2026-08-31 · midday

The book changed during the session. Four contracts of the GOOG Jan-2028 260 calls were closed and an eight-lot GOOGL Sep-2027 300/440 call spread was opened - the first structural change since the August restructure, and neither leg appears in the runbook's description of what the book is. Nothing new fired: the same four structural rails, the runbook sweep is clear, and the standing long-leg close reprices to $37.91. The two Alphabet tickers now disagree about which trigger is nearest.

What to do

280P x5 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 · ~$18,955 debit ($37.91 net per spread on this run's 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 as prior runs, repriced from the book's own marks. $37.91 against $37.77 at this morning's open and $36.53 on Friday's closing marks - the upward drift continued through the session and this is the dearest the close has been in the series. Closing the pair also cuts defined max loss, so it moves the program_cap rail the same direction rather than against it. The 8x Dec-28 300P also trip this rail and remain the separate matter; no order is proposed against them.

Still over, unchanged

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

rail	now	limit
program_cap	$175,059	$100,000 cap	1.8x, $75,059 over
gross_leverage	1.733x	1.50 red / 1.30 amber	red
net_cash / NLV	-0.071	-0.05 amber / -0.10 red	amber
uncapped_shorts	0 uncapped, 10 GOOG short / 10 long and 30 GOOGL short / 38 long	every short leg capped	clear, and measured this run

Why each one reads the way it does

program_cap
Unchanged from this morning to the dollar. The put program was not touched by today's trade; the two new legs are calls. Every leg carried a contract_description this run, so the number is complete.
gross_leverage
Up from 1.709x at the open. The move is arithmetic, not risk: the new Sep-27 spread replaces a single long call leg with two legs, and this rail sums gross position value, counting a short leg's absolute value as more exposure rather than less. The IPS row it proxies measures total DELTA notional / NLV, which is a different number and is not measured here. See plumbing.
net_cash / NLV
Cash -$99,051 on NLV $1,397,964, essentially the open-run balance. Signed at -14.8% and already red that day, so this is the improved end of a known accepted state. The written amber action is quoted verbatim in the findings below, because today's trade lands inside it.
uncapped_shorts
Measured on complete descriptions and passing. The 8 new short Sep-27 440 calls are covered by the 8 long Sep-27 300 calls; they sit outside the put program, so this rail is unaffected either way.

Everything else clear

Nearest trigger is tranche 2 weakness (the quote the script reads) at $317, 5% away.

All 21 triggers swept, with today's number

price trigger	now	fires at	distance
GOOG · tranche 2 weakness (the quote the script reads)	$335.10	≤ $317	5.4% down
GOOGL · tranche 3 confirm	$338.73	≥ $360	6.3% up
GOOGL · tranche 2 weakness	$338.73	≤ $317	6.4% down
GOOGL · thesis break	$338.73	≤ $280	17.3% down
GOOGL · Rule 8 crash powder	$338.73	≤ $260	23.2% down
GOOGL · Rule 6 close every short put	$338.73	≥ $434	28.1% up
MU · roll the collar	$945.43	≥ $1,250	32.2% up
GOOGL · long spreads near max	$338.73	≥ $500	47.6% up
GOOGL · shares capped by the 600Cs	$338.73	≥ $600	77.1% up

level trigger	now	fires at	distance
Dec-28 30-delta put IV	35.6%	> 43%	+7.4 pt
Dec-28 put IV · Part 3 crash trigger	35.6%	> 50%	+14.4 pt
GOOGL Dec-28 340P extrinsic	$56.76	< $5	+$51.76
Dec-28 300/400 call spread vs width	41.5%	>= 90%	+48.5 pt
Sep-27 300/440 call spread vs width (new today)	38.2%	>= 90%	+51.8 pt
Dec-28 350/500 call spread vs width	29.3%	>= 90%	+60.7 pt
Excess liquidity / NLV	74.7%	< 40%	+34.7 pt
NLV change over one week	-0.30%	-8% in one week	+7.7 pt

calendar	date	away
Wash-sale window opens for the 6/05 GOOG loss lots	2026-09-17	17d
Q3 print, the first scheduled adjudicator	2026-10-28	58d
Deploy the $100k reserve at market if no crash by then	2026-10-29	59d
First GOOG vest collar leg (leg 9, +66 sh)	2027-03-13	194d

Spot prices

	last	day
GOOG	$335.10	-2.3%
GOOGL	$338.73	-2.3%
AMZN	$259.16	-2.7%
BRK B	$504.29	-0.1%
DIS	$107.71	-0.4%
EQT	$54.03	-1.0%
INTC	$89.68	+0.2%
META	$572.40	-1.0%
MU	$945.43	+1.3%
NVDA	$219.66	+1.0%
RDDT	$147.35	-3.7%
TSLA	$367.54	+5.4%
VOO	$704.66	-0.4%

The check itself

found this run the book moved during the session and the runbook no longer describes it.
Two changes since the 09:14 read. The 4x GOOG Jan21'28 260 CALL position went to zero, and an 8x GOOGL Sep17'27 300/440 call spread was opened - 8 long 300 calls at an average of $75.56 against 8 short 440 calls at $22.19, both marked within a dollar of that and showing today as their first day of P&L. The runbook's WHAT THE BOOK IS section still lists '4x Jan-28 260C' and knows nothing of any Sep-2027 expiry. The runbook is a doc Gabriel approved and this run does not edit it; the divergence is the finding. The 90%-of-width row in Part 2 does apply to the new spread on its own terms, and it was swept: 38.2% of a $140 width, nowhere near.
found this run the trade lands inside the net_cash amber action, and the test that would judge it is the one rail that has never been measured.
net_cash / NLV is amber at -0.071. The IPS writes the amber action verbatim as 'no new leverage, no new delta in the top two names. Re-run the stress scenario within seven days.' Alphabet is the top name and a long call spread was opened in it today, while a long call position in the same name was closed. Whether net Alphabet delta rose or fell is exactly what top2_name_delta_ratio would answer, and that rail has been unmeasured since signing because summary.json carries no per-leg deltas. No verdict is offered here and none can be computed from this run's inputs. Flagging the collision only.
found this run the two Alphabet tickers now disagree about which trigger is nearest.
On GOOGL $338.73, tranche-3 confirmation at $360 is 6.28% above and tranche-2 weakness at $317 is 6.42% below, so t3 is nearest by 0.14 of a point. On GOOG $335.10, the quote rails.py actually reads, t2 is 5.40% below against t3 at 7.43% above, so t2 is nearest by more than two points. This morning both tickers agreed t2 was nearest; six hours later they do not. The gap between the two quotes is the persistent share-class spread, not a data error, but it is now large enough relative to the distance to the triggers that the answer to 'which line is closest' depends on which ticker is asked. Both readings appear in the watch table above rather than one being picked.
will happen again the gross_leverage rail has never measured the policy row it cites.
rails.json maps gross_leverage to gross_position_value / net_liquidation and tests it against thresholds the IPS wrote for total delta notional / NLV. rails.json carries its own note saying so: '1.57 at signing on gross; true delta leverage ~1.0x'. Today makes the mismatch visible: opening a defined-risk call spread pushed the rail from 1.709 to 1.733 because the short leg's absolute value is added rather than netted, even though a spread caps risk relative to the naked long leg it partly replaced. The rail's red has never been a statement about the policy metric.
will happen again rails.py reads GOOGL price levels off the GOOG quote.
Every spot rail in the googl_put_campaign program is written against GOOGL prices in the playbook and the runbook, but the script evaluates them on whichever underlying it resolves first, which is GOOG. On most days the share-class spread is too small to change any verdict. See the new finding above for the day it changed which trigger is reported as nearest.
will happen again the cron has no trading-day guard.
'35 6,12 * * *' fires seven days a week. Eight weekend runs so far in this series, none of which could carry new information by construction.

Rails that could not be evaluated

stressed_excess_liquidity - needs IBKR Risk Navigator (-30% price, +50% IV) with the TWS gateway up. Never yet measured, and the IPS amber action triggered today asks for exactly this within seven days.
top2_name_delta_ratio - needs per-leg option deltas, which summary.json does not carry. ~0.90 at signing and already red. This is the rail that would have judged today's trade.
short_dec28_340p_extrinsic - rails.json still classes this unmeasured pending a live Dec-2028 chain pull, but the position mark answers it: $56.76 of extrinsic against the <$5 watch, with the put $1.27 in the money at GOOGL $338.73.
The vast GPU series still does not reach 90 days, so the H100 tripwire's own test cannot run.
SemiAnalysis has published no composite since 2026-04, so the anchor is four months old. Tomorrow is the first 6:35 run of September and the next scheduled attempt to refresh it.
gpu-spot-daily.csv still has no row for 2026-08-29, because the 6:35 pull never fired that Saturday. Nothing was synthesized to fill it.
The GPU intraday health check still cannot see a capture that ran and collected nothing; two of today's captures were empty and printed no warning.

Where the numbers came from

get_account_positions, get_account_summary and get_price_snapshot all succeeded. All 13 underlyings returned top_status REALTIME with live last-trade stamps between 12:36:31 and 12:37:18 PT.
This run fired at 12:37 PT, close to its 12:35 cron slot.
positions.json carries 33 rows this run against 31 at the open. contract_description was present on all 33; no label-drop.
prior-close came back EMPTY on all 13 tickers again, the recurring pattern. Every prior_close is derived as last minus the broker's own change field.
All 12 stock priors were then independently re-checked against each position's own daily_pnl divided by share count, a second broker number that never touches the change field. All 12 of 12 match to the cent. GOOGL holds no shares and has no second check; its $346.59 is unchanged from the open run, where it was confirmed against the broker's own daily price history returning 346.59 as the 2026-08-28 close.
Dec-28 30-delta put IV read from GOOGL Dec15'28 340P (844342518): option-midpoint-iv annualIv 0.355807 with isValid TRUE, the identical midpoint value the open run read. implied_vol annual_iv 0.357881 agrees to within 0.21 of a point.
GOOG contract id 208813720, GOOGL 208813719.
Sections 4b (GPU spot) and 5 (forecast pass) are 6:35-only and correctly did not run; both ran at 09:16 today.