2026-11-02 17:00 ET: Settled after-hours: ON's −6% shrinks to −2.4% during its 5 PM call, PLTR/AAPL/VRTX drift +0.5% to +1.4%; the 16:49 post-close run replaces the stale buckets (LM = REGN only; 14 MR names, BKNG prints Tue AMC); no Tuesday-BMO print reaches a candidate at ρ ≥ 0.7; later AH snapshots beat 16:30 as a gap estimate

Slot: Mon 2026-11-02 17:00 ET hourly research. Status: DRAFT, written ~17:05–17:40 ET (EST). The whiggity consult was folded into §9 at ~17:15–17:25 ET; §3 (PLTR, AAPL), §4 (PFE), §6A and §10 were updated after verification. Screening/research, not investment advice.
TRAI data used: /api/latest = post-close run `2026-11-02_postclose_2026-11-02T21-49-57-818Z` (4:49 PM ET, `marketSessionDate` 2026-11-02), data-quality gate passed (0 stale symbols, VIX complex stamped 16:15:01 ET). This is the first fresh run today: earlier slots used the Friday-dated premarket run. Regime stress (VIX9D/VIX 1.062), riskPerTradePct 0.25. /api/events (generated 7:38 AM ET), /api/correlations (60d, asOf 10-30), /api/scorecard, /api/regime, /api/macro: all 200, read ~17:00 ET. Production is still on an older deploy because of the Vercel 100-builds/day limit; the API data above is live.
Quotes: Yahoo chart API 5-min bars incl. extended hours, read 17:05 ET (scripts/research-calcs/2026-11-02-settled-ah-snapshot.py); cross-checked for ON/PLTR/AAPL/VRTX/PFE against Public.com's after-hours pages. Unofficial, may be delayed. AH prints are thin. Futures last bar 16:58 ET (CME daily break 17:00–18:00). All replay statistics are my own calculations (scripts/research-calcs/2026-11-02-ah-timing-vs-gap-and-tue-links.py).
Previous slots today: 09:00, 10:00, 11:00, 12:00, 13:00, 14:00, 15:00, 16:00 (headline numbers of tonight's four prints, SHOP/PFE consensus and implied moves, Tuesday macro calendar). Those are not repeated; this slot covers what changed after 16:32 ET.

1. Focus and why

• The candidate list changed. The 16:49 post-close run is the first run dated 11-02. Every bucket call made today was on Friday-dated buckets. The new list decides which Tuesday catalysts matter.
• The AH moves settled differently from the 16:30 read. ON went from −5.9% (16:32) to −2.4% (17:05) during its 5 PM call. That tests S-131 (use the ≥ 16:30 AH move as the gap estimate) on the day it was proposed.
• Tuesday pre-market: map the BMO slate (SHOP, PFE, MPC, ETN in-universe; FISV, APO, KDP, SYY, ADM, YUM, APTV, LDOS outside it) onto the new candidates by ρ.

2. The 16:49 post-close run (/api/latest)

• Regime block: VIX 21.15, VIX9D 22.46, VIX3M 21.60, VVIX 110.47; VIX9D/VIX 1.062, VIX/VIX3M 0.979. SPY −0.48%, QQQ −1.36%, SMH −2.71%, all closing at 0.04–0.07 of the day's range. [V-P: /api/latest]
• The 16:00 report used Yahoo closes 21.42 / 22.91 (≈ 1.07). The engine's 16:15 snapshot is lower (1.062). Both are ≥ 1.05 with |SPY| < 0.5%, so S-129's trigger holds either way. VIX 21.15 < 22, so the coded S-092 halving does not fire.
• Own-print proximity (from `/api/events`): only BKNG prints within 2 sessions (Tue 11-03 AMC, sessionsToReaction 2). Next nearest: OXY Mon 11-09 AMC (6 sessions). LOW 11-18, WMT 11-19. All others have reported already (REGN 10-30, XOM/CVX 10-30, PYPL 10-27, DASH 10-29, the banks/AXP/GE/HON/ABT in October). [V-P: /api/events; issuer dates as verified in earlier reports]

3. Settled AH on tonight's AMC prints (Yahoo 5-min, 17:05 ET; Public.com cross-check)

Sources: ON GlobeNewswire release 16:10 ET 2026-11-02 (call at 5 PM ET), Seeking Alpha 4:28 PM, Benzinga 4:31 PM, Investing.com 4:20 PM; PLTR Palantir IR, 2026-11-02 [V-P], Investing.com 4:49 PM, Fortune 4:47 PM; AAPL Apple Newsroom, 2026-11-02 [V-P] (call 2:00 PM PT), CNBC live blog, updated 4:46 PM, AskTraders 21:41 UTC (Dec-qtr analyst est. $153.15B / EPS $2.91; these are analyst numbers); VRTX SEC ex-99.1 [V-P], Seeking Alpha (headline only). AH: Yahoo chart API; Public.com ON, PLTR, AAPL, VRTX (pages stamped "5:30 PM EST", read ~17:10 ET; the stamp looks ahead of the clock, so treat it as "latest").

• Data caveat: Yahoo shows only ~17k AH shares for ON's post bars, which is implausibly low for a −6%→−2% swing. The price agrees with Public.com, so I keep the price and ignore Yahoo's ON AH volume.
• Peers: ADI +0.57% (16:18, stale), TXN +0.77% (16:59), MCHP −0.17% (stale, 0 AH volume). SMH +0.39%, QQQ +0.33%, SPY +0.17% at 17:05. ES 7587.25 / NQ 28,227.5 at 16:58 (−0.35% / −1.07% vs Friday's close per Yahoo chartPreviousClose, so this includes Monday's session; it is not an AH move). The S-132 "ADI/MCHP/TXN ~−2%" estimate from 16:00 is now too big: with ON at −2.4%, the reporter gap is below S-132's 5% threshold, so the tag would not fire.

4. Tuesday 11-03 pre-market slate vs the new candidates

In-universe BMO names (SHOP, PFE, MPC, ETN) and Tuesday macro were covered at 16:00 §4; here only the link to the new list.

• Own-print risk inside the session: BKNG (MR fade-short, Tue AMC; implied ±7.5% per the 15:00 report). TRAI is flat by 15:55, so the print itself can't hit a BKNG intraday trade. The issue is trading a fade into a print day (S-111 "print-adjacent", half risk). S-127 found the pre-print-day move doesn't predict the reaction.
• AMC Tuesday: AMD, SMCI, ANET, BKNG, AMGN (from /api/events). AMGN–ABT ρ .45, AMGN–REGN .48. These matter for Wednesday, not Tuesday.

5. Read-through to Tuesday gaps (my calc)

Net: no Tuesday-BMO print reaches a candidate at ρ ≥ 0.7. The largest links are PFE→REGN (0.60) and MPC→CVX/OXY/XOM (0.50–0.54), and §6B shows the refiner link is weak in practice.

6. Tests (my calc)

6A. Which AH snapshot best predicts the AMC reaction gap? (extends 16:00 §6A and S-131)

Same 36 AMC prints in /api/events (09-10 → 10-29), Yahoo 5-min, AH price vs the regular close at each time:

• Later is better. The error falls by ~40% from 16:30 to 19:55, and the "×1.19" scaling in S-131 goes to ~1.0 (the last AH print is roughly the gap).
• In-sample, no ≥ 3% 16:30 move halved by 19:55 (0 of 20). ON tonight (−5.9% at 16:32 → −2.4% at 17:05) would be the first such case, if it holds to 19:55. One case is not evidence, but it shows the 16:30 read can be overtaken by the call, and most AMC calls start at 16:30–17:00.
• Caveats: n 36, one 8-week window, Yahoo AH data, overlapping symbols. Not split by call time.
• Look-ahead and sample check (after the consult, §9): Yahoo stamps 5-min bars with their start time, so each snapshot above used prices up to T+5 min. Re-run with only bars that had ended by T: r 0.86 (16:30) → 0.95 (19:55), essentially unchanged. The 36 prints are 36 different symbols but fall on only 10 report dates (18 on 10-28/10-29), so they are not independent. Resampling whole dates: r(16:30) 95% CI 0.71–0.93; the 19:55-minus-16:30 gain is +0.09 (95% CI +0.01 to +0.15). Later-is-better survives; the size of the gain is uncertain. Script: scripts/research-calcs/2026-11-02-ah-snapshot-lookahead-check.py.

6B. Tuesday BMO links in practice

• MPC/PSX → integrateds (8 reaction days in /api/events): XOM/CVX/OXY gaps ran −2.22% to +1.25%. On MPC's −8.52% (2025-11-04), XOM −0.33%, CVX −0.40%, OXY −1.71%, with SPY −1.06%. On MPC +0.59% (2026-08-04), the three gapped −1.6% to −2.2%. Sign agreement with the refiner is poor; the integrateds track crude and SPY more.
• FISV → PYPL (7 FISV gaps ≥ 5%, 2y): same sign 2/7; median |PYPL|/|FISV| 0.07. 60d close-close ρ 0.29. FISV's −43% (2025-10-29) moved PYPL −0.44%.

7. Analysis and calls

1. The fresh run cuts most of today's Tuesday worries. The 16:00 calls on CMCSA/TMUS/SPOT/AMZN/PSX day-2 ranking refer to names that are no longer LM candidates. PSX moved to the suspended SM bucket, so the MPC→PSX (ρ 0.91) gap-risk flag no longer touches an active candidate.
2. REGN (sole LM): the only overnight links are PFE (ρ .60, BMO, call 10:00) and VRTX (ρ .61, AH +1.2%). Both are below the 0.7 bar. Keep the 16:00 rule: no REGN entry before the 10:00 PFE call is under way, because S-128 fires on PFE (implied 2.9× TRAI's history) and PFE is already +1.6% AH for no reason I can find.
3. Energy fade-longs (OXY/XOM/CVX): this is a one-cluster bet on crude (S-114: cluster-penalized sizing; the engine caps at 1 per cluster). MPC's print is a weak driver for them (§6B). S-130 says don't chase oil-shock moves on day 2; the same logic warns against assuming a snap-back. OXY's own print is Mon 11-09 AMC (6 sessions), outside the S-111 window.
4. BKNG fade-short is print-adjacent (Tue AMC). Under S-111 it is half risk or skipped. It is also in the LOW/BKNG cluster; take LOW instead if a Consumer Disc. slot is wanted.
5. Financials fade-shorts (PYPL/COF/AXP/WFC, one cluster): FISV is the only Tuesday-BMO print near them, and history says it barely moves PYPL (§6B).
6. Sizing: S-129 holds on the engine's numbers (1.062, |SPY| 0.48%). Keep the by-hand 0.125% override from the 15:00/16:00 reports for Tuesday. The 16:00 §9 Election Day range check found the day itself hasn't been wide; that supports a range-scale, not a further cut.
7. AMC reporters for Wednesday's watch: ON at −2.4% and PLTR at +1.2% are under S-131's 3% threshold now, so neither gets an expected-gap estimate. The final number is the 19:55 print, not this one (§6A).

8. Strategy implications (proposals; the 6 PM session decides)

• S-133 (new, amends S-131): use the latest AH print (≥ 18:00 ET, ideally 19:55 ET) as the AMC reaction-gap estimate, with scale ≈ 1.0. Keep 16:30 only as a provisional flag. Evidence: §6A.
• S-134 (new, extends S-132 / S-122): for each Tuesday-type BMO print, only tag candidates at ρ ≥ 0.7. Record links of 0.5–0.7 on the dashboard as "watch", not "gap risk". Refiner→integrated and FISV→PYPL show the 0.3–0.55 links don't carry (§6B).
• Evidence for existing items: S-111 (BKNG is the live print-adjacent MR case), S-128 (PFE pre-print AH drift +1.6% with S-128 firing), S-129 (engine values 1.062 / 0.48%), S-132 (estimate dropped below threshold as ON recovered).

9. Research partner (whiggity) consult

I consulted whiggity in a new Monica chat after the draft (chat, reply ~17:13 ET; saved at /workspace/trai-consults/2026-11-02-1700-whiggity.md). It ran out of tool budget after four onsemi searches and answered none of the four questions (AAPL guide, PFE news and PLTR TCV/AH price were not searched; ON call content unknown). Below I keep only points I could check against a cited public source. [W] marks points that came from whiggity. Its citation numbers don't match its links: the URLs it listed under [1]–[4] include a Yahoo page for onsemi's Q4 2025 announcement and onsemi's Q3 2025 event page, which don't support the claims they are attached to. I re-sourced everything below myself.

Kept (verified):

• [W] onsemi call at 5:00 PM ET Mon 11-02: verified. The issuer release (GlobeNewswire, reposted by Nasdaq, published Oct 15, 2026 4:05 PM EDT) says results "after market close on Monday, November 2, 2026" and a call "at 5 p.m. Eastern Time (ET) on November 2, 2026" (Nasdaq repost). The 11-02 results release (§3) repeats the time. [V-P]

Rejected:

• [W] Investing.com headline "ON Semiconductor's results, guidance beat estimates but stock slips after hours": wrong quarter. I opened the article. It is dated 05-05-2026 and covers Q1 2026 (EPS $0.64 on $1.51B; Q2 guide $1.535–1.635B; "fell about 4% in premarket trading on Tuesday"); the current headline on the page says "dips premarket" (Investing.com India, 2026-05-05). It says nothing about tonight. The Q3 2026 sources are in §3.

Whiggity's handoffs (done here where quick):

• (1) Timings from issuer IR: all verified. BKNG releases at ~4:00 PM ET Tue 11-03, call 4:30 PM ET, so Tue AMC (Booking Holdings IR, 2026-10-06). PFE issues its performance report "that morning", call 10:00 AM ET Tue 11-03, so BMO (Pfizer, 2026-09-22); Pfizer gives no exact release time. AAPL did report today: FQ4 revenue $114.5B (+12%), GM 47.9% incl. ~1 pt from tariff refunds, EPS $2.10 incl. $0.05 from refunds, call 2:00 PM PT (Apple Newsroom, 2026-11-02). [V-P] I did not build the full Tue BMO list from IR pages; §4 has the in-universe names.
• (2) §6A look-ahead and n: checked (see §6A). The snapshots peeked up to 5 minutes ahead because of Yahoo's bar stamps. Fixing that changes nothing material. The real problem is clustering: n 36 on 10 report dates. The r level (0.85–0.95) is fine, with a date-clustered CI of 0.71–0.93 at 16:30. The gain from waiting is smaller and less certain than the point estimates suggest. S-133 stands as a proposal, with "+0.01 to +0.15 r" as the evidence. Separately, a 16:49 screen can't see the 19:55 print. S-133 needs a post-20:00 update step; it is not a fix to the 16:49 run.
• (3) Correlation clusters vs PFE→REGN 0.60 / MPC→CVX 0.54: whiggity's premise is wrong, and the answer is mixed. The code clusters at average ρ ≥ 0.5 (average linkage, CLUSTER_RHO = 0.5 in lib/research/correlations.ts), not 0.7. 0.7 is the S-132/S-134 peer-print tag bar, a separate rule. In /api/correlations (asOf 10-30), MPC and CVX are in the same cluster (#3, 8 energy names, avg ρ 0.707) even though their pair ρ is 0.54. PFE (cluster #4, 8 pharma names, avg ρ 0.625) and REGN (singleton) are in different clusters, even though their pair ρ is higher (0.60). That follows from average linkage, so the code is consistent with itself. It does mean the cluster id and the pairwise S-134 test can disagree. Also, the dashboard's "cluster Energy ×3" flags are sector counts (lib/engine.ts), not these correlation clusters. For the 6 PM session: say which one S-134 uses.
• (4) Should screens flag candidates whose post-close peer news is still unsourced at 17:00? Not decided here; that is yoo's design question. Tonight it would touch REGN (PFE's unexplained AH move, ρ 0.60) and nothing else at ρ ≥ 0.5.

My re-checks on the four original questions (~17:15–17:22 ET; not from whiggity):

• AAPL December-quarter guide: still not found. Apple gives it on the call, not in the release. CNBC's live blog (updated 5:14 PM) says Apple "will provide an outlook for the current quarter on a call" and had not yet posted it (CNBC). 24/7 Wall St.'s card says Apple "gave no explicit financial guidance", but it is stamped 4:30 PM, before the call, so it describes only the release (24/7 Wall St.). Search results quoting "10–12% growth, GM 47–48%" are Apple's October 2025 guide (Motley Fool transcript, 2025-10-31). Rejected for this quarter. Pre-call consensus: ~$153B, ~+6% (analyst numbers, §3). [UNV]
• PFE AH move: cause still unknown. No Pfizer release, filing or news after the close turned up. Search results only restate the price (≈ $31.05–31.15 vs $30.67 close). [UNV]
• ON call content: not found. Coverage found so far (SA 4:28 PM, Benzinga 4:31 PM, Investing.com 4:20 PM, Boursorama 22:54 CET) predates or ignores the call. [UNV]
• PLTR: TCV/RDV figures found in the release (§3: TCV $2.918B +6% y/y, US commercial RDV $6.517B +4% q/q) [V-P]. CNBC (updated ~17:00 ET) has only the release and a Karp interview, not call commentary (CNBC). Call commentary: [UNV].
• Latest AH (Yahoo 5-min, last bar 17:17 ET, vs §3 closes; unofficial, thin): ON 68.13 (−2.8%, giving back some of the call recovery), PLTR 178.92 (+0.3%), AAPL 338.55 (+0.1%), VRTX 558.00 (+0.9%), PFE 31.05 (+1.2%), BKNG 154.91 (+0.7%, 17:06). All are under S-131's 3% threshold. The 19:55 print is still the one to use (§6A).

Effect on the calls (§7): unchanged. Nothing from the consult changes a bucket, a size or an entry rule. REGN's "no entry before the 10:00 PFE call is under way" rule stays, because PFE's AH move is still unexplained.

Not from whiggity: §2–§8, the verification sources, the IR checks, the code reading in (3) and every [CALC] are mine. Only the points marked [W] came from whiggity.

10. Open questions

• ON/AAPL/PLTR/VRTX 19:55 ET last prints (the §6A-preferred estimate). Not available at writing.
• PFE's AH move: cause unknown (re-searched ~17:25 ET).
• Apple's December-quarter revenue and gross-margin guide (given on the call; not found by ~17:22 ET).
• ON and PLTR call commentary (ON's −5.9% → −2.4% → −2.8% path is unexplained).
• S-134: use pairwise ρ or the correlation-cluster id? They disagree for MPC/CVX and PFE/REGN (§9 (3)).
• yoo: flag candidates whose peer's post-close move is unsourced at 17:00? (§9 (4)).
• §6A: split by call start time (16:30 vs 17:00) and extend past 60 days with a different AH source.
• Tuesday's premarket run will refresh buckets on Tuesday's open; these candidates are from Monday's close.