2026-09-09 · midday

Alphabet has clawed back part of this morning's fall and the one order on this page got cheaper with it. The long-leg repair - buy back the 5 short Dec-2028 360 puts, sell the 5 long 280 puts - now costs $19,570.59, against $19,885.61 at the open run seven hours ago. That $315 reduction is the same mechanism this morning's page described, running the other way: GOOGL rose $2.27 and the repair fell, at roughly $139 of debit per dollar of GOOGL measured across the two reads of this one session. Worth saying plainly, because this morning's framing of a repair that gets more expensive through a decline could be read as a reason to hurry. It is not. The price moves both ways and Gabriel picks the moment. Nothing else broke. The same three structural rails are over, every runbook trigger in Parts 2 through 4 is clear, and tranche-2 weakness at $317 has backed off to 3.4 percent below GOOG from 2.8 percent this morning. Nothing was traded: 32 rows, same contracts, cash unchanged to the cent for a sixth session. The one thing about the check itself: the midday cron slot fires at 12:35 Pacific, which is 25 minutes before the close, not the hour the written slot describes - a second scheduling discrepancy that survives fixing the timezone anchor everything was blamed on last week.

What to do

280P x5 and 300P x8 sit above the $260 long-leg ceiling.
BUY TO CLOSE 5 GOOGL Dec15'28 360 PUT / SELL TO CLOSE 5 GOOGL Dec15'28 280 PUT
one spread, limit, day · ~$19,570.59 debit ($39.1412 net per spread on live marks)
Your playbook: “long leg never above $260 (must sit at or above the $230-250 floor zone, never below $230)”
Ninth read of the same standing order. It printed $18,769.60 on six frozen-market runs, $18,793.92 on Tuesday's live chain, $19,885.61 this morning and $19,570.59 now. The $315 drop since this morning came entirely from GOOGL recovering $2.27, which puts a measured number on this morning's claim: about $139 of repair debit per dollar of GOOGL, established across two reads of the same session rather than asserted from the structure. The direction was already stated this morning; what is new is that it was witnessed running the other way within hours, which is the part that argues against reading the morning note as urgency. Only the 5x 280P leg is reachable by this order; the 8x 300P leg has no written repair. Reprice before sending, the chain is still moving and the close is 23 minutes away.

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
net_cash_ratio	-0.066	amber -0.05, red -0.10	amber
gross_leverage	1.764x	amber 1.30x, red 1.50x	red
top2_name_delta_ratio	0.66x	amber 0.70x, red 0.85x	clear
total_delta_ratio	0.94x	amber 1.30x, red 1.50x	clear

Why each one reads the way it does

long_leg_band
The rail the action item above addresses. No leg has been traded.
net_cash_ratio
Cash -$93,230.12 against NLV $1,411,505.20. The cash figure has not moved by a cent since 2026-09-04, a sixth consecutive session, so this ratio tracks NLV alone. NLV rose $8,136 since this morning and the ratio improved from -0.0664 to -0.0661 without anything being bought or sold.
gross_leverage
Gross $2,489,623.90 over NLV $1,411,505.20. Down from 1.773x this morning, which was the highest of the series, and again the move is arithmetic rather than positional: gross grew $839 while NLV grew $8,136. Carried caveat, unchanged: this divides GROSS position value by NLV against thresholds the IPS wrote for delta notional, and the delta rows read 0.66x and 0.94x, so the red reading overstates true leverage.

delta by name

Delta notional per name as of 2026-09-09: 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	$786,448	55.7%	1543 sh, 16 legs
MU	$143,408	10.2%	525 sh, 2 legs
META	$94,455	6.7%	144 sh, 0 legs
VOO	$86,262	6.1%	123 sh, 0 legs
RDDT	$75,401	5.3%	513 sh, 0 legs
NVDA	$50,252	3.6%	285 sh, 2 legs
INTC	$31,797	2.3%	300 sh, 0 legs
AMZN	$25,213	1.8%	100 sh, 0 legs
BRK B	$20,292	1.4%	40 sh, 0 legs
EQT	$10,878	0.8%	200 sh, 0 legs
TSLA	$369	0.0%	1 sh, 0 legs
DIS	$104	0.0%	1 sh, 0 legs

Everything else clear

Nearest trigger is tranche 2 weakness (nearest written level, and what rails.py reads) at $317, 3% away.
No position moved more than 5% today. Biggest was META +6.92%.
META +6.9%, now $655.94. No written rule for it.
Named, not acted on. META is on the no-written-rule list, so this check proposes nothing for it. It closed at $613.48 yesterday and has extended from +5.21% at the open run to +6.92% now, the book's largest single-name gain today.

All 27 triggers swept, with today's number

price trigger	now	fires at	distance
GOOG · tranche 2 weakness (nearest written level, and what rails.py reads)	$328.24	≤ $317	3.4% down
GOOGL · tranche 2 weakness	$330.64	≤ $317	4.1% down
GOOGL · tranche 3 confirmation	$330.64	≥ $360	8.9% up
GOOG · tranche 3 confirmation	$328.24	≥ $360	9.7% up
GOOG · thesis break, close regardless of price	$328.24	≤ $280	14.7% down
GOOG · washout, stop selling spreads and buy stock	$328.24	≤ $260	20.8% down
GOOGL · crash powder, deploy the $100k reserve	$330.64	≤ $260	21.4% down
MU · collar roll-up	$1,027.00	≥ $1,250	21.7% up
GOOGL · close every short put	$330.64	≥ $434	31.3% up
GOOGL · long spreads near max, apply the buy-it-fresh test	$330.64	≥ $500	51.2% up
GOOGL · shares capped by the 600Cs, roll the collar up	$330.64	≥ $600	81.5% up

level trigger	now	fires at	distance
Dec-28 30-delta put IV	35.88%	> 43%	+7.1 pt
GOOGL IV, crash trigger	35.88%	> 50%	+14.1 pt
11x Dec-28 300/400 call spread, mark vs width	40.07%	>= 90%	+49.9 pt
8x Sep-27 300/440 call spread, mark vs width	35.63%	>= 90%	+54.4 pt
10x Dec-28 350/500 call spread, mark vs width	27.53%	>= 90%	+62.5 pt
GOOGL Dec-28 340P extrinsic	$51.34	< $5	$46.34
Excess liquidity / NLV	74.64%	< 40%	+34.6 pt
NLV change on the week	+1.70%	-8% in one week	+9.7 pt
Defined max loss vs program cap	$175,059	$200,000	$24,941 headroom
Biggest single-name move today	META +6.92%	>= 5%	fired, no written rule

calendar	date	away
Wash-sale window opens for the 6/05 GOOG loss lots	2026-09-17	8d
Micron fiscal Q4 print (not a runbook trigger row; the five Seeking Alpha pieces filed this morning all name it)	2026-09-30	21d
Alphabet Q3 print · scheduled thesis adjudicator	2026-10-28	49d
Reserve deploys at market if no crash by this date	2026-10-29	50d
Earliest date the GPU 90-day tripwire test can evaluate at all	2026-11-16	68d
First GOOG lot vest, collar put leg 9	2027-03-13	185d

Spot prices

	last	day
GOOG	$328.24	-2.1%
GOOGL	$330.64	-2.3%
MU	$1,027.00	+2.7%
AMZN	$252.13	-1.9%
BRK B	$507.29	+0.3%
DIS	$104.36	-0.7%
EQT	$54.39	-1.3%
INTC	$105.99	+1.5%
META	$655.94	+6.9%
NVDA	$223.87	-0.8%
RDDT	$146.98	-1.6%
TSLA	$368.92	+0.2%
VOO	$701.32	-0.4%

The check itself

found this run the midday slot fires 25 minutes before the close, not the hour it is written to fire.
The written slot is '12:00 PM Pacific (one hour before the close)'. The cron entry is '35 6,12 * * *', which on a Pacific clock means 12:35 Pacific, and the market closes at 13:00 Pacific. This run's quotes are stamped 15:35 to 15:37 Eastern, 23 minutes before the bell. The discrepancy is 35 minutes and it is not the timezone problem. Last week's runs attributed every scheduling failure to the missing timezone anchor, and the machine returning to Pacific this morning was reported as that problem going quiet. This is a separate one that the anchor would not fix: the cron minute simply does not match the specification the prompt writes for it. It cost nothing today - a quote 23 minutes before the close is a live quote and every number here is a live number - but a slot defined as an hour of runway and delivering 25 minutes is a smaller window than intended for anything Gabriel would actually want to send before the bell. Not changed: the cron entry is written configuration.
found this run prior_close service rate is a function of the time of day, not of the feed's health.
This morning's run reported prior_close served on 1 of 13 equities and recorded it as the worst rate of the series. The same call against the same 13 contracts on the same broker seven hours later served 5 of 13 - GOOG, BRK B, DIS, RDDT and VOO. Nothing was fixed in between. The eight unserved names were derived as last minus the broker's change field, exactly as this morning, and every one of the thirteen - the five served and the eight derived - reproduces this morning's independently established 2026-09-08 closes to the cent. That cross-check is the point: the two runs arrived at the same thirteen numbers by different routes. The practical finding is that a low service rate is not evidence of a degraded feed and should not be reported as one, which is close to how this morning's entry read.
resolved no repeat of Tuesday's stale positions row.
The 09-08 midday run found the positions feed returning GOOG at Friday's close to eight decimals while the quote feed had it live, and noted the check has no consistency test between the two. That test was run by hand again this run. GOOG reads 328.19 in positions against 328.24 last and 328.2450 plprice; MU 1028.14 against 1027.00; META 654.83 against 655.94; INTC 106.095 against 105.99. All four are live and mutually consistent within a normal tick. Still done by hand, still not enforced anywhere in the check.
will happen again the open run's report was archived before this one overwrote it.
report-2026-09-09.json was copied to report-2026-09-09.open.json as the first write of this run, before anything else touched the directory. This is the ordering the 09-08 midday run got wrong and had to recover from a git commit. It is still enforced by nothing but the run remembering to do it first.

Rails that could not be evaluated

The midday cron slot is 35 minutes later than the prompt specifies, leaving 25 minutes to the close instead of an hour. Separate from the timezone anchor and not fixed by it.
The cron still has no timezone anchor and no trading-day guard. The machine is on Pacific today and the slots happen to mean roughly the right thing; nothing was fixed, and last week's six runs against one frozen session are what the failure looks like.
The check has no automated consistency test between the positions feed and the quote feed. Tuesday's stale-row artifact was caught by hand and this run's clean result was confirmed by hand.
Nothing enforces archiving the open run's report before the midday run overwrites it. Done correctly this run by remembering to.
The GPU intraday capture can write empty rows on schedule and the health check will call that healthy, because it tests arrival and not contents. Reported this morning; no GPU pull runs at midday, so nothing here re-tests it.
stockanalysis.com re-dates analyst rows and the ledger holds three rows for one Bank of America call and three for one Bernstein call. Reported this morning; no forecast pass runs at midday.
The GPU tripwire is defined at a 20 percent fall on an instrument that moves 20 percent or more on roughly half of all days, and cannot evaluate at all before 2026-11-16.
The SemiAnalysis composite is five months stale at 2026-04 $2.82 and the runbook's tripwire prose is defined on it, while later months are paywalled.
stressed_excess_liquidity - needs IBKR Risk Navigator (-30% price, +50% IV) with the TWS gateway up. Never yet measured, and the IPS amber's 7-day stress re-run has never run.
short_dec28_340p_extrinsic is still listed unmeasured in rails.json because it needs a live Dec-2028 chain pull. The runbook's own version of the row was swept by hand and is clear at $51.34 against the $5 trigger.
gross_leverage divides GROSS position value by NLV against thresholds the IPS wrote for delta notional, so the red reading overstates true leverage. Unchanged and unresolved.
rails.py reads GOOGL levels off the GOOG quote, which is why tranche-2 sits 3.42 percent away on GOOG and 4.13 percent away on GOOGL in this report.
rails.py keys option-leg matching on contract_description with no contract_id fallback.
The 11x Dec-28 300/400 line is 11 long 300C against 1 short 400C, not eleven matched spreads.
The MU collar roll cannot be executed as the runbook writes it: Part 2 says 'BUY back 1400C' but there is no 1400C - the collar is 5x Dec-28 1740C short against 5x 900P long. Only the trigger row and IPS Rule 7 line 242 are stale. First reported 08-18, and the trigger is 21.7 percent away.
The runbook's own 2026-09-04 note already says the MU roll-up as written is not premium-neutral: selling the 900P and buying the 1150P costs about $79k net on the 5-lot, while the roll down-and-out is a $41k credit. Re-price on the day.
The 8x GOOGL 300P long leg sits outside the $230-260 band and has no written repair order. Only the 5x 280P leg is reachable by the standing action item.

Where the numbers came from

Market OPEN, 23 minutes before the close, live stock and options book.
Run label decided from the rails log rather than the clock: an entry for today already reads markets-open, so this is the midday run and sections 4b (GPU) and 5 (forecasts) were both skipped as the prompt requires.
get_account_positions, get_account_summary and 14 get_price_snapshot calls all succeeded. No get_price_history call was needed: every prior close this run either came served or reproduced the open run's already-verified values exactly.
All 13 equities returned top_status REALTIME with last-trade stamps 1788982526-1788982655. The GOOGL Dec-28 340P returned top_status DELAYED with option-midpoint-iv valid at 0.358216 for a third consecutive run; implied_vol's 0.358772 was used for continuity and the two agree to 0.056 vol points, the closest agreement of the series and far below both the 43 percent and 50 percent IV rails.
prior_close served on 5 of 13 (GOOG, BRK B, DIS, RDDT, VOO). The other 8 were derived as last minus the broker's change field. All 13 match the open run's independently established 2026-09-08 closes to the cent. No price estimated.
The positions feed and the quote feed were checked against each other by hand. GOOG 328.19 in positions against 328.24 last and 328.2450 plprice, MU 1028.14 against 1027.00, META 654.83 against 655.94, INTC 106.095 against 105.99. All live, no repeat of Tuesday's artifact.
Book: 32 rows, the same 32 contract_ids as this morning, zero quantity changes, all 32 marks moved. Nothing was traded. Cash is unchanged to the cent since 2026-09-04.
rails.py fired 3, cleared 18 and left 2 unevaluated. The delta block is the script's own output, carried verbatim.
Runbook Parts 2, 3 and 4 were swept by hand against this run's numbers, every row clear. The Part 2 wash-sale calendar row opens in 8 days and is not yet a trigger.
Nothing in rails.json was edited. Where the book and the written rules disagree, that is reported in gaps rather than reconciled.