2026-09-20 · midday

Second run of a shut Sunday, and it found nothing new, which it verified rather than assumed. Every input was re-fetched: 35 position rows against the same 35 contract ids with zero field diffs, an account summary matching to the cent at $1,482,973.27 NLV and -$53,891.02 cash, no working order, all twelve price series re-pulled as regular-hours daily bars and every close identical to this morning's, and the Dec-28 340P implied vol identical to nineteen decimal places. Two rails fire, the same structural pair as the last eight checks: thirteen long puts above the $260 ceiling, and gross leverage 1.6814x against a 1.5x red. Neither is actionable before Monday's open. Runbook Parts 2 through 4 were swept row by row from this run's own marks and are clear on every price row, with tranche 3 at $360 still the nearest written level at 2.99% above GOOGL. The one genuinely new number came from outside the broker: the forecasting pipeline's last record is now 47.8 hours old, up from 41.8 this morning, and it is Friday's failed run.

What to do

13 long puts sit above the $260 long-leg ceiling: 280P x5 and 300P x8.
BUY TO CLOSE 5 GOOGL Dec15'28 360 PUT / SELL TO CLOSE 5 GOOGL Dec15'28 280 PUT
one combo x5, limit near mid, day · ~$17,750.45 debit ($35.5009 net per spread at Friday's closing marks)
Your playbook: “"long leg never above $260 (must sit at/above the $230-250 floor zone, never below $230 where protection stops paying before history bottoms)"”
The same standing order at an unchanged price. $35.5009 net per spread, identical to four decimal places across all four weekend reads, because the 360P and 280P marks have not moved since Friday's close. It addresses the 280P x5 leg only; the 8x Dec-28 300P has no written repair order anywhere, so this order leaves 8 of the 13 offending contracts in place. Not executable before Monday's open.

Still over, unchanged

These fire every run until you act on them. No order attached.

rail	now	limit
long_leg_band	280P x5, 300P x8	$230-$260 band	13 contracts outside
gross_leverage	1.6814x	1.5x red (1.3x amber)	0.1814x over red
top2_name_delta_ratio	0.62x	amber 0.70x, red 0.85x	clear
total_delta_ratio	0.86x	amber 1.30x, red 1.50x	clear

Why each one reads the way it does

long_leg_band
The rail the action item addresses. No leg has been traded on it.
gross_leverage
Flat to four decimal places for a fourth consecutive read. It rose on five consecutive reads before the weekend (1.6546, 1.6614, 1.6673, 1.6756, 1.6814) and has been pinned since, but a shut market is the entire reason - this is absence of evidence, not the drift stopping. Monday's open is the first read that can distinguish them.

delta by name

Delta notional per name as of 2026-09-20: stock at delta one, each option leg at the Black-Scholes delta implied by its own mark. Top two: GOOG, MU.

name	delta notional	of NLV
GOOG	$759,003	51.2%	1543 sh, 18 legs
MU	$155,689	10.5%	525 sh, 2 legs
META	$95,868	6.5%	144 sh, 0 legs
VOO	$86,319	5.8%	123 sh, 0 legs
RDDT	$77,386	5.2%	513 sh, 0 legs
NVDA	$50,866	3.4%	285 sh, 2 legs
AMZN	$25,371	1.7%	100 sh, 0 legs
EQT	$10,000	0.7%	200 sh, 0 legs
INTC	$8,560	0.6%	300 sh, 2 legs
TSLA	$364	0.0%	1 sh, 0 legs
DIS	$103	0.0%	1 sh, 0 legs

Everything else clear

Nearest trigger is GOOGL - tranche 3 confirmation (nearest written price level) at $360, 3% away.

All 29 triggers swept, with today's number

price trigger	now	fires at	distance
GOOGL - tranche 3 confirmation (nearest written price level)	$349.54	≥ $360	3.0% up
GOOGL - tranche 2 weakness	$349.54	≤ $317	9.3% down
GOOGL - thesis-break exit	$349.54	≤ $280	19.9% down
GOOGL - no-chase ceiling	$349.54	≥ $430	23.0% up
MU - roll the collar up (the only written MU level, and the only one above spot)	$1,015.80	≥ $1,250	23.1% up
GOOGL - Rule 6, close every short put	$349.54	≥ $434	24.2% up
GOOGL - washout, stop selling spreads and buy stock	$349.54	≤ $260	25.6% down
GOOGL - Rule 8, deploy the $100k crash reserve	$349.54	≤ $260	25.6% down
GOOGL - long spreads near max, apply the would-I-buy-it-fresh test	$349.54	≥ $500	43.0% up
GOOGL - shares capped by the 600Cs, roll the GOOG collar or accept the cap	$349.54	≥ $600	71.7% up

level trigger	now	fires at	distance
Every input re-fetched, and every one came back identical - the run found no new information in the book	0 of 48 figures changed	a second run of a shut market	The 35 position rows, the 12 price series, the account summary and the Dec-28 340P implied vol were all pulled fresh rather than carried forward, and all of them match the 06:49 run exactly: same 35 contract ids, zero diffs across position, mark, market value, average price and both P&L fields, NLV $1,482,973.27, gross $2,493,525.65, cash -$53,891.02, IV 0.3622516801180204. get_account_orders is empty. The last session was Friday 2026-09-18 and the next is Monday, so the identity is the correct answer rather than a stale read - it is reported because it was checked, not assumed.
Widest long call spread against the sell-it-always rule	44.00% of width (Dec-28 300/400)	90% of width	All five long call spreads recomputed from this run's marks: Dec-28 300/400 44.00%, Sep-27 300/440 41.78%, Sep-27 300/445 41.10%, Dec-28 350/500 31.35%, Dec-28 350/600 25.39%. The widest is less than half the trigger.
GOOGL Dec-28 30-delta put IV against the tranche-2 and crash alternates	36.23%	43% (tranche 2) / 50% (Rule 8 crash)	Taken from implied_vol annual_iv on the Dec-28 340P, 0.3622516801180204, identical to nineteen decimal places across all four weekend reads.
Dec-28 340P extrinsic, the assignment watch	$54.15, all of it extrinsic	< $5	Spot $349.54 is above the $340 strike, so the put is out of the money and its entire $54.152099609375 mark is extrinsic. The runbook notes this row only bites if GOOGL is under roughly $250 in 2028.
NLV one-week change against the 8% freeze	+3.99%	a fall of 8% in one week	Struck against the matching Sunday seven days back, 2026-09-13 at $1,426,052.75 from the rails log. The rule tests a FALL, so a rise cannot fire it either way.
Excess liquidity against both defence rows	75.83% of NLV	40% amber / 25% red	$1,124,514.56 of $1,482,973.27. Nearly double the amber.
Program defined max loss against its cap	$175,059	$200,000	$24,941 of headroom. Unchanged, because no leg traded.
Net cash against the IPS floor	-0.03634	-0.05 amber / -0.10 red	Cash is -$53,891.02 for a seventh consecutive read, identical to the cent. $20,257.64 of borrowing away from amber.
The 2026-09-17 wash-sale row is three days past due and its precondition still fails	DO NOT SELL	"IF selling: confirm zero GOOG/GOOGL acquisitions (shares, calls, DRIP) in the prior 30 days"	Re-derived this run from a fresh get_account_trades DAYS_30 pull rather than carried: 8x GOOGL Sep-27 300 CALL bought to open 2026-08-31T19:21:31Z across five exchanges at $75.40-$75.66, and 3x GOOGL Sep-27 445 CALL written 2026-09-14T16:25:18Z. Either is an acquisition inside the 30-day window on the narrow reading; the written call only counts on the broad one. Clean dates 2026-10-01 narrow, 2026-10-15 if a written call is also replacement property, which investing/tax-rules still does not say. The same 09-14 order also BOUGHT 3x Sep-27 305 PUT - a long put is not an option to acquire stock, so it moves neither date.

calendar	date	away
MU fiscal Q4 print, after the close - finviz carries it as Sep 30 AMC	2026-09-30	10d
Earliest clean sale date for the 6/05 GOOG loss lots if only bought calls count as replacement property	2026-10-01	11d
Earliest clean sale date if a WRITTEN call is also replacement property	2026-10-15	25d
GOOG print, first scheduled thesis adjudicator	2026-10-28	38d
Deploy the $100k reserve at market if no crash has come	2026-10-29	39d
GOOG capex guide, second scheduled adjudicator	2027-02-03	136d
GOOG lots vest long-term, add collar put leg 9 (+66 sh)	2027-03-13	174d
GOOG lots vest long-term, add collar put legs 10-11 (+200 sh)	2027-04-02	194d
GOOG lots vest long-term, legs 12-14 (+350 sh) - 193 of them clock-stopped by the 09-14 collar	2027-06-05	258d
GOOG lots vest long-term, leg 15 (+50 sh) - all 50 clock-stopped by the 09-14 collar	2027-06-25	278d

Spot prices

	last	day
GOOG	$344.41	+0.2%
GOOGL	$349.54	+0.6%
MU	$1,015.80	+3.9%
AMZN	$253.71	+1.0%
DIS	$102.67	-2.5%
EQT	$50.00	-0.7%
INTC	$108.60	-0.2%
META	$665.75	-2.4%
NVDA	$222.27	+1.3%
RDDT	$150.85	-1.3%
TSLA	$364.27	-0.5%
VOO	$701.78	+0.1%

The check itself

broken The forecasting pipeline's last record is now 47.8 hours old, and it is a failed run.
ops/status.py --check exits 11 on three reasons: 'latest daily record is from 2026-09-18, not 2026-09-20', 'latest daily record is 47.8h old, over the 26.0h threshold', and 'latest daily record status is failed' naming 27 steps. The staleness has grown from 41.8h at this morning's run and will reach roughly 71h before Monday's 12:45 slot can write anything. Two of the three reasons are pure calendar - the LaunchAgent's StartCalendarInterval is Weekday 1-5 only - but the third is not: Friday's run genuinely failed.
broken Friday's failure is 27 of 115 steps, and every one of them is a peripheral name.
house_forecast AMZN/GOOGL/META/MU/NVDA, street and scorecard and book_ev across BRK-B/DIS/EQT/RDDT/TSLA/VOO plus INTC, sync_positions BRK-B. Quoted from the status record, not re-derived. This is the same failure set this morning's run reported; nothing has run since to change it.
flag The gate examined 40 of 155 candidates on Friday, and 115 were skipped.
gate_candidates 155, gate_examined 40, gate_skipped 115, gate_losses 0, committed true at f05b886, duration 275.7s. The historical zero-count signature is ABSENT, so the gate did run against a real list. But 74% of candidates skipped is a large fraction and the record carries no field distinguishing an intended skip from a filter quietly matching nothing, which is exactly how both prior incidents presented. Flagged, not claimed wrong.
expected No record for today or yesterday, because the job does not run at weekends.
launchctl print gui/$UID/com.gabriel.stock-dashboard.daily reports StartCalendarInterval entries for Weekday 1 through 5 at 12:45. A Sunday with no record is the job behaving correctly, and the check still has no weekend-aware branch, so it reports the absence as a fault either way. Not proposed for change.
broken The Dec-28 340P's option-midpoint-iv returns a negative volatility, for a fourth consecutive run.
annualIv -15.874507866387544, dailyIv -1.0, isValid false. A volatility cannot be negative, so the field is not merely stale, it is wrong. The run fell back to implied_vol annual_iv 0.3622516801180204 is_valid true, which is the figure every IV row in this report uses. Nothing in any script checks the sign, so a future consumer picking option-midpoint-iv would read a -1587% volatility as data.
broken On a frozen book get_price_snapshot's `last` is Friday's after-hours print, not the close.
Fourth consecutive check, and confirmed from this run's own pull rather than carried: get_price_snapshot on GOOGL returns last $350.86 with is_close FALSE and top_status FROZEN, against the $349.54 regular-hours close, while its prior-close field returns the regular-hours $347.33. The snapshot therefore pairs an after-hours last with a regular-hours prior close and reports the day as +1.02% when the actual session was +0.64%. MU shows the same shape in this run's position marks, $1,017.79 against a $1,015.80 close, and on INTC the effect flips the sign of the day. Every price in this report is therefore a get_price_history ONE_DAY RTH bar, which is also the basis the runbook's daily-close rows are written against. No script enforces this; it is a per-run judgement each time.
broken deploy.sh still needs PATH=/opt/homebrew/bin prepended by hand.
Sixteenth consecutive run. The script does not set its own PATH and wrangler is not on the cron PATH.
open markets.dottl.xyz has no Cloudflare Access gate.
Fifteenth run reporting it. The page carries full position sizes, strikes and account NLV at a guessable URL. Not changed unilaterally.

Rails that could not be evaluated

MU has NO written price level anywhere below spot, and the fiscal Q4 print is now 10 days out. It closed $1,015.80, it is $155,689 of delta notional - second-largest in the book - and the runbook's only MU rows are the $1,250 collar roll-up 23.06% above and a Part 5 thesis break on HBM contract pricing that is a qualitative judgement rather than a level. The 900P floor is a position, not a rule: nothing written says what to do between $1,015.80 and $900. A bear target of $840.24 went on the tape yesterday and could not be acted on for the same reason.
The 8x GOOGL Dec-28 300P has no written repair order. It is 8 of the 13 contracts that fire long_leg_band every run, and the only order the playbook supplies addresses the 280P x5. Rolling them 300->260 at this run's marks would cost $10,898.64, but that number is this run's arithmetic and not a rule Gabriel has approved.
stressed_excess_liquidity has still never been measured. It needs Risk Navigator at -30% price and +50% IV with the TWS gateway up, and it is the one account rail with no number at all - the 75.83% unstressed figure says nothing about it.
No rail covers short calls against shares held. Re-derived from this run's rows: 45 short GOOG/GOOGL calls against 29 long calls is 16 net short contracts, 1,600 shares of obligation against 1,543 GOOG shares held, leaving 57 SHARES uncovered - just over half a contract. Small today, but nothing written tests it, so it would go unreported at any size.
The three Sep-2027 GOOGL legs and the whole INTC collar are invisible to every rail in rails.json. They are real risk that the check cannot see.
investing/tax-rules is silent on whether a WRITTEN call is replacement property, which is the only thing standing between a 2026-10-01 and a 2026-10-15 clean date for the 6/05 GOOG loss lots.
The SemiAnalysis H100 composite anchor is still the 2026-04 print at $2.82/GPU-hr, now five months old. No GPU pull ran this check by the midday rule, but the anchor's age is a standing gap and the whole above-contract reading of the daily marketplace numbers rests on it.
rails.json still lists short_dec28_340p_extrinsic as needing a live Dec-2028 chain pull when the position row's own mark answers it directly at $54.15. The rail reports unevaluated every run for an input the run already holds. Not edited - the file is only changed from an approved doc.
The markets prompt's section 4a fallback globs stocks-dashboard/cron/logs/, a path that does not exist in the repo. It has not been needed since ops/status.py landed, but the fallback would fail silently if it ever were.

Where the numbers came from

Run label decided from the rails log, not the clock: investing/checks/rails-log.jsonl already carried a 2026-09-20 markets-open entry at 06:49:22, so this is the MIDDAY run. Sections 4b (GPU) and 5 (forecast pass) were skipped by that rule, and gpu and forecasts are empty objects rather than values carried forward from the open run.
THE MARKET IS SHUT. Today is Sunday and this is the second consecutive shut day; the last session was Friday 2026-09-18, now two calendar days back, and the next is Monday 2026-09-21. Nothing in this report is a live price and no intraday figure exists for today.
EVERY INPUT WAS RE-FETCHED, NOT REUSED, and the identity was verified programmatically rather than eyeballed. get_account_positions: 35 rows, the same 35 contract ids, and a field-by-field diff across position, market_price, market_value, average_price, unrealized_pnl and daily_pnl returned ZERO differences against the 06:49 run. get_account_summary: NLV $1,482,973.27, gross $2,493,525.65, cash -$53,891.02, excess liquidity $1,124,514.56, identical to the cent. get_account_orders: empty. All 12 get_price_history series re-pulled: every close and prior close identical. The Dec-28 340P IV: identical to 19 decimal places.
PRICES ARE REGULAR-HOURS DAILY BARS, NOT THE SNAPSHOT `last`. On a frozen weekend book get_price_snapshot's `last` is Friday's final after-hours print and it differs from the close on 11 of 12 names, flipping INTC's sign. Every last/prior_close pair in quotes.json is a get_price_history ONE_DAY RTH bar, Friday 09-18 close over Thursday 09-17 close, which is the basis the runbook's daily-close rows are written against.
IV came from implied_vol annual_iv on the GOOGL Dec-28 340P, 0.3622516801180204 is_valid true, because option-midpoint-iv returned annualIv -15.874507866387544 isValid false for a fourth consecutive run. A negative volatility is not a stale reading, it is a wrong one, and it is reported under plumbing rather than quietly worked around.
RUNBOOK PARTS 2-4 WERE SWEPT ROW BY ROW from this run's own marks, not carried from the open run. The five long call spread percentages, the four GOOGL distance rows, the MU $1,250 row, the IV rows, the 340P extrinsic row, the excess liquidity rows and the NLV week row were each recomputed. All clear.
The NLV week comparison is struck against 2026-09-13, the matching Sunday seven days back, at $1,426,052.75 from the rails log, giving +3.99%. Part 3 tests for a FALL of 8%, so no verdict turns on the choice of baseline.
The wash-sale precondition was re-derived from a fresh get_account_trades DAYS_30 pull this run rather than carried forward. The last fill in the account is still 2026-09-16T15:18:16Z, the INTC 87.5/140 collar.
READ-ONLY throughout on ~/code/financials: nothing was edited, committed, reverted, rerun or cleaned, and the LaunchAgent was not reloaded. rails.json, build_markets_report.py and googl-mu-runbook.md were likewise not edited.