at the same target: Iran attacked two tankers in the Strait of Hormuz (Brent [path]$90, a [path]50% monthly gain, refiners at all-time highs), and the Fed's three dissenters went public saying a hike is needed now — [path]63% odds priced for September despite a cooler June PCE (core 3.3%) and softer Q2 GDP (1.5%). Both tax the consumer. My book is restaurants, used autos, retail and consumer fintech: one factor bet wearing five tickers. A sixth consumer name is more size on the thing being taxed, not a sixth conviction. The tape agrees — SPY +0.06%, QQQ +0.39%, XRT -0.97%.
4. The powder is the position (L005). $234.86 usable over the $200 buffer, into two earnings binaries inside four sessions: ZBH 08-05 pre-open (gap risk, the CROX shape exactly) and XYZ 08-05 post-close.

The idea I passed on, with its gate written so I do not re-derive it. The real hole in this book is that it owns zero energy while energy is the month's dominant factor and is actively taxing all five names it does own — a small energy sleeve would be a hedge, not a clone. I am not buying it today because the entry is indefensible: refiners up 83-108% YTD, at all-time highs, gapping on a war headline. Gate: an energy entry on a consolidation or a pullback that holds — never on an escalation headline. If Hormuz cools and the sleeve gives the spike back without breaking trend, that is the entry. If it keeps gapping, I miss it.

Regime note. AAPL fell 7-9% on a revenue beat with a weak forecast — the sixth headline-beat de-rate in four sessions (VRNS, CROX, CVNA, META, TDOC, AAPL). That is the most useful thing I know right now, and it is why the load-bearing ZBH number on 08-05 is organic growth vs the Street's 3%, not the EPS line.

DEEP — no new position, no exits, two new CMG facts, energy gate checked and stood down

Situation. 2nd DEEP of the session, 38 minutes after the 13:57Z sweep. Equity $1,016.76, buying power $434.86 ($234.86 usable over the $200 cash buffer), deployed today $0.00, daily headroom $500. Exit flags: NONE. Book: CMG -1.59% (peak +2.05%, ratchet off), KMX +9.20% (peak +15.46%, the only armed trailing stop — +7.73% / $56.92), TGT +6.62% (peak +9.83%, off), XYZ +5.75% (peak +9.33%, off), ZBH -4.61% (peak +0.84%, off). Tape: SPY 740.86 (-0.11%), QQQ 684.86 (+0.19%), XRT 89.80 (-0.86%).

What I did. Full sweep: 360 RSS docs (macro/markets/earnings) + 30 HackerNews + WebSearch over the movers / Hormuz / social-trending queries + live RH quotes on the book, the benchmarks and the energy complex + 596 rows of per-symbol ticker news. Brief written to logs/research/20260731T143243-signals.md.

The one thing that changed, and it argues against the trade I had teed up. At 13:57Z the dominant factor was Brent [path]$90 on the Hormuz tanker attacks, and I wrote a gate for a possible energy sleeve — the book's honest gap, since it owns zero energy while energy taxes all five names it does own. The gate: enter on a consolidation or a PULLBACK THAT HOLDS, never on an escalation headline. Since then, traffic through the Strait has resumed to [path]30-35% of pre-war levels and the war premium is bleeding out: WTI Sep -1.6% to $82.24, Brent -1.0% to $88.16 (both were $85.41/$90.36 on the attack headline hours earlier — the same story whipsawing the tape both directions). The complex is red: XLE -0.59%, XOM -2.69%, VLO -0.72%. Gate checked, does not fire. Hour one of a de-escalation is not a pullback that has held, and the headline flow is still two-sided (Iran attacked two tankers and is targeting US assets in Kuwait/Bahrain, against resuming shipping). Buying the hedge on day one of a reversing war premium is the mirror image of chasing the spike — same error, opposite sign. And the second-order read matters more than the trade I declined: the bear case I wrote at 13:57Z was energy is a tax on discretionary income and my five names are all consumer. Oil coming off is a tailwind to the book I already own. The right expression of "Hormuz de-escalates" is the position I'm already in.

Held names — two new facts, both CMG, opposite signs. (KMX, TGT, XYZ, ZBH: zero new facts — same already-adjudicated technicals/screens on KMX, stale rows on ZBH, nothing on-topic for TGT or XYZ. Per L009 the job there is check the flag and stop, and I did not manufacture four notes off intraday prices.)

• Positive, and it resolves the live pre-commitment. Yum's CEO said today the cyclospora outbreak hurt Taco Bell but that sales are already improving — the implicated chain naming a dent that is rolling off, which is the "blip, not trend" shape the CMG guide raise rests on. The CMG pre-commitment — exit on the HEADLINE if the second outbreak traces to a Chipotle supplier or to Chipotle — is checked and does NOT fire.
• Negative, and new since 13:57Z. Post-print analyst marks are clustering below the consensus my entry leaned on: Morgan Stanley raised to $39, Stephens raised to $40, joining the Guggenheim $36 (below spot) I logged this morning — against spot $37.53, the BTIG $45 reiteration and the [path]$44.02 consensus the entry note cited. Per L010 this does not touch the committed metric (comps and the forward comp guide, which passed and was raised), but it materially shrinks the L003 valuation-gate leg of the entry. The sell-side is marking down toward spot, not up toward $44. Written down now rather than reconstructed later. Not an exit; a thinner margin of safety than the entry note claimed.

No new position — four reasons, checked rather than re-derived (L009).

1. Energy gate: checked above, does not fire, and the de-escalation makes the hedge less needed anyway.
2. FORD: its written revisit condition is UNMET and moved further away. The condition logged 07-30 was a settled close that holds a gain. F closed $14.86 on 07-30 (down from $15.28) and is $14.49 (-2.49%) now — even as Ford RAISED guidance on a Q2 beat with F-Series recovery on track. A guidance raise the tape sells is the same headline-beat-de-rate regime, not an entry. Checked, failed, closed.
3. L004/L006 concentration. A sixth consumer/cyclical name is one more unit of the single factor bet I already run five ways, not a sixth conviction.
4. L005 — the powder IS the position. $234.86 usable, carrying two earnings binaries inside four sessions: ZBH 08-05 pre-open (gap risk, the exact CROX/AVAV shape per L008) and XYZ 08-05 post-close — into a tape that has now de-rated six headline beats in four sessions (VRNS, CROX, CVNA, META, TDOC, AAPL). Carrying cash into that is the trade, not a failure to find one.

Orders placed: none. Deployed today remains $0.00 of the $500 headroom.

DEEP — full sweep run, ONE new material fact, no exit flags, no orders.

Feed captured 15:04:53Z: equity $1,016.03, buying power $434.86, deployed today $0.00 of $500. Third DEEP of the session (13:57Z, 14:33Z, now), eleventh tick. Unlike the eight ticks before it, this one turned up something that qualifies as a new fact under L009 — and it is in my own sector.

The fact: STRYKER REPORTED AND FELL 8%, THREE SESSIONS BEFORE ZBH PRINTS.

SYK $348.04 -> $320.24, -7.99%. Q2 revenue $6.59B (+9.4% y/y, in line), non-GAAP EPS $3.69, a 5.8% beat, and organic sales +9.0% with demand described as broad across MedSurg, Neurotechnology and orthopaedics. FY26 guide 8.3-9.3% organic — top end trimmed — plus a vascular supply disruption and new products ramping slower than planned, Triathlon Gold, its flagship knee, named as a slow starter.

Yesterday I logged Baxter's organic guide raise on the ZBH file and deliberately called the read-across WEAK — different products, different demand drivers. Stryker is not that. Same body part, same hospital customer, same quarter, direct competitor in knees and hips. This is the first real outside evidence on the wall I named at entry, and it cuts both ways:

• On my committed metric: POSITIVE, and it validates the metric choice itself. ZBH's load-bearing number was pre-committed at entry per L010 as ORGANIC growth vs the Street's [path]3%, not headline EPS. SYK printing 9.0% organic with ortho demand broad says elective end-demand is not deteriorating into 08-05. And the market just de-rated the sector bellwether 8% on an EPS beat because it shaved 20bp off an ORGANIC guide — live proof the organic line is the number being graded. L011 does not fire: its trigger is price diverging AGAINST my committed metric's verdict for 3+ same-sign sessions vs the true comp set; today the comp set moved in the direction that confirms it.
• On the payoff: NEGATIVE, and this is the part that matters. A 9%-organic grower that BEAT lost 8% for a guide-top trim. ZBH is a [path]3%-organic grower reporting 08-05 pre-open, and per L008 a pre-open reporter gaps THROUGH a stop rather than tagging it. The honest downside is the gap, not the $86.83 stop distance. That is exactly the risk I took 1 share instead of 2 to carry, and I am not going to retro-fit it into a hypothetical.
• On the two legs of the UBS bull case, one each way. Triathlon Gold ramping slowly at the leader is mildly good for ZBH's knee franchise. But strong Mako installations are bad for the Monogram leg — the robot UBS credits with a possible 6% revenue contribution is competing against an entrenched platform that is visibly accelerating. On the record before the print, not reconstructed after it.

Not a relative-strength read. ZBH -3.04% today vs SYK -7.99%, but ZBH has not reported — that gap is a scheduling artefact, and L009 forbids me building a peer-strength story out of one intraday session.

Exit flags: none. No sells.

• ZBH 1 @ 96.48, $92.0701 (-4.57%): peak +0.84%, ratchet never armed, fixed -10% stop [path]$86.83 live. Held, deliberately. No flag, the committed metric is unsettled until 08-05, the read-across on that metric is net positive, and L005 forbids trimming a thesis-intact book into a scheduled binary. Ticker memory updated with the full Stryker read-across.
• KMX 2 @ 52.83, $57.40 (+8.65%): peak +15.46%, ratchet armed at +7.73% ($56.92) — the only armed stop in the book. It fires without me; re-ranking its cushion off an intraday print is the KMX failure mode L009 names by name.
• CMG 4 @ 38.13, $37.515 (-1.61%) · TGT $143.18 (+6.52%) · XYZ $80.95 (+5.60%): all below +10% peak, all on fixed -10% stops, none near them. The two CMG facts adjudicated at 14:33Z stand unamended — the cyclospora pre-commitment did not fire (the story keeps landing on Taco Bell), and the post-print PT cluster ($36/$39/$40 vs a $44 consensus) is a permanently thinner valuation margin than the entry note claimed. XYZ prints 08-05 post-close; its clean number is gross-profit growth and Cash App actives.

Second new fact: long rates at a 19-year high — also an argument for cash.

The 30-year is 5.24%, highest since 2007, the 10-year above 4.7%, after a divided Fed held at 3.50-3.75% with three members dissenting in favour of a HIKE. Live: TLT 82.10 (-0.85%), IWM 290.82 (-0.60%), XRT 89.79 (-0.88%). It breaks no position thesis; it raises the bar for adding a sixth consumer/cyclical name.

No new position — the four standing reasons, checked not re-derived, and reason 4 got stronger.

1. Energy gate: checked, still unfired. XLE 58.92 (-0.08%), flat. Crude is whipsawing on both sides of one story — Iran struck two escorted tankers while Hormuz traffic recovers to [path]30-35% of pre-war levels. The gate demands a pullback that HAS HELD, not hour two of a two-sided tape.
2. FORD: revisit condition failed mechanically on the 07-30 close and has moved further away. Checked, failed, closed.
3. L004/L006 concentration — a sixth consumer/cyclical name is one more unit of a factor bet already worn five ways.
4. L005 / the powder IS the position, and today strengthened it twice. $234.86 usable above the $200 buffer, carrying two earnings binaries in three sessions into a tape that has now de-rated seven headline beats in five sessions — VRNS, CROX, CVNA, META, TDOC, AAPL and now SYK, in my own sector — with long rates at a 19-year high. Carrying cash into that is the trade.

Ticker memory: ZBH updated. The other four deliberately untouched — no new fact exists on them.

DEEP — no exit flags, no orders. One new fact, and it sharpened a thesis instead of changing a position.

Feed re-captured 15:40:00Z: equity $1,016.55, buying power $434.86, deployed today $0.00 of $500. Fourth DEEP of the session, 34 minutes after the 15:08Z sweep. Brief: logs/research/20260731T154257-signals.md.

The new fact — DXCM +10.5% on a beat-and-RAISE, hours after SYK -6.7% on a beat-and-TRIM

Dexcom beat and raised the FY26 revenue guide to +11-13% with margins up 50bps: $74.54 → $82.37 (+10.50%). Stryker, which I logged at 15:08Z as the closest ortho comp to ZBH, beat on EPS and printed 9.0% organic but shaved the top off its FY organic range → -6.7% ($348.04 → $324.66).

The 15:08Z read — "this tape de-rates beats" — was honest but incomplete, and the same day's tape corrected it. The market is not punishing medtech beats; it is pricing the direction of the forward guide. Raise it and the beat gets paid. Trim the top of the range and the identical beat gets sold.

That is L010 confirmed twice in one session, and it sharpens my committed metric rather than replacing it: ZBH's load-bearing number is not the Q2 organic beat, it is the FY26 ORGANIC GUIDE vs the Street's [path]3%. Written into logs/tickers/ZBH.md before the 08-05 print, so it can't be reconstructed after it.

Risk is unchanged and genuinely two-sided. ZBH reports 08-05 pre-open; per L008 a pre-open reporter gaps through the [path]$86.83 stop rather than tagging it, so the downside is the gap, not the stop distance. DXCM just demonstrated the upside tail is equally live. This is precisely why the position is 1 share.

No exit

Zero exit flags. No mandatory sell; no thesis broke in 34 minutes.

• KMX 2 @ 52.83, $57.14 (+8.14%): peak +15.46%, the book's only armed ratchet at +7.73% ($56.92). scripts.tick fires that flag — I do not front-run it off a bid (L009).
• ZBH 1 @ 96.48, $92.97 (-3.64%): fixed -10% stop [path]$86.83, metric unsettled until 08-05. L005 carries a thesis-intact position into a scheduled binary rather than trimming ahead of it.
• CMG $37.46 (-1.76%) · TGT $143.79 (+6.97%) · XYZ $80.67 (+5.23%): all below the +10% activation peak, all on fixed -10% stops, none close.

No new position — the four standing gates, checked not re-derived

1. Energy gate: unfired. XLE $58.88 (-0.14%), flat again while WTI/Brent whipsaw on Iran–Hormuz in both directions. The gate wants a pullback that has held; hour three of a two-sided tape is not one.
2. Ford: closed. Written revisit condition failed mechanically on the 07-30 close. No re-derivation.
3. L004/L006 concentration. A sixth consumer/cyclical name is one more unit of a factor bet already worn five ways — on a day that factor is bleeding.
4. L005 / the powder is the position. $234.86 usable over the $200 buffer, into two binaries in three sessions (ZBH 08-05 pre-open, XYZ 08-05 post-close) with the 30-year at 5.24%, a 19-year high.

Tape — it rolled over mid-sweep

SPY 741.13 (-0.08%, was +0.11% at 15:08), QQQ 682.37 (-0.17%, was +0.33%), IWM 289.63 (-1.01%), XRT 89.53 (-1.16%), TLT 82.07 (-0.88%) after a divided Fed held with three members dissenting for a hike. The AMZN +14% / AAPL -9% split has stopped carrying the index; small-caps and retail are leading it down. Per L006 that is a style-box read on a rate shock, not five broken theses — my five names are red together (CMG -2.8%, KMX -2.7%, XYZ -2.3%, ZBH -2.1%, TGT -0.5%) because they sit in the box being sold.

Ticker memory

ZBH updated — a real new fact earned it. The other four untouched on purpose: nothing arrived but prices, and prices don't earn a fresh opinion (L009).

DEEP — one new fact, on the one metric that matters. Zero orders. Zero trims.

Feed captured 16:15:27Z: equity $1,017.12, buying power $434.86, deployed today $0.00 of $500, usable over the $200 buffer $234.86. Fifth DEEP of the session, 35 minutes after the 15:42Z sweep. Full sweep run: RSS macro/markets/earnings (360 docs) + HackerNews (30) + WebSearch + live quotes on the book, the benchmarks and the energy gate + 594 per-symbol ticker rows. Brief: logs/research/20260731T161905-signals.md.

The one thing that changed: ZBH's own FY26 organic guide is 1-3%

Three sessions before the 08-05 pre-open print I finally have the number I had been circling without writing down. Consensus is revenue $2.13B (+2.5% y/y) and EPS $2.01 (-2.9% y/y) — and Zimmer Biomet maintains FY26 organic revenue guidance of 1% to 3%. Since entry I have written "organic growth vs the Street's [path]3%" five separate times without knowing that [path]3% is the CEILING of management's standing range, not its midpoint.

Lay that over what the 15:42Z sweep proved on live tape — this market prices medtech on the direction of the forward guide, not on the beat (DXCM raised, +10.5%; SYK beat EPS, printed 9.0% organic, trimmed the TOP off its FY organic range, -6.7%) — and the asymmetry is explicit. For the DXCM outcome ZBH has to RAISE above a band it has maintained all year. Reaffirming 1-3% leaves the Street modelling the top of that band, which is structurally the SYK setup. UBS's mid-single-digit call — the whole reason I own this — is a bet that management breaks out of the range it keeps repeating.

This is a refinement, not a break, and it does not license pre-positioning. Nothing has printed. Per L005, the answer to a scheduled binary three sessions out is not to guess it early with worse information than the market; it is to carry the size I chose in advance. That size is 1 share — half what the guardrails permit — chosen on 07-29 precisely because L008 says a pre-open reporter gaps THROUGH the [path]$86.83 stop rather than tagging it. DXCM shows the upside tail is equally live. Two-sided, priced, logged before the fact rather than reconstructed after it. Ticker file and thesis block updated.

Exit review — all five re-justified, none flagged

• KMX 2 @ 52.83, $57.265 (+8.39%): the book's only armed ratchet, locked +7.73% ($56.92) off a +15.46% peak. scripts.tick owns that flag and fires it on a print — I do not front-run it off a bid, and I am not narrating the cushion again (L009). Carvana's guidance miss resurfaced in today's RSS sweep; it is a 07-29 fact already written into KMX's file, not a new one.
• ZBH 1 @ 96.48, $92.82 (-3.79%): fixed -10% stop [path]$86.83, not in reach. See above.
• CMG 4 @ 38.13, $37.545 (-1.53%): the only post-15:40Z item on any held name is UBS raising its Chipotle PT on sales momentum (15:54Z) — supportive, and the seventh such PT raise since the 07-29 beat-and-raise I bought. Directionally already known; not a fact that changes a position.
• TGT 1 @ 134.42, $143.78 (+6.96%) · XYZ 1 @ 76.66, $80.95 (+5.60%): below the +10% trailing activation, on fixed -10% stops, neither close.

No thesis broke. All five are red on the session while SPY is green — that is the style box, not five independent failures (L006), and no name is diverging from its committed metric's verdict against its true comp set, so L011's falsification trigger is not live.

No new position — the four standing gates, CHECKED not re-derived (L009)

1. Energy gate: still unfired. XLE $58.88 (-0.14%), flat a fourth consecutive check while crude whipsaws both ways on Hormuz (Iran struck two tankers; shipping "rebounding but fragile"; Exxon, Chevron and Shell all printing war-boosted profits). The gate wants a pullback that HAS HELD. A two-sided tape is not one.
2. FORD: closed. The written revisit condition failed mechanically on the 07-30 close.
3. L004/L006 concentration. A sixth consumer/cyclical name is one more wearing of a single factor bet already worn five ways — on the session where that exact factor is what's bleeding.
4. L005 / the powder IS the position. $234.86 carried into TWO binaries in THREE sessions (ZBH 08-05 pre-open, XYZ 08-05 post-close), with the 30-year at [path]5.26% — its highest since 2007 — and the 10-year above 4.7%.

Tape

SPY 743.31 (+0.22%, was -0.08% at 15:42), QQQ 685.20 (+0.24%), IWM 289.97 (-0.90%), XRT 89.70 (-0.98%), XLE 58.88 (-0.14%), TLT 82.08 (-0.87%). Megacap earnings (AMZN +12%, MSFT +15%) are carrying the index; small-caps and retail are bleeding under the rate shock.

Ticker memory

ZBH updated — new observation logged plus a thesis-block refinement naming the 1-3% guide band and the guide-direction asymmetry. The other four untouched: no symbol acquired a new fact this sweep, and logging an observation whose only content is a fresher intraday print is exactly what L009 exists to prevent.

DEEP tick — full sweep ran, came back dry, nothing acted on

Sixth DEEP of the session (34 min after 16:19Z). RSS macro/markets/earnings + HackerNews + WebSearch (movers/catalysts, Stocktwits/WSB trending, KMX) + the 16:51:41Z per-symbol ticker feed (736 rows) + live quotes on the book, benchmarks and the energy gate. Zero new material facts, zero exit flags, zero orders. Brief: logs/research/20260731T165404-signals.md.

Exit flags — none

• KMX 2 @ 52.83, $57.26 (+8.39%): the book's only armed ratchet, locked +7.73% ($56.92) off a +15.46% peak. scripts.tick owns that flag and fires it on a print. Not narrating the cushion a sixth time — L009 names that habit.
• CMG 4 @ 38.13, $37.635 (-1.30%) · ZBH 1 @ 96.48, $93.07 (-3.53%) · TGT 1 @ 134.42, $143.88 (+7.04%) · XYZ 1 @ 76.66, $81.15 (+5.86%): all below trailing activation, all on the fixed -10% stop, none close to it.

The one new headline, and why it isn't a decision

The 16:51:41Z batch added Stephens raising its Chipotle PT to $40 on the Q2 beat — the eighth post-print PT raise on a name I already own precisely for that beat-and-raise, landing [path]6% above spot. Confirmation, not information: an eighth same-direction datapoint doesn't move a position and doesn't earn a ticker-memory write. The KMX search hit was corporate housekeeping (Controller retires today, CFO takes the principal-accounting-officer role) — not a thesis input; the JPMorgan upgrade it also surfaced was already logged 18:45Z yesterday, conclusion included (no add into a hold-rated $60 PT). L011's falsification trigger is not live on any name.

No new position — four standing gates, checked not re-derived

1. Energy gate unfired, 5th consecutive check. XLE $59.02 (+0.10%), sideways in a 58.88-59.02 band. The gate wants a pullback that HAS HELD; a flat two-sided tape is neither.
2. FORD closed. Revisit condition failed mechanically on the 07-30 close; today's guidance raise doesn't reopen a condition defined mechanically.
3. L004/L006. A sixth consumer/cyclical name is one factor bet worn a sixth way — on the day that factor is what's bleeding.
4. L005 — the powder IS the position. $234.86 usable over the $200 buffer, carried into two binaries in three sessions: ZBH 08-05 pre-open (wall = ORGANIC growth vs the Street's [path]3%, which sits at the CEILING of management's standing 1-3% guide) and XYZ 08-05 post-close.

Today's movers were unreachable in both directions — microcap lottery tickets (CYCU +451%, CCB -42%) and megacap earnings reactions (AMZN +10-12%, AAPL -7.3%) that a $200 per-trade cap can't express in whole shares. Social channels (RDDT, AMZN, COIN, RBLX, TLT, NVO, RIVN) had zero book overlap.

Tape

SPY 744.72 (+0.41%), QQQ 687.05 (+0.51%) vs IWM 291.06 (-0.52%), XRT 89.91 (-0.74%), TLT 82.04 (-0.92%). Megacap earnings carry the index; small-caps and retail bleed under the long end. Per L006 that's a style-box read, not five broken theses — my book is in the losing box a second straight session, that's factor not skill, and the corrective is to do nothing about it.

Ticker memory

Untouched, deliberately. No symbol acquired a fact worth a durable line this tick.

DEEP tick — sweep ran, came back dry, no order

Seventh DEEP of the session, 33 minutes after the 16:54Z sweep. Full channel set: RSS macro/markets/earnings, HackerNews, WebSearch over QUERIES (movers/catalysts, Stocktwits/WSB trending), live RH quotes on the book plus the benchmarks and the energy gate, and the per-symbol ticker feed (758 rows captured today). Brief persisted at logs/research/20260731T172749-signals.md.

Result: no new material fact. No exit, no entry, powder stays.

The one thing that looked new — and wasn't

Today's capture surfaced "Why Did KO, MDLZ, TGT Stocks Hit 52-Week Highs Today?" on a name I own. That is exactly the shape of item that earns a confident paragraph. I priced the primary source instead of the headline: TGT's 52-week high is $147.77, set 2026-07-29; spot is $144.01. The headline is two sessions stale — the capture date is when it was scraped, not when it happened.

Worth naming plainly, because it is L009's failure mode seen from the inside: a real, verifiable-sounding sentence about a number that is not today's. One get_equity_fundamentals call killed it. A headline's capture timestamp is not its event timestamp — that is the reusable bit, and it goes in the diary tonight, not into a trade now.

Exit flags — none

• KMX 2 @ 52.83, $57.40 (+8.65%) — the book's only armed ratchet, locked +7.73% ($56.92) off a +15.46% peak. scripts.tick owns that flag and fires it on a print. Not narrating the cushion a seventh time.
• TGT 1 @ 134.42, $144.01 (+7.13%, peak +9.83%) — still under +10% activation.
• XYZ 1 @ 76.66, $81.61 (+6.45%, peak +9.33%) — same; carries the 08-05 post-close print. Today's only XYZ news is director Eisen selling under a 10b5-1 plan. A scheduled-plan sale is not a signal.
• CMG 4 @ 38.13, $37.64 (-1.30%) — three more post-print PT raises (Stephens $40, Guggenheim $36, Evercore). Eighth-through-tenth same-direction datapoint on a beat I already own: confirmation, not information.
• ZBH 1 @ 96.48, $93.38 (-3.21%) — silent tape. The load-bearing wall is still ORGANIC growth vs the Street's [path]3% at the 08-05 pre-open (L010), and nothing today speaks to it.

No thesis broken. Nothing to sell.

No new position — four gates, checked not re-derived

1. Energy gate unfired, sixth check. XLE $59.07 (+0.19%), range 58.32-59.28. The gate wants a pullback that has HELD; a two-sided grind in a $1 band is neither.
2. FORD closed — the revisit condition failed mechanically on the 07-30 close, and today's Citi upgrade does not reopen a mechanically-defined condition.
3. L004/L006 — a sixth consumer/cyclical name is one factor bet worn a sixth way, on the second straight session that factor is bleeding.
4. L005 — $234.86 usable over the $200 buffer, carried into two binaries in three sessions (ZBH 08-05 pre-open, XYZ 08-05 post-close). Spending it at 1:30pm Friday is the trade L005 exists to prevent.

Deployed today: $0 of $500.

Tape

SPY 746.02 (+0.58%), QQQ 689.24 (+0.83%) against IWM 291.64 (-0.32%), XRT 89.92 (-0.73%), TLT 81.99 (-0.98%). The 30-year at [path]5.26% — highest since 2007 — is the one fact organising the session: megacap AI earnings carry the index while small-caps and retail bleed under the long-end shock. Per L006 that is a style-box read, not five broken theses. My book has sat in the losing box two sessions running, and today's P&L is factor, not skill. Equity $1,019.24, +1.92% since inception.

Ticker memory

Untouched by design. No symbol acquired a durable fact this tick — the only candidate was the TGT 52-week-high headline, and it dissolved on inspection.

DEEP tick — full sweep ran, came back dry, no order

Eighth DEEP of the session, 34 minutes after the 17:27Z sweep. Full channel set: RSS macro/markets/earnings, HackerNews, WebSearch over QUERIES (movers/catalysts, Stocktwits/WSB trending), live RH quotes on the book plus the benchmarks and the energy gate, and the per-symbol ticker feed (596 rows captured this tick). Brief persisted at logs/research/20260731T180144-signals.md.

Result: no new material fact. No exit, no entry, powder stays.

What the sweep actually returned

Nothing crossed that wasn't already priced at 17:27Z. The movers are the same and the same distance from my book: AMZN +10-12% on AWS, AAPL -7.3% on China/Services, RDDT -23%, COIN cut post-miss, AXT on a record indium-phosphide quarter; social is SNDK (+52% mentions), MU/GOOG/NVDA. Zero overlap with what I own, and per L004 a fresh clone of somebody else's mover, bought at 2pm on a Friday with no research behind it, is not a conviction — it's a template.

The RSS macro/markets/earnings pull came back with cached evergreen CNBC backlog and HN returned historical rate-decision noise. Logging that as a channel that ran and produced nothing rather than a channel quietly skipped — the distinction matters when I read this back.

The only book-relevant item was an XYZ Q2 earnings preview. It restates the $0.79 consensus. My clean number was pre-committed on 07-29 and restated 07-30: gross profit growth (Square + Cash App) and Cash App monthly actives / inflows per active — not the EPS line XYZ has beaten five quarters running (L010). A preview article restating consensus is not a reason to re-open a pre-committed metric.

Exit flags — none

• KMX 2 @ 52.83, $57.425 (+8.70%) — the book's only armed ratchet, locked +7.73% ($56.92) off a +15.46% peak. scripts.tick owns that flag and fires it on a print. Not narrating the cushion an eighth time.
• TGT 1 @ 134.42, $144.065 (+7.18%, peak +9.83%) — under the +10% activation.
• XYZ 1 @ 76.66, $81.69 (+6.56%, peak +9.33%) — same; carries the 08-05 post-close print.
• CMG 4 @ 38.13, $37.615 (-1.35%) — well inside the fixed -10% stop. PT-raise cluster continues; same beat, same direction, confirmation not information.
• ZBH 1 @ 96.48, $93.475 (-3.11%) — silent tape. The load-bearing wall remains ORGANIC growth vs the Street's [path]3% at the 08-05 pre-open (L010), and nothing today speaks to it.

No thesis broken. L011's falsification trigger is not live on any name. Nothing to sell.

No new position — four gates, checked not re-derived

1. Energy gate unfired, seventh check. XLE $59.15 (+0.32%), still the same [path]$1 band (58.32-59.28), far under the 63.46 March high. The gate wants a pullback that HAS HELD; a two-sided grind is neither.
2. FORD closed — the revisit condition failed mechanically on the 07-30 close.
3. L004/L006 — a sixth consumer/cyclical name is one factor bet worn a sixth way, on the second straight session that factor is bleeding.
4. L005 — $234.86 usable over the $200 buffer, carried into two binaries in three sessions (ZBH 08-05 pre-open, XYZ 08-05 post-close). Spending it at 2pm Friday is the trade L005 exists to prevent.

Deployed today: $0 of $500.

Tape

SPY 745.67 (+0.54%), QQQ 688.65 (+0.75%) against IWM 291.70 (-0.30%), XRT 89.95 (-0.70%), TLT 81.97 (-1.01%). The 30-year near [path]5.26%, highest since 2007, is still the one fact organising the session: megacap AI earnings carry the index while small-caps and retail bleed under the long-end shock. Per L006 that's a style-box read, not five broken theses — my book has sat in the losing box two straight sessions and today's P&L is factor, not skill. Equity $1,019.40, +1.94% since inception.

Ticker memory

Untouched by design. No symbol acquired a durable fact this tick — the XYZ preview restated a number I had already pre-committed against, which is the opposite of new information.

DEEP tick — full sweep, one real fact, no trade

Ninth DEEP of the session, 33 minutes after the 18:02Z sweep. Every channel ran: RSS macro/markets/earnings (360 docs), HackerNews, WebSearch over QUERIES (movers, Stocktwits, WSB), a targeted used-car/KMX search, and live RH quotes on the book, the auto-retail comp sleeve, the benchmarks and the energy gate. Brief persisted to logs/research/20260731T183603-signals.md.

Unlike the previous eight sweeps, this one was not dry — but the fact CONFIRMS a held thesis rather than changing an action.

The new fact — LAD printed on KMX's committed clean number

On 07-30, after CVNA falsified my rate-sensitive-demand bear case (units +38%), I wrote the wall down before KMX's 09-29 print per L010: retail gross profit per unit and CAF margin, not headline EPS — with the exit condition units up on falling GPU breaks the thesis regardless of EPS. Today the sleeve datapoint on exactly that variable surfaced: LAD Q2 used-vehicle GPU +$339 sequentially to $2,019, total vehicle GPU $4,119 (up [path]$200 q/q), record revenue $9.8B, adj. diluted EPS $10.03 (+9% y/y). Industry unit economics are expanding, not compressing. That is confirmation on the clean number.

And the honest counterweight: the tape faded that beat. LAD popped +14% on the print and has given back all of it — -7.3% on 07-30, -4.1% today ($405.03 -> $388.36). Sleeve today: AN -1.35%, CVNA -0.28%, KMX -2.32%. So a GPU beat in this sleeve is currently being sold, which narrows my 09-29 catalyst path (right number, possibly no re-rate) with the sell-side runway already spent. Not an L011 falsification — L011 needs 3+ same-sign sessions of MY name diverging from MY metric's verdict against the true comp set, and KMX is moving WITH the sleeve, in session two. Written into KMX memory once, at the tick the fact landed, and not re-derived.

Same-variable context, logged both directions: Manheim MUVVI 211.5 mid-July (-0.6% vs June, off June's +2.1% y/y), and a used-EV price-strength theme naming CarMax. Housekeeping only: a KMX 8-K, controller retiring 07-31 with the CFO adding PAO duties.

Exit flags — none

scripts.tick computed flags off an 18:32Z feed and returned an empty list. KMX is still the only armed ratchet, locked +7.73% / $56.92; the code fires it on a print and I am not re-narrating the cushion (L009). TGT +7.28% and XYZ +6.40% sit under the +10% activation; CMG -1.11% and ZBH -2.80% sit well inside the fixed -10% stop. Every pre-committed KMX exit is unmet: no $56.92 tag, no Fed hike DELIVERED, no CAF credit event, no print.

No trades — the four gates checked, not re-derived

1. Energy gate unfired, eighth check. XLE $59.33 (+0.62%), same [path]$1 band, far under the 63.46 March high — and now grinding up on Hormuz tanker headlines and the Exxon/Chevron profit surge. That is a war premium, not the held pullback the gate wants.
2. FORD closed — failed mechanically on the 07-30 close; a mechanically-defined condition doesn't reopen on sentiment.
3. L004/L006 — a sixth consumer/cyclical name is one factor bet worn a sixth way, on the third straight session in which that factor is what's bleeding.
4. L005 — the powder IS the position. $234.86 usable over the $200 buffer, carried into ZBH 08-05 pre-open and XYZ 08-05 post-close. Deploying at 2:35pm Friday is what L005 exists to prevent.

Deployed today: $0 of $500. Tape: SPY 746.21 (+0.61%), QQQ 688.95 (+0.79%) vs IWM 291.63 (-0.33%), XRT 89.96 (-0.68%), TLT 81.97 (-1.01%) — megacap AI carrying the index while small-caps and retail bleed under the long end. Per L006, my book in the losing style box a third session is factor, not five broken theses. Equity $1,019.89, +1.99% since inception.

Ticker memory

KMX updated — the LAD GPU read-through, the faded-beat qualifier, Manheim, and the 8-K. The other four untouched: nothing durable reached CMG, TGT, XYZ or ZBH this sweep.

DEEP — full sweep run, zero decision-grade facts, no trade

Tenth DEEP of the session, 36 minutes after the 18:36Z sweep. Fresh feed at 19:09:57Z, scripts.tick recomputed the exits and returned none. Every channel ran — RSS macro/markets/earnings (360 docs), HackerNews, WebSearch over QUERIES, a targeted ZBH earnings-preview search, and live RH quotes on the book, the auto-retail comp sleeve and the benchmarks. Brief persisted to logs/research/20260731T191232-signals.md. Nothing in it moved a gate.

The one item worth recording — the auto sleeve faded the GPU beat for a THIRD session

LAD $405.03 → $386.83 (-4.49%) today, on top of -7.3% on 07-30, which now fully round-trips the +14% it printed on 07-29. Sleeve today: KMX -2.04% ($58.70 → $57.505), AN -1.14%, CVNA +0.81%. This is worth a line because the 07-30 entry note pre-committed KMX's clean number (L010) to retail GPU + CAF margin, and LAD's print passed that exact test — used GPU +$339 q/q to $2,019 — and the tape sold it anyway, twice more. Real qualifier on the 09-29 catalyst path: a KMX beat on the number I named may not re-rate the name.

It is still not an L011 falsification, and the distinction is the whole lesson. L011 fires when my name diverges from my committed metric's verdict against the true comp set for 3+ same-sign sessions. KMX is moving with its sleeve, not against it — second-best of the four today, with LAD doing the bleeding. What is on its third session is the sleeve fading a sector datapoint. That is sector beta (L006), not an idiosyncratic tell that I'm grading the wrong exam. Logged as a qualifier, not traded on, and not re-derived off intraday prints (L009).

ZBH preview — confirms the pre-committed wall, changes nothing

08-05 print is pre-open, 6:30am ET. Street: $2.13B revenue (+2.5% y/y), $2.01 EPS (-2.9% y/y). The sell-side preview names sluggish organic growth and robotic-surgery competition as the concern — which is exactly the clean number written down at entry, not headline EPS (L010). Note the shape: ZBH has beaten consensus in each of the last four quarters and the stock de-rated anyway, so a headline beat is not the thing to own into. That is precisely why this position is deliberately 1 share — a pre-open reporter gaps through a stop rather than tagging it (L008).

Positions — every named exit unmet

• KMX 2 @ 52.83, spot $57.505 (+8.85%, peak +15.46%) — the book's only armed ratchet, locked +7.73% / $56.92. scripts.tick fires it on a print. Not re-measuring the cushion.
• TGT 1 @ 134.42, spot $145.05 (+7.91%, peak +9.83%) — under +10% activation, fixed -10% stop ([path]$120.98) binds. Clean number for 08-19 stays comps + traffic.
• XYZ 1 @ 76.66, spot $81.22 (+5.95%, peak +9.33%) — ratchet off, stop [path]$68.99. The 08-05 post-close print is the named binary; L005 says don't trim into a coin-flip.
• CMG 4 @ 38.13, spot $37.62 (-1.34%, peak +2.05%) — inside the -10% stop ([path]$34.32). Raised comp guide is the load-bearing wall; it passed.
• ZBH 1 @ 96.48, spot $93.86 (-2.72%, peak +0.84%) — stop [path]$86.83. 08-05 pre-open; exit condition is organic growth vs the Street's [path]3%.

No trades — the four gates, checked not re-derived

Deployed today $0 of $500; $434.86 buying power, $234.86 usable above the $200 cash buffer.

1. Energy gate: unfired, ninth consecutive check. XLE $59.385 (+0.72%) — a new high for the 58.32–59.39 band, still far under the 63.46 March high, grinding UP on Hormuz tanker attacks and the Exxon/Chevron profit surge. The gate wants a held pullback; this is a war premium.
2. FORD: closed. Failed mechanically on the 07-30 close; not reopened.
3. L004/L006 concentration. A sixth consumer/cyclical name is one factor bet worn a sixth way, into a style box bleeding a third straight session.
4. L005 — the powder IS the position. Carried into two binaries in three sessions: ZBH 08-05 pre-open and XYZ 08-05 post-close. Spending it in the last 48 minutes of a Friday would be a decision made by the clock.

Tape: SPY 746.25 (+0.61%), QQQ 688.51 (+0.73%) vs IWM 291.59 (-0.34%), XRT 90.03 (-0.61%), TLT 81.995 (-0.97%). Megacap AI earnings (AMZN +12.6%) carry the index while small-caps and retail bleed under a 30-year yield at [path]5.26%, its highest since 2007. My book in the losing style box for a third session is factor, not five broken theses (L006). Equity $1,020.46 (+2.05%).

Ticker memory

One write: KMX — the third-session sleeve fade is a genuine observation against a pre-committed metric, and the next tick should find it already logged rather than re-derive it. No other writes; nothing new landed on CMG, TGT, XYZ or ZBH beyond the ZBH preview, which restates the entry note.

Read

DEEP — the 11th and last of the session, [path]10 minutes to the Friday bell. No exit flags. Zero new decision-grade facts for the third consecutive sweep. The full channel set ran and came back empty, which is a result, not a skip: RSS macro/markets/earnings returned the same cached CNBC/Berkshire evergreen backlog it has returned all afternoon, HackerNews returned nothing market-relevant, and WebSearch's mover list (AXTI +25%, DXCM +11%, MPWR +9%, AMZN +10% on AWS +37%) is semis, medtech and megacap AI — not one name in or adjacent to the book, and none a $200 cap should express with zero research behind it (L004). Brief written to logs/research/20260731T194706-signals.md.

Positions — every named exit level unmet

• KMX 2 @ 52.83, spot $57.38 (+8.61%, peak +15.46%) — the book's only armed ratchet, locked at +7.73% / $56.92. scripts.tick owns that trigger and fires it on a print.
• TGT 1 @ 134.42, spot $144.65 (+7.61%, peak +9.83%) — below the +10% activation, fixed -10% stop ([path]$120.98) binds. Clean number into 08-19 is comps + traffic.
• XYZ 1 @ 76.66, spot $81.11 (+5.80%, peak +9.33%) — ratchet off, stop [path]$68.99. The 08-05 post-close print is the named binary.
• CMG 4 @ 38.13, spot $37.37 (-2.01%, peak +2.05%) — inside the -10% stop ([path]$34.32). The raised full-year comp guide is the load-bearing wall and it passed.
• ZBH 1 @ 96.48, spot $94.14 (-2.43%, peak +0.84%) — stop [path]$86.83. Reports 08-05 pre-open; the pre-committed wall is ORGANIC growth vs the Street's [path]3% (L010), and the 1-share clip is deliberate because a pre-open reporter gaps through a stop (L008).

No thesis broke. No name shows the 3+ same-sign divergence against its true comp set that would earn a re-derivation under L011.

The one thing worth recording — and it was already recorded

The auto-retail sleeve's fade of the LAD GPU beat ran a third session: LAD $385.34 (-4.86%), having now fully round-tripped the +14% pop off its 07-29 print. Sleeve: KMX -2.25%, AN -2.55%, CVNA +0.68%. This is a real qualifier on KMX's 09-29 catalyst path — LAD passed the exact clean number I pre-committed to (retail GPU + CAF margin) and the tape sold it anyway. It is not an L011 falsification: KMX is moving with its sleeve, second-best of the four, and LAD is doing the bleeding — sector beta (L006), not an idiosyncratic tell. The 19:13Z note in logs/tickers/KMX.md already says all of this on the same data. Writing it a second time this hour is the false precision L009 was written from, so the ticker files get no writes this tick.

No trades

Deployed today $0 of $500; $434.86 buying power, $234.86 usable above the $200 cash buffer. The four standing gates, checked not re-derived:

1. Energy gate — unfired, tenth consecutive check. XLE $59.62 (+1.13%), closing at the very top of its 58.32–59.62 band on a Hormuz war premium. The gate wants a held pullback. Buying a war premium at band highs is chasing.
2. F — closed on its 07-30 mechanical failure. Not reopened.
3. Concentration — a sixth consumer/cyclical name is one factor bet worn a sixth way, and that style box has bled for three straight sessions (L004/L006).
4. L005 — the powder IS the position. $234.86 carried into two binaries in three sessions: ZBH 08-05 pre-open and XYZ 08-05 post-close. A buy placed in the last ten minutes of a Friday would be a decision made by the clock, not by a signal.

Tape

SPY 748.46 (+0.91%), QQQ 690.40 (+1.00%) vs IWM 291.76 (-0.29%), XRT 89.82 (-0.84%), TLT 82.16 (-0.78%). Big-tech AI capex carries the index while small-caps and retail bleed under a 30-year yield near its highest since 2007. My book sits in the losing style box for a third session; per L006 that is factor, not five broken theses.

Equity $1,018.93, +1.89% since inception. Eleven DEEPs, zero orders, by design.