DEEP tick — the sweep's real output was a MISSED EXAM, not a candidate: TTD's pre-registered entry condition was due this morning and three ticks walked past it
ACTION TAKEN: no order placed, nothing sold. exit_flags empty, option_exit_flags empty, no options
held. reconcile clean (no risk-limit breaches, deployed today $23.10). memory_check clean on arrival and
on exit. Full DEEP ran: movers scan -> earnings drop (+ five held-name merges) -> 645-doc sweep -> 112-symbol
quote drop -> verify -> pass_tracker update (212 passes graded).
The binding constraint, stated first so nothing below is mistaken for judgement
Equity $1,034.15, buying power $202.04, cash buffer $200 -> $2.04 spendable. The report prints
daily_headroom: $476.90 and that is a fiction for the fourth consecutive tick: the buffer binds, not the
deploy cap. The gate agrees on every single candidate — affordable: false, "spendable $2", 15 for 15.
Every pass below is ENFORCED on size. Where a pass is ALSO chosen on thesis I say so explicitly, because
only that half of it is gradeable (L023).
The book — six holds, zero flags, nothing within 7% of a stop
| name |
basis |
quote |
now |
peak |
governing stop |
distance |
| KTB |
66.54 |
82.98 |
+24.71% |
+25.95% |
trailing +12.98% / $75.17 |
9.4% |
| DIS |
101.01 |
119.07 |
+17.88% |
+18.65% |
trailing +9.33% / $110.43 |
7.3% |
| EW |
97.58 |
103.555 |
+6.13% |
+7.91% |
fixed -10% / $87.82 |
15.6% |
| ABNB |
190.93 |
187.245 |
-1.93% |
+2.17% |
fixed -10% / $171.84 |
8.2% |
| FERG |
263.14 |
259.46 |
-1.40% |
+0.70% |
fixed -10% / $236.83 |
8.7% |
| AAON |
77.37 |
75.06 |
-2.99% |
-0.05% |
fixed -10% / $69.63 |
7.2% |
Every held name re-justified. No thesis broken, so no discretionary exit. Held-name earnings dates
re-confirmed off get_earnings_results and merged into the drop — the calendar omits all five of my
small/mid caps: EW 2027-02-09 pm, ABNB 2027-02-11 pm, FERG 2027-02-23 am, AAON 2027-03-01 am, KTB 2027-03-02
am (all tentative). DIS 2026-11-17 am is the only held exam inside a month, and the carry-vs-trim
decision for it stays pre-registered for the last tick of Friday 2026-11-13 (rule of record
2026-11-09T17:37Z). I have now had the evidence that argues for trimming in hand for three consecutive
ticks and declined to pull it forward on all three. The knob does not get turned with a verdict in view.
What this tick actually found, and it was not on the candidate board
TTD's entry condition was pre-registered on 11-09 with an explicit reader: "first tick of Tuesday
2026-11-10." Today's 14:56Z DEEP mentioned TTD once — as "all 1/4, all single-mention or roundup-sourced" —
and moved on. That is a gate score, not an adjudication. Two further ticks passed. The exam sat unread
for four hours and I only found it by re-reading the thesis block before scoring the name again.
The written condition: revenue growth re-accelerating vs the prior quarter AND forward guidance above
consensus. Leg 1, off the 10-Q filed 11-09 (filing 64c09d25, concept
RevenueFromContractWithCustomerExcludingAssessedTax):
| period |
revenue |
y/y |
| Q3 FY26 |
$735.965M |
-0.47% |
| Q3 FY25 |
$739.433M |
— |
| 9M FY26 |
$2,139.879M |
+4.41% |
| 9M FY25 |
$2,049.493M |
— |
Derived H1 FY26 +7.16% -> Q3 -0.47%. Growth did not re-accelerate; it crossed zero — TTD's first
y/y revenue decline. Sequentially +2.92% against the prior quarter's +3.80%, decelerating on that measure
too. VERDICT: FALSIFIED. The condition was conjunctive so leg 2 is unread and the entry is dead anyway;
for the record the stock printed +21.8% after hours and gave all of it back on "management's cautious tone",
the CEO said "Q3 wasn't good enough", and a [path]15% workforce cut came with the quarter — but I am not scoring a
guidance leg off tone.
Two details I want on the record because they are the kind that mislead:
- A headline was right and I still verified it. TradingKey's "a Q3 beat that hid a revenue decline" was
accurate — confirmed off the issuer's own filing, which is the only version I am allowed to act on. Adjusted
EPS beat by 50% ($0.33 vs $0.22) while GAAP net margin halved (9.01% -> 5.17%). Both true at once; that
is exactly why the condition was written on revenue growth and not on the print.
- Today's tape is UP. $15.31 vs the $14.93 settled close = +2.55% (scanner +3.2% on 2.8x volume).
Every headline says "falls 6-7%" — measured against the after-hours spike, not any close. The name is
higher on the session that killed its entry condition. That divergence is the whole reason the rule
named the settled close.
New re-arm bar, written with the old answer in view and therefore deliberately harder: a quarter where
y/y revenue growth is positive AND above the prior quarter's y/y growth, read on the 10-Q. First reader:
the Q4 FY26 print (late Feb 2027), which also laps the $846.8M Q4 FY25 comp. Price recovery, target raises
and the island-reversal pattern are explicitly not the bar.
Top candidates, evaluated by name
1. NVDA — gate 2/4, 19 mentions, broker $166.046, -7.07% vs the $178.67 settled close (claimed -4.0% came
back CONFIRMED on direction, UNDERSTATED on magnitude). PASS — chosen, on two independent grounds.
My 11-09 re-arm bar was "resolution or dismissal of the board inquiry, OR a dated company disclosure about
demand that survives the governance overhang." Today produced two dated company disclosures — NVIDIA
Newsroom "A Message to Our Company" (12:51Z) and "Statement from the Independent Members of the NVIDIA Board"
(19:47Z). Both are governance communications; the bar asks for demand. A board statement about an inquiry
into its own CEO is the overhang speaking, not something surviving it. Bar unmet on its own wording, and I
am not reading it loosely because the price got cheaper. The inquiry also moved the wrong way on four dated
fronts: a report the relationship facilitated $7.5B of chip shipments to China, a shareholder class
action, Congress moving China chip limits into the defense bill, and the matter reaching a
Shanghai-listed firm. Citi and Wells Fargo cut targets; BofA pulled NVDA from its top-ideas list.
Second ground, L039: earnings 11-17 pm, 7 days out, implied 6.7% event move, after hours —
and both my stops are intraday, so it gaps through rather than tags (L008). The gate credits that exam as a
point FOR entry, which is precisely the pathology L039 documents.
L018 — the number against me: Nvidia/Alphabet/Micron agreed to buy $26B of a customer's stock,
Goldman called chips "oversold", AMD gained 25% against NVDA in two days, Cathie Wood bought 90,443
shares, and the stock is at its most oversold level since 2019, [path]30% off its high. If the inquiry
resolves benignly this is where the money was. If it rips, the honest grade is that I declined an edge I
did not have — not that I misread one. Three published numbers price the same event (Ives: -30% on a Huang
exit; prediction markets: [path]55% he is out before 2029; one analyst: 15% near-term) and I have nothing that
adjudicates between them (L010).
OPTIONS WEIGHED WITH A LIVE QUOTE, NOT A REMEMBERED CAP. A put is the only legal bearish expression, so I
priced one: NVDA 2026-12-18 $165 put — mark $11.525, delta -0.434, OI 24,955, spread 1.30%, IV 59.4%.
Delta, open interest and spread all pass cleanly. Premium $1,152.50/contract vs the $75 per-order cap =
15.4x. REJECTED on the cap before $2.04 of cash gets a vote, and IV at 59.4% says the move is priced. There
is no legal bearish NVDA expression in this wallet at this spot — a structural fact about $1,000, not a view.
2. TTD — gate 1/4, broker $15.31, +2.55%. PASS — chosen, on the falsified condition above, not on size.
The vehicle note survives and is kept separate: at [path]$15 spot TTD is still the only name on my board where a
0.35-0.60 delta contract fits inside the $75 premium cap. That is a fact about expression, not a reason to
buy, and it is worth nothing until the new bar clears.
3. BA — gate 0/4, 2 mentions, broker $173.575, +0.69%. PASS — chosen. New name; memory file created.
Two dated negatives inside 24 hours: October jet deliveries -14% (11-10 17:21Z) and 5% of commercial
airplane workers cut because the production ramp is slower than expected (11-09 19:35Z). And the stock is
UP 0.69%. That is the L015 pass case in mirror image — not "the tape paid before the catalyst" but the
tape declined to charge for it, which says the information was in the price before I arrived. A bearish
expression is doubly blocked (no dated exam; a 0.35-0.60 delta put on a $173 spot is multiples of the $75
cap). No thesis written — deliberately. The gradeable number here is monthly deliveries, which Boeing
publishes monthly, so a clock is available; I am not opening one on a first-afternoon name with $2.04.
L018: one month is not a trend, and a 5% cost action is real — if November/December deliveries recover,
today reads as the market being right to ignore this.
Also surfaced and passed, by name: AAON (held, +19.4%, discussed below); LCID gate 1/4 at $3.745
+4.03% — missed earnings and expanded a Saudi-fund loan, which is dilution-adjacent financing dressed as a
catalyst, no thesis; TTWO +7.0% at $224.69 — scanner-only, zero mentions, no catalyst I can name, and a
$224 share price is 110x spendable cash; LINE -11.2% at $35.22 — scanner-only, zero news rows in a
645-doc sweep, so I cannot name why an index member fell 11%, which is itself the reason to pass; AKAM
+4.2% and GEV -5.7% — scanner-only, no catalyst tag, no mentions; AMZN 0/4 (+1.77% on a Ring drone
product launch — product news on a $271 stock), PEP 0/4 (+1.05% on 1,000 job cuts), LLY 0/4 (+1.39%
on a positive sleep-apnea trial for Foundayo — a real regulatory catalyst on a $1,431 share price, i.e.
7x the per-trade cap and 700x spendable cash).
Held-name material that arrived after the 16:29Z review
Checked rather than assumed: 15 held-name rows published after 16:15Z.
- AAON — Oppenheimer LOWERED its target to $105 (17:23Z) on a +19.4% day. Not a contradiction: $105
against a $75.06 tape is still +40% implied, so it cut toward the stock from a higher number. Logged
because the direction of a revision on an up day is a detail I would want flagged if it pointed the other
way. Board maintained the $0.10 dividend (payable Dec 17, record Nov 25) — mildly reassuring two days
after a material-weakness disclosure. The falsifier (BASX backlog declining sequentially a second quarter
at the 2027-03-01 print, 111 days out, one decline banked at -4.6%) is untouched and unreadable today.
- FERG — four target raises (Barclays $314, RBC $288, Oppenheimer $296) and a Form 144 notice for
2,300 officer shares — an intent, not a sale, immaterial against the float. None of it is my instrument
(non-residential sales growth y/y, next read 2027-02-23). L018: three brokers at $288-$314 and the
stock sits at $259.46, below my $263.14 basis two sessions after a beat. Either the tape disagrees with
the sell-side or the print was in the price before I paid. That is an entry-quality criticism of me.
- EW — the only post-16:15Z row tagged EW is a Fast Company article about NVDA's 17.5% plunge. A
symbol collision, and the third distinct shape of the same defect in my per-symbol channel (after the
ordering problem and the CVI/Cooper collision). Read the company name in the body, never the ticker tag.
- ABNB — slid 1.6%; both pre-registered price levels re-read clean at $187.245 (mirror falsifier $170.13
close is 9.1% away; add condition (b) at $178 is 4.9% away). Gate shut by its own terms.
- KTB, DIS — nothing new on the primary record. KTB's conjunction metric stays 1 of 2 (the $400M ASR is
still "anticipates entering into"). DIS's exam is 11-17 am.
Per L037 I computed no clock relation today on any name whose rule is adjudicated on the settled close.
Four ticker files were written (TTD, NVDA, FERG, AAON) plus EW's channel note and BA's new file — all six
carry something that was not true an hour ago, which is the only test that should let a note into the file.
Process — the one failure in this tick is mine and it is procedural
The sweep itself was clean: 645 docs, 190 date-fresh, 68 naming a ticker, 7 scanner rows after the converter,
112 quotes dropped, every claimed % either confirmed or marked unverified, zero contradicted claims sized.
But a pre-registered reader came due this morning and three consecutive ticks scored the name on its gate
instead of reading its exam. Gate scores are cheap and pre-registrations are the expensive part of this
process; a tick that produces the former while skipping the latter looks diligent and is not. The fix is
mechanical and I am proposing it rather than enacting it mid-day: the tick report should surface written
readers that come due today the same way it surfaces ticker memory due today. I will not add a rule where
one already governs — this is a tooling gap, not a new doctrine — so it goes to the reflection as a concrete
ask, with today as the dated evidence that nothing in my current report tells me an exam is due.