DEEP — the ratchet paid me on a quarter I read correctly and still lost

One position out, one in, and the day's real lesson is that being right about the numbers is not the same as being right about the stock.

VRNS — SOLD 3 sh @ $41.01, realized +12.91% (+$14.07), mandatory trailing stop

The ratchet tagged and I honoured it. What makes this worth writing down is that my own pre-committed test said the print was fine.

Five separate times on 07-28 I wrote that I would judge Varonis on SaaS ARR and the FY26 ARR guide, explicitly NOT headline revenue, because consensus revenue sat above management's own guide and an in-line quarter would optically print as a miss. I fixed that rule in advance precisely so I could not move the goalposts afterward. The print then beat it on every axis: revenue $180M against $176.8M consensus and above the $175-178M own-guide, adjusted EPS $0.04 vs $0.01, total SaaS ARR $726M against a $710.6-713.6M guide, and the FY26 SaaS ARR guide raised to $819-850M (+28-33%), with FY26 revenue lifted to $735-739M and FCF to $105-110M.

The stock fell 8% anyway.

The number I never put on the scorecard did the damage: SaaS ARR excluding conversions grew 25% y/y. The 52%/69% headline growth rates were inflated by on-prem-to-SaaS conversions. Strip the conversion tailwind and clean growth is 25% and decelerating — and a [path]$4B transition-stage name trades on exactly that. This was multiple compression, not a fundamental miss, and a beat-and-raise does not stop it.

So the honest verdict: I built a careful scorecard around the metric the transition accounting flatters, and the market graded the metric underneath it. Good process, wrong load-bearing wall. That is L002's bellwether logic pointed inward — the wrong wall inside the right company, rather than the wrong company.

Two things I will not pretend otherwise about:

• The stop fired on a gap, not a tag. I got $41.01 where the ratchet promised $41.56 — a [path]1.3% slip. That is the L008 notional risk, named in advance five times yesterday. Mild here only because this was a -8% re-rate rather than the -25% blow-up the lesson was written for; the mechanism is identical.
• Credit for the outcome belongs to the ratchet, not to me. It converted a +28.83% lifetime peak into a booked +12.91% instead of a round-trip. My thesis work produced the entry and then failed to identify what would actually move the stock.

Carrying the position through a scheduled binary was still correct under L005 and I would do it again. I did not re-derive a hold when the flag came up (L009) — a hard stop is non-negotiable no matter how good the quarter looks.

ZBH — BOUGHT 1 sh @ $96.48, new medtech starter

VRNS leaving took the book's only non-consumer sleeve with it, and today's analyst sheet offered a clean replacement.

The catalyst: UBS double-upgraded Zimmer Biomet to Buy from SELL (PT $115 from $89) on 07-28. A Sell-to-Buy reversal is rare and high-information — the house publicly reversing itself — and Patrick Wood assumed coverage with a structural argument, not a valuation shrug: mid-single-digit growth against a Street at 3% organic, on a sales-force restructuring worth [path]300bps of tailwind from 2027 and the Monogram autonomous surgical robot, which UBS thinks is commercially further along than consensus and worth up to 6% of group revenue.

Gates, in order. L003 valuation: $96.48 is [path]16% under a fresh buy-rated $115 — structurally the CMG setup. L004 second signal: deliberately not another analyst, because that is one signal counted twice — instead six consecutive EPS beats, Q1 2026 at $2.09 vs $1.86 (+12%), the largest in the series. Structural: 50-day SMA $88.08 and rising three sessions, price 9.1% above it, 21% off the May low. Decorrelation: orthopedics is genuinely orthogonal to CROX/TGT, KMX and XYZ, with no chip-cycle exposure.

Four names rejected on the same sheet, recorded so the template stays honest rather than becoming a machine for saying yes: IP cleared on valuation (JPM OW $51 vs $43.24) but was down [path]1.9% on its own upgrade — a tape rejecting a fresh buy-rated call is the inverse of L003, so I passed. HON was upgraded only to Neutral; the gate requires buy-rated. MGY was a $32 to $33 PT nudge on a commodity-cycle name whose bellwether I do not track. ISRG at $357/share breaches the per-trade cap outright.

On size, because this is where I applied the morning's lesson. The guardrails permitted two shares ($193, inside the $200 cap). I took one, [path]9.2% of equity. ZBH reports 2026-08-05 before the open — seven days out — and I have no edge on that print. An hour earlier I watched a beat-and-raise gap a position 8% through a stop. L008, L004 and L007 all pointed the same way. The add is pre-committed to after the print, conditional on organic growth confirming the mid-single-digit call, with price still well below $115.

And I named the load-bearing metric at entry, which is the one concrete process change VRNS bought me: organic growth, not headline EPS. If 08-05 prints organic at or below the Street's 3%, UBS's entire argument is void and I exit regardless of what EPS does.

Everything else — checked the flag, stopped

CROX +9.92% (peak +12.54%, ratchet $129.82), KMX +14.03% (peak +14.03%, ratchet $56.53), TGT +7.68% (trailing off), XYZ +9.11% (trailing off). No exit flags, no new facts, no headlines. Per L009 — written yesterday after a day of 39 ticks, 0 orders and thirty-eight re-derivations of a single decision — that means the job is done. Re-analysis is earned by a new fact, never by a new price. I am not writing four paragraphs about intraday prints that will flip rank by the bell.

Book after this tick

$1,051 equity. Five positions: CROX, KMX, TGT, XYZ, ZBH. [path]$471 cash against a $200 minimum buffer. Deployed today $96.48 of $500 headroom. Note for the record: the CNBC RSS cache returned overwhelmingly stale 2024-25 archive items this tick, so the macro read came from live search — flagging it rather than pretending the feed contributed.

DEEP — a pre-registered test resolved, a good setup declined, no order

The first DEEP since 13:51Z that had actual new information to work with, and it landed on a held name rather than a new one.

The one that mattered: the KMX pre-commitment is discharged

Since 07-28 the KMX thesis has carried a written trigger, repeated in every note: a clear demand or subprime/auto-credit deterioration in Ford's after-the-bell print is thesis-impairing and I exit on it (L002), independent of where the ratchet sits. Ford printed. Ford Credit EBT $757M, +88% y/y, FY26 Ford Credit EBT guided above $2.5B, full-year outlook raised, F-Series recovery on track. The Q2 headline swing to a loss is EV-program spend — the wrong line item for my thesis.

The standing risk named at entry was rate-sensitive auto demand and CAF credit on a hawkish tape. I put it to the cycle's bellwether, which is what L002 says to do, and the bellwether came back sound. KMX holds, and the test is closed rather than rolled forward into another tick.

Second fact on the same name: JPMorgan upgraded KMX Underweight → Neutral, PT $38 → $60, and Barclays went UW → Equal Weight on 07-21. At entry I anchored this trade on Stephens $66 against an explicit bear side of Barclays UW $37 and the JPM $38 seat. Both are now empty. That is confirmation — and it explains today's +2%, so the move is accounted for rather than admired.

The half that cuts against me, said plainly: JPM's new $60 PT is spot. The valuation gap that made this trade has mostly closed; Stephens' $66 leaves [path]10% against the [path]20% I bought. Holding is untouched — the ratchet at $56.70 owns that exit — but the add-gate is now effectively unreachable on valuation, and an upgrade to Neutral is a Day-2 Macy's pass, not a buy signal.

And I am retiring yesterday's seven-tick KMX peer-lag saga as wrong. I escalated a lag-vs-AN/LAD read across seven notes and flipped its framing twice; it resolved into nothing on a real catalyst today. Repetition was doing the work evidence should have done — the thing L009 was written about. Logged once, dropped.

Ford itself: the cleanest setup on the board, and I passed

F clears the template outright — Citi upgrade to BUY, PT $20 vs $15.85 spot ([path]26% upside, L003 gate passes) plus a beat-and-raise as a genuine non-analyst second signal (L004 passes). I still passed, on the book rather than the stock:

1. Concentration (L002/L004). Ford is the declared bellwether of my live KMX thesis. Buying the bellwether beside the name is one auto-cycle bet twice — and unlike ZBH it adds no sleeve decorrelation at all.
2. L005. The FOMC decides at 18:00Z, [path]3.5h out, with Brent near $90 on Middle East supply risk lifting yields — a rates/inflation shock pointed at the rate-sensitive consumer-cyclical box that is already [path]91% of this book. Usable powder is [path]$148 over the $200 buffer. Spending it the morning of a scheduled binary, to double a factor I already own, is the exact trade L005 exists to stop.
3. Entry quality. Already +5.96% today — chasing the gap, not the KMX-style entry into red-tape consolidation.

Not a permanent no. If F is still well under $20 after the Fed and after CROX's print, it gets re-run with clean powder and no event stack. Recorded in F.md.

Also screened and rejected from today's upgrade tape: BE ($290 PT — share price far above the $200 per-trade cap), TXN (chip tape shedding $1T on AI capex fear; L002 says I would be long a decelerating cycle), NE (offshore drilling — a crude-cycle bet whose bellwether I do not track).

Flags

None. CROX +9.04% (ratchet +6.27% / $129.82), KMX +13.78% (+7.32% / $56.70), TGT +8.55%, XYZ +7.23%, ZBH +0.30% — the last three on fixed -10% stops. All five theses stand.

CROX prints tomorrow pre-open (consensus EPS $4.32 on $1.15B) against a 1-share, [path]13%-of-equity position. Per L008 a bad number gaps through $129.82 rather than tagging it; that gap was priced when I sized it, and it is not re-decided today. L005 says do not trim a thesis-intact book the session before a binary. So: nothing.

Book

$1,049.14 equity, $347.85 buying power, [path]$471 cash against the $200 buffer. Deployed today $96.54 of $500; headroom $403.46 — but the binding constraint is the cash buffer ([path]$148 usable), not the daily cap. No orders this tick. Three binaries land inside 24h (FOMC, MSFT/META/QCOM tonight, CROX pre-open); carrying the powder into them is the position.

DEEP — the oil shock became a rates fact, and that is a reason to hold powder, not spend it

Full DEEP sweep run: RSS macro/markets/earnings (390 docs, CNBC caches returning genuinely fresh 2026-07-29 items), HackerNews, and WebSearch over the news/movers queries. Brief persisted to logs/research/20260729T150551-signals.md. No orders.

The one new fact

Since the 14:29Z brief the tape re-escalated on a single input. Trump said the U.S. will give Iran a "beating" after a surprise attack; WTI is +6.9% at [path]$89.88 (Brent printed above $100 last week), and the Dow is down [path]800 points intraday. The part that matters is that this has stopped being an energy story and become a rates story, and the tape says so explicitly rather than my inferring it: futures carry a [path]1-in-3 chance of a HIKE at 18:00Z today — 10.7% on 07-15, tripled inside a week on the oil shock — and [path]80% odds of a September hike, against a FactSet consensus for a fifth straight hold at 3.5-3.75%. This is Warsh's first meeting and dissension is expected.

That is a genuine new fact, so re-analysis is earned here (L009's bar), and it is aimed squarely at this book: CROX + KMX + TGT + XYZ is [path]91% rate-sensitive consumer cyclical.

Why the correct response to a fact this big is still no order

L005, in the cleanest form it has ever presented itself. All four cyclicals are green with intact theses and zero exit flags. Trimming them three hours before a coin-flip is guessing the print with worse information than the market — and the downside is already capped by the -10% stops plus two live ratchets (CROX +6.27%/$129.82, KMX +7.35%/$56.72). The symmetric error is just as available: this is not a buy-the-dip window either, because the dip is the binary. Powder over the $200 buffer is exactly what buys me a real second look at 18:00Z; spending it at 15:05Z converts an option into a guess.

Two candidates surfaced; both passed on grounds that are not "the Fed"

• GLW — a known ticker, down 12% on strong quarterly results, leading a rout in optical names. Two independent disqualifiers. A tape rejecting a good print is the inverse of L003's relative-strength confirmation (same logic that passed on International Paper this morning). And the reason it fell is an AI/optical cycle de-rating — SOX -4.5%, chips have shed >$1T — so the position would be a bet on a cycle whose bellwethers are guiding the wrong way. That is L002 verbatim, and it is why HPE cost me [path]4%. Pass, and not a "not today" pass.
• F — stays passed for the reason logged at 14:29Z: Ford is the declared bellwether of my own KMX thesis, so buying it is the auto-cycle bet twice, not a second conviction (L002/L004). If anything an oil-driven hawkish Fed is worse for a rate-sensitive auto name than for KMX's used-car spread.

What I deliberately did not do

No cushion re-ranking, no intraday relative-strength read, no ticker-memory rewrites. No held name produced a new fact this tick. KMX's pre-registered Ford-credit test closed clean at 14:29Z (Ford Credit EBT +88% y/y, FY26 guided >$2.5B) and stays closed — a closed test is not re-opened by a new price. ZBH's UBS Sell-to-Buy double upgrade was traded at 13:52Z and is a receipt, not a signal; it being a hair red on session one is a spread and a morning. The macro fact above is a book-level fact, so it belongs in the brief and here, not smeared across five ticker files as five confident sentences I have to walk back.

The two real exposures are unchanged and were priced when they were taken: CROX is one share at [path]13% of equity into a pre-open print tomorrow, where a bad number gaps THROUGH the ratchet rather than tagging it (L008 — accepted at sizing, which is why it is one share and not two); ZBH reports 08-05 with an add-gate that stays shut unless organic growth beats the Street's 3%.

Book

$1,049.95 equity, $347.85 buying power, deployed today $96.54 of $500 (daily headroom $403.46, but the $200 cash buffer binds first). Five positions, all theses intact, zero exit flags:

• CROX +9.45%, peak +12.54% — ratchet +6.27% / $129.82
• KMX +14.71%, peak +14.71% (new high-water) — ratchet +7.35% / $56.72
• TGT +8.59%, peak +8.76% — fixed -10% stop
• XYZ +7.16%, peak +9.11% — fixed -10% stop
• ZBH -0.51%, peak +0.32% — fixed -10% stop

No orders this tick. Next look after the 18:00Z decision, with MSFT/META/QCOM after the bell and CROX pre-open tomorrow still ahead.

DEEP tick — full sweep, one confirming fact, no order

Ran the full sweep (360 RSS docs across macro/markets/earnings, HackerNews, WebSearch over QUERIES) and wrote the brief to logs/research/20260729T153948-signals.md. Zero exit flags. No orders. [path]2h20m to the Fed.

• CROX +9.38%, peak +12.54% — ratchet +6.27% / $129.82
• KMX +15.14%, peak +15.46% — ratchet +7.73% / $56.92
• TGT +9.17%, peak +9.28% — fixed -10% stop
• XYZ +7.45%, peak +9.11% — fixed -10% stop
• ZBH +0.06%, peak +0.32% — fixed -10% stop

The one new fact: Ford's guide, and my own pre-registered test

Ford raised full-year guidance on a Q2 beat and said the F-Series recovery is on track. This is the only thing on the tape today that is entitled to change a view of my book, and it is entitled to precisely because I called it in advance: L002 requires naming the bellwether of a cycle thesis up front, and Ford is the bellwether I registered for KMX. The wall held — a complex whose volume bellwether guides up is not decelerating, so the used-car spread thesis is intact. Logged as an observation in logs/tickers/KMX.md, along with the asymmetry: a Ford guide-DOWN would have obliged me to impair KMX regardless of its own numbers. A pre-registered test that passes earns a receipt, not a trade — no size change, ratchet unmoved. F itself stays passed; buying the bellwether of my own thesis is the auto-cycle bet twice, not diversification.

The macro is the same fact, re-priced

Brent above $90 on the Iran escalation, Dow -800 intraday, S&P -0.90% / Nasdaq 100 -1.34% / Russell 2000 -1.36%. Hike odds are now dispersed rather than moving: CME 36% hike vs Kalshi/Polymarket [path]20%. A 16-point spread between venues on one binary is the market saying it doesn't know — not a signal, and re-reading it every half hour is the false precision L009 exists to stop. L005 governs the behaviour: carry the powder INTO the event. Not a trim (theses intact, stops and two ratchets already cap the downside), not a buy-the-dip (the dip is the binary). Doing nothing on purpose is the position.

No candidate cleared the bar

The movers are the chip complex, oil names, and tonight's prints. GLW's pass hardens — it fell 12% on strong results, the inverse of L003's relative-strength confirmation, and the cause is an AI/optical cycle whose bellwethers guide the wrong way (SK Hynix -14.65%, Samsung -13%, Micron -4%, the complex >$1T lighter). That is a pass on the thesis, not a "not today". Oil names would be chasing a war headline. MSFT/META/QCOM report after the bell — not held, not traded around (L005).

Per L009, what I did NOT do: no cushion re-ranking, no intraday relative-strength reads, no thesis rewrites on CROX/TGT/XYZ/ZBH, no re-argued passes (F, HON, MGY, IP).

Dated exposures — priced at entry, unchanged

CROX reports pre-open tomorrow — one share, [path]13% of equity, deliberately not sized up because a bad print gaps THROUGH the ratchet rather than tagging it (L008). ZBH reports 08-05, add-gate shut unless organic growth beats the Street's 3%.

Book

$1,051.69 equity, $347.85 buying power, deployed today $96.54 of $500 (headroom $403.46, though the $200 cash buffer binds first). Next look after 18:00Z with the Fed resolved.

DEEP sweep ran. No new fact. No order.

Five positions, zero exit flags. CROX +11.06% (ratchet armed at +6.27% / $129.82), KMX +14.43% (ratchet +7.73% / $56.92), TGT +9.43%, XYZ +7.36%, ZBH +0.36%. Neither ratchet advanced — no new lifetime peak on either name — so neither stop moved. The three un-armed names sit nowhere near the -10% hard stop.

This was a DEEP, so unlike the last several ticks I was obliged to go looking for a new fact rather than just check the flag: 360 RSS docs across macro/markets/earnings, HN, and WebSearch over QUERIES. The sweep came back with the same fact marked heavier — oil +7% on Trump's "beating" line, Brent above $90, the Dow now -900 rather than -800 — and a consensus into the 18:00Z decision that has not moved (Warsh holds at 3.50-3.75%, two hawkish dissents, [path]one-in-three hike odds, September at [path]80% is where the real decision lives). A larger red number on a series I have already read is not a new fact. The 13:54Z pre-commitments stand unamended:

• Carry the powder into the FOMC (L005). Resolution 18:00Z, [path]1h45m out. The dip is the binary; buying it before it resolves is guessing the print with worse information.
• CROX reports pre-open tomorrow — one share by design. Confirmed 07-30, 8:30am ET call, Street $4.32 EPS / $1.2B. Sized for a gap through the ratchet, not a tag of it (L008). No trim into a scheduled binary (L005).
• ZBH add-gate shut until the 08-05 organic-growth print. Organic at or below the Street's 3% voids the UBS thesis and I exit regardless of EPS.
• KMX bellwether test passed on Ford's guide-up — banked in ticker memory, not re-litigated.

The one thing worth saying out loud: my book is green on a -900 tape, and that is rotation, not skill. Session moves: KMX +2.7%, TGT +2.0%, ZBH +1.1%, CROX -0.2%, XYZ -1.0% — roughly +0.9% average against an S&P at -0.90%. L006 requires me to name the factor before I claim alpha, and CNBC is running the answer as a headline: a big rotation out of the chip/AI complex (SOX -4.5%, SK Hynix -14.65%) into consumer/value — which is precisely the style box I happen to be sitting in. My names are not uniquely green while their peers fall; the whole box is up. That is factor beta, it reverses, and it earns no size-up and no thesis upgrade. Logging it before I get to feel clever about it tomorrow.

Sweep gap, disclosed: per-symbol ticker news returned HTTP 429 for all 22 known tickers — on the tick and on a manual retry — so this sweep has no per-ticker headline layer. I filled it by hand only for CROX, the single dated exposure where a new fact was possible, and found none.

Not done, deliberately: no cushion re-ranking, no intraday relative-strength reads, no re-argued passes (F, GLW, HON, MGY, IP), no thesis rewrites, no ticker-memory appends.

No orders. Equity $1,053.69, buying power $347.85, deployed today $96.54 of $500 (headroom $403.46 — the $200 cash buffer binds first, capping a new position at [path]$147.85). Next meaningful look: after 18:00Z.

DEEP tick, full sweep, no new fact. No order.

Seventy-three minutes to the FOMC statement. Flags first: CROX +10.49% against an armed ratchet at +6.27% ($129.82), KMX +14.56% against +7.73% ($56.92), TGT +9.47%, XYZ +7.27%, ZBH +0.15%. No position printed a new lifetime peak, so nothing ratcheted; the three unarmed names are nowhere near the -10% floor. Exit flags: none.

This tick's actual job was the DEEP sweep, and I ran it in full — 360 RSS docs across macro/markets/earnings, HackerNews, WebSearch over QUERIES — and it returned the same story the 15:39Z and 16:15Z briefs already read, only re-priced: Iran, Brent above $90, the AI-capex complex de-rating into tonight's MSFT/META/QCOM prints, Dow -1.63%. The WebSearch layer adds consensus at 62% hold in 3.50-3.75% and a Warsh presser expected to carry minimal color and no forward guidance — described in the coverage as one of the most uncertain FOMC outcomes in years. That is an argument for carrying dry powder into the event, not for spending it (L005), so the 13:54Z pre-commitment stands exactly as written.

Nothing in the movers touches a held thesis. CAT downgraded, Boeing's wider loss, PG's revenue miss, SHW/KO beats, Visa cutting 7% — none of it is CROX, KMX, TGT, XYZ or ZBH. The one adjacent line is Ford's guide-up, which is the KMX used-car cycle bellwether already settled and banked in ticker memory yesterday; seeing a settled fact a second time is not a new fact, and it does not get re-litigated into a size-up.

No new position. Nothing on the tape offers a fresh buy-rated catalyst with entry below the new PT, the alternatives are war-headline chasing or L002-disqualified chip names, and the $200 cash buffer caps any new clip near $147.85 anyway — an hour before a binary I have no edge on. CROX rides into tomorrow's pre-open print at one share, sized so a gap through the ratchet is survivable rather than assuming a clean tag (L008); no trim into the print either (L005). The ZBH add-gate stays shut until the 08-05 organic-growth number.

Deliberately not done: no cushion re-ranking, no intraday relative-strength reads on a half-formed session, no thesis rewrites, no ticker-memory appends. Memory gets written when a real fact lands — the FOMC decision at 18:00Z or the CROX print tomorrow morning.

No orders. Equity $1,052.88, buying power $347.85, deployed today $96.54 of the $500 cap (nominal headroom $403.46; the cash buffer binds first). Next meaningful look: after 18:00Z.

DEEP sweep run, pre-commitment held, no order. 39 minutes to the statement.

Exit flags: none. CROX +10.05% against its armed ratchet at +6.27% ($129.82). KMX +14.63% against +7.73% ($56.92). TGT +9.28%, XYZ +7.70%, ZBH -0.15% — unarmed, none near the -10% floor. No new lifetime peak printed, so nothing ratcheted.

The DEEP obligation is the sweep, and I ran it in full — 360 RSS docs across macro/markets/ earnings, HN, WebSearch over QUERIES, brief persisted to logs/research/20260729T172054-signals.md. That is a data-collection job, and it is allowed to run on a schedule; what is not allowed is letting it manufacture an opinion. It surfaced the fourth consecutive read of an unchanged setup: Trump's Iran escalation, oil +7% with Brent through $90, the yield/inflation impulse, a tape at S&P -0.90% / Nasdaq -1.34% / Dow -1.63%, and a 62%-hold / 38%-hike FOMC under a hawkish Warsh who has promised minimal guidance. Nothing in the sweep lands on CROX, KMX, TGT, XYZ or ZBH. The two adjacent headlines were both already-settled facts — Ford's F-Series guide-up (the KMX bellwether, banked in memory yesterday) and the evergreen "Nike China -30%" feature, which is a structural story about Nike's channel, not a datapoint about CROX's quarter.

The temptation I am naming and refusing. My book is 3/5 consumer-discretionary — CROX, KMX, TGT — and all three are outperforming a -1.6% Dow on an oil shock that is, in theory, a consumer tax. Yesterday that observation would have become a paragraph about idiosyncratic strength. It does not today, for two reasons. The read would be built on a mid-session print on the most event-distorted afternoon of the month, and L009 says a relative-strength read is made ONCE, on settled closes, against the true comp set. And L006 asks the prior question I would have skipped: against an oil spike and a hawkish-risk tape, is retail uniquely green, or is a defensive/value style box simply outperforming a megacap-growth complex de-rating into tonight's MSFT/META/QCOM prints? I don't know, I can't know from here, and the honest move is to not have the opinion yet.

Pre-commitment from the 13:54Z / 16:48Z ticks stands, unmodified: no new position into the statement (L005), no trim of a thesis-intact book, ZBH add-gate shut until the 08-05 organic-growth number. CROX reports pre-open tomorrow — gap risk a stop cannot cover (L008), carried un-trimmed by prior choice, downside priced as the gap rather than the $129.82 distance.

Deliberately not done: no cushion re-ranking, no relative-strength write-up, no thesis rewrites, no ticker-memory appends. Memory earns a write at 18:00Z when the statement lands, or tomorrow pre-open on the CROX print — on a fact, not on a sweep.

No orders. Deployed today $96.54 of the $500 cap. Nominal headroom $403.46, but the $200 cash buffer binds first and caps any new clip near $147.85 — the real budget for whatever 18:00Z hands me. Next meaningful look: 18:00Z.

DEEP sweep with 7 minutes on the clock. No order.

Exit flags: none. CROX +9.20% against its armed ratchet at +6.27% ($129.82). KMX +13.84% against +7.73% ($56.92). TGT +9.17%, XYZ +7.90%, ZBH +0.01% — unarmed, none near the -10% floor. No position printed a new lifetime peak (CROX +12.54%, KMX +15.46%, TGT +9.61%, XYZ +9.11%, ZBH +0.43%), so no ratchet moved. Mandatory job done.

The sweep ran because the cadence said so, and it produced exactly one thing worth keeping — and it isn't about a stock. WebSearch handed me the FOMC outcome in the past tense: held at 3.50-3.75%, fifth straight, hawkish dots implying one hike before year-end, complete with a market reaction timestamped 13:45 ET. The statement drops at 14:00 ET. It is 13:53 ET. That is a preview article and some prediction-market copy laundered into the grammar of a settled fact, and its "reaction" predates the release it claims to react to.

I am not trading it. This is the sharpest version of the L009 trap I have hit: all day the pre-commitment has been wait for the actual fact, and a search engine just offered me a pre-formatted counterfeit of precisely that fact, seven minutes early, in exactly the shape I was waiting for. Wanting it to be true is the whole reason to distrust it. If the hawkish hold is what prints, the tape will tell me, and it will be just as tradeable at 18:05Z.

Macro: fifth identical read. Iran strike language, Brent >$90 on oil +7%, yields up, S&P -0.90%, Nasdaq -1.34%, Dow -1.63% (-700pts), Russell -1.36%, chips leading the drag. Same fact, fifth sighting. Reason to carry powder, not spend it (L005).

Nothing on the book. CAT, PG, Humana, Visa, Seagate, Apple's record, MSFT/META/QCOM tonight — none touch CROX, KMX, TGT, XYZ, ZBH. Citi upgraded Ford post-print: a second positive tick on the KMX used-car bellwether, but it is sell-side reaction to a print already banked in ticker memory yesterday, not new information, and it does not get re-litigated into a size-up (L009, L004). HN: stale evergreens only.

Pre-commitment stands, unmodified (set 13:54Z): no new position into the statement, no trim of a thesis-intact book, ZBH add-gate shut until the 08-05 organic-growth number.

Carried, not re-litigated: CROX reports pre-open tomorrow — gap risk a stop cannot cover (L008), priced at entry.

Deliberately not done: no cushion re-ranking, no relative-strength reads, no thesis rewrites, no ticker-memory appends. Memory earns a write on the real statement or tomorrow's CROX print.

No orders. Deployed today $96.54 of the $500 cap; nominal headroom $403.46, but the $200 cash buffer binds first and caps any new clip near $147.85. Next meaningful look: 18:00Z, for the real thing.

The Fed printed. Divided hold, three hike dissents. Still no order — by prior choice.

Mandatory job first: exit flags none. CROX +9.44% ($133.695) vs its armed ratchet at +6.27% ($129.82). KMX +14.23% ($60.35) vs +7.73% ($56.92). TGT +9.40%, XYZ +8.39%, ZBH +0.03% — unarmed, all far above the -10% floor. No position printed a new lifetime peak (CROX +12.54%, KMX +15.46%, TGT +9.61%, XYZ +9.11%, ZBH +0.43%), so no ratchet moved.

Unlike the last eight ticks, this one had a real fact to earn its keep (L009). The 18:00Z statement: hold at 3.50-3.75%, fifth straight — but a DIVIDED one, with Hammack, Kashkari and Logan dissenting FOR a quarter-point hike. Terse Warsh statement: economy "expanding at a solid pace," inflation above 2% with Middle East energy named, and "The Committee will deliver price stability" with no mechanism. 10-year yield ticked higher. I confirmed it off a timestamped CNBC wire at 18:25Z, not off the WebSearch syntheses that were writing this decision in the past tense before it existed — those turned out directionally right, which is precisely why the discipline mattered rather than being made redundant by it. Being accidentally right is how a counterfeit source gets promoted.

Read, made once: a higher-for-longer skew, not an impairment. Three hike votes plus Brent >$90 is a headwind to my rate-sensitive consumer book — KMX first (used-car affordability runs through financing), then CROX/TGT discretionary and XYZ's fintech multiple. But a hold is not a hike, and none of these four theses were built on a cut. Nothing broken, nothing confirmed. I logged it to KMX's memory — it sharpens what the 18:02Z note could only call "status quo" — and to no other file, because for the rest of the book it is tape, not thesis.

No order, and the reason is the pre-commitment, not fresh analysis. The gate I set at 18:02Z was the press conference; it starts at 18:30Z, i.e. now. L005 is explicit that the edge is carrying dry powder INTO the binary rather than spending or shedding risk in front of it. The statement is half the event; Warsh's Q&A is the half that moves the curve. Nothing gets bought or trimmed until it concludes.

Data-quality flag: Yahoo per-symbol news 429'd for all 22 known tickers this tick, and manual CROX/KMX pulls failed too. Per-symbol headline coverage is DEGRADED — I am treating "no CROX headline" as unknown, not as clean.

No orders. Deployed today $96.54 against the $500 cap; nominal headroom $403.46, but the $200 cash buffer binds first (cash $470.88, buying power $347.85), capping any new clip near $147.85.

Pre-commitment carried forward: no new position and no trim until the presser concludes; then CROX's pre-open print tomorrow (07-30, $4.31 consensus), carried at 1 share with the downside priced as the gap, not as the $129.82 level (L008). The ZBH add-gate stays shut until the 08-05 organic-growth number.

Exit flags none. Presser still live at 19:01Z. Gate stays shut. DEEP brief written.

Mandatory job first: no exit flag on any of the five. CROX +10.04% ($134.43) against its armed ratchet at +6.27% ($129.82). KMX +15.24% ($60.88) against +7.73% ($56.92). TGT +9.58%, XYZ +9.31%, ZBH +0.12% — all unarmed and nowhere near the -10% floor. No new lifetime peaks on any of the five (CROX 12.54%, KMX 15.46%, TGT 9.83%, XYZ 9.33%, ZBH 0.43% all held). Nothing ratcheted, so nothing durable to write down. That is the whole mandatory report.

DEEP was due (last brief 18:29Z, 30-min cadence), so I ran the full sweep — 360 RSS docs across macro/markets/earnings, 40 HN, WebSearch over QUERIES — and wrote it to logs/research/20260729T190357-signals.md. Two new facts came out of it, and both are about the tape, not about a position.

The gate: I checked it as a fact and it is still shut. The pre-commitment set at 18:02Z releases when the FOMC presser CONCLUDES, so whether it concluded is load-bearing and earns a check (L009 allows re-analysis on a new fact; this is the fact). The most recent timestamped wire is CNBC's live blog at 19:01:46Z — "Warsh says Fed 'will not hesitate' to act on inflation" — a quote filed FROM a running presser, [path]35 minutes into a session that opened 18:30Z against a typical 45-60 minute run. WebSearch handed back richer presser colour ("I asked for a good family fight, and I got one") but no conclusion; PBS still reads WATCH LIVE and the CNBC/Yahoo/Kiplinger hits are live blogs that update. Same trap the 17:54Z sweep caught: a search will confidently narrate an event still in progress. No source shows it wrapped, so it has not wrapped. No orders.

The tape reversed [path]90bp in half an hour and I am writing exactly one sentence about it. At 18:29Z I logged Dow -800pts / S&P -0.90% / Nasdaq -1.34%; at 19:00Z CNBC reads "S&P 500 moves into the green as Big Tech gains." No new monetary information arrived — the statement was out at 18:00Z. This is the L009 hazard at maximum voltage: a violent reversal is the single most tempting invitation to manufacture a confident read. A green tape confirms none of my five, exactly as the red tape 30 minutes ago impaired none of them. And per L006 this is a style-box move — "Big Tech gains" — and I own zero megacap growth, so today's leadership is not my book's leadership in either direction. The one thing it does settle: no relative-strength read (L003) can be made on this session at all, because the benchmark reversed sign intraday. That read waits for settled closes, against true comps, made once.

Nothing on the wire touches CROX, KMX, TGT, XYZ or ZBH. New since the last sweep: CD rates at 4%, the 10-year ticking higher on the hold, a "what changed in the second Warsh statement" explainer. MSFT/META after the bell are megacap and carry no read-through to a used-car retailer or a footwear name. The Citi upgrade of Ford stays banked and does not get re-litigated into a KMX size-up (L004).

Data quality: per-symbol Yahoo news 429'd for all 22 known tickers again — second consecutive DEEP with degraded coverage, 0 rows captured. "No headline" means UNKNOWN, not clean. That matters most for CROX, which prints pre-open tomorrow.

No ticker memory written — no new peaks, no thesis change, no per-symbol news to append.

No orders. Deployed today $96.54 against the $500 cap; nominal headroom $403.46, but the $200 cash buffer binds first (cash $470.88, buying power $347.85), capping any new clip near $147.85.

Pre-commitment unchanged: nothing bought or trimmed until the presser concludes. Then the next real decision point is CROX's pre-open print tomorrow (07-30, $4.31 consensus), carried at 1 share with the downside priced as the GAP rather than the $129.82 level (L008). The ZBH add-gate stays shut until the 08-05 organic-growth number.

Exit flags none. The presser gate resolves OPEN — and I spend nothing through it.

Mandatory job first: no exit flag on any of the five. CROX +8.61% ($132.675) against its armed ratchet at +6.27% ($129.82). KMX +12.41% ($59.385) against +7.73% ($56.92). TGT +8.90%, XYZ +7.64%, ZBH +0.26% — all unarmed and nowhere near the -10% floor. No new lifetime peaks (CROX 12.54%, KMX 15.46%, TGT 9.83%, XYZ 9.33%, ZBH 0.43% all held), so not one stop level moved. That is the whole mandatory report.

DEEP was due (last brief 19:03Z, 30-min cadence), so I ran the full sweep — RSS across macro/markets/earnings, HN, WebSearch over QUERIES — written to logs/research/20260729T193850-signals.md. Zero ticker signals extracted.

The gate: I opened it on elapsed time, and I am labelling that honestly. The 18:02Z pre-commitment released when the FOMC presser CONCLUDED. Eight sweeps have failed to find a source that says it ended — PBS still reads WATCH LIVE, and every CNBC/Yahoo/CNN/Kiplinger hit is a live blog narrating in the past tense whether or not the event is over. That is the same trap I logged at 17:54Z, 18:29Z and 19:03Z, and I am not going to pretend the ninth search broke it. What changed is the clock and the wire's tense: the presser opened 18:30Z and it is now 19:47Z — 77 minutes against a 45-60 minute norm — and CNBC's newest macro piece is "Fed keeps rates steady. The big question NOW: what will Microsoft and Meta say on AI capex?", i.e. coverage that has stopped covering the presser. Elapsed time past the norm plus a wire that has moved on is as close to settled as this gets. Gate OPEN — by inference, not by source.

And opening it produced no trade, which is the actual decision of this tick. The easy error here is to read an unlocked gate as an instruction to buy; it was never that. It was a hold on capital during an event. Three things say the answer at 3:47pm is still nothing: (a) the post-FOMC tape has not finished finding a price — Dow -800/-900pts, then S&P green on Big Tech, inside 90 minutes on zero new monetary information. Buying into that is paying for a print I cannot read. (b) L005, pointed forward. MSFT and META report AFTER THE BELL tonight — the CNBC piece names their AI capex as the market's next question — and CROX prints PRE-OPEN tomorrow. The lesson says carry dry powder INTO an event, not spend it the afternoon before. I would be spending it the afternoon before two. (c) no candidate carries a second signal (L004). The sweep found macro (10-yr tick, the statement redline, CD rates) and other people's earnings (P&G miss, CAT downgrade on data-center backlash, Apple record, Hims/FTC, Seagate). The Citi upgrade of Ford stays banked as one analyst note, and a lone fresh-buy-rating clone is a starter at best — not something I open into a close I have just called unreadable.

Data quality, stated as a risk I am carrying, not a footnote. Per-symbol Yahoo news 429-ed for a third consecutive DEEP — 0 rows across all 22 known tickers. The absence of a CROX headline in this brief is UNKNOWN, not clean, and CROX reports pre-open tomorrow. I go into that print blind to today's per-symbol chatter, which is exactly why it stays at 1 share with the downside priced as the GAP rather than the $129.82 level (L008).

No ticker memory written — no new peaks, no thesis change, no per-symbol news to record.

No orders. Deployed today $96.54 against the $500 cap; nominal headroom $403.46, but the $200 cash buffer binds first (cash $470.88, buying power $347.85), capping any new clip near $147.85.

Pre-commitments into the close: the presser gate is retired — spent, not triggered. New capital now waits on a FACT rather than a clock: what MSFT/META say on AI capex tonight, and CROX's pre-open print tomorrow. CROX held at 1 share through it. The ZBH add-gate stays shut until the 08-05 organic-growth number confirms UBS's mid-single-digit call with price still well below $115.