2026-11-02 15:00 ET: WTI −7% rotation (travel/retail up, energy down) into BKNG (Tue AMC) and MPC (Tue BMO, PSX's ρ 0.91 twin); VIX9D jumps 25% from its low with SPY flat, a real front-end inversion into the midterms

Slot: Mon 2026-11-02 15:00 ET hourly research. Status: DRAFT, written ~15:05–15:40 ET (EST). The whiggity consult was folded into §9 at ~15:30 ET, and §4 (Treasury) and §5 (MPC revenue) were updated after verification. Screening/research, not investment advice.
TRAI data used: /api/latest = premarket run 2026-10-30_premarket_2026-11-02T13-58-09-807Z (8:58 AM ET; marketSessionDate 2026-10-30; regime stress, VIX9D/VIX 1.015). Stale for intraday purposes: production is still pinned to an older deploy (Vercel's 100-builds/day limit), and no intraday run has been published since 8:58 AM. /api/events (asOf 2026-11-02), /api/correlations (60d, asOf 10-30), /api/regime, /api/macro, /api/scorecard: all 200, read ~15:05 ET.
Quotes: Yahoo chart API, 5-min and 15-min bars, last equity bar 15:07–15:10 ET; CBOE indices (^VIX, ^VIX9D) last bar 14:55 ET (Yahoo chart API). VWAP = my calc from 5-min typical price × volume; OR30 = 9:30–10:00. Unofficial, may be delayed. All replay statistics are my own calculations (script: scripts/research-calcs/2026-11-02-oil-crash-day2-and-vol-bid-flat-spot.py).
Previous slots today: 09:00 (oil −4% pre-open, PSX against signal, MAR raise → BKNG read-through), 10:00 (software), 11:00 (financials, REGN), 12:00 (semis), 13:00 (media/telecom, ETN), 14:00 (tonight's PLTR/AAPL/ON/VRTX, AI-hardware fade into AMD/SMCI/ANET). This hour covers the other side of the tape: the oil-driven rotation (which extended from −4% to −7% after the 09:00 report), the two non-tech Tuesday prints tied to it (BKNG, MPC), and the volatility bid building into the close. AI hardware and tonight's prints are not repeated.

1. Focus and why

• Long momentum (active): PSX (energy, +3.37% on 10-30) and AMZN (relVol 2.75, residualZ +1.23 on 10-30) are both on the list, and both are moved by today's macro story rather than their own news. PSX's 60d twin MPC (ρ 0.91) prints Tue 11-03 BMO (/api/correlations, /api/events).
• Not candidates but moving most: consumer discretionary (XLY +2.5%), airlines (JETS +4.3%), cruise, online travel (BKNG +4.3%, ABNB +4.9%, EXPE +3.3%) vs energy (XLE −3.4%). BKNG prints Tue 11-03 AMC.
• Regime: VIX +9% and VIX9D +17% vs Friday with SPY flat. The 8:58 AM stress label was a Monday/stale-input artifact (09:00 report, S-102/S-112); by 14:55 the front-end inversion is real (VIX9D/VIX ≈ 1.06) and VIX is ~0.8 pt below the 22 risk-halving line.

2. Tape at ~15:07 ET (vs Fri 10-30 close)

• Two different afternoons in one tape. The travel/airline names gapped up on oil and then went sideways below VWAP (gap ≈ day's move). The big-box retailers and home-improvement names (HD/LOW/WMT/COST) kept climbing all day and sit at highs above OR30: a rates/consumer bid, not a one-shot oil gap. The 10y is 4.90% (−3 bp on Yahoo ^TNX; MarketDash cites −5 bp to 4.89% at midday).
• Oil did its move before the open. WTI was 97.55 at 09:00 ET and has traded 95.49–97.55 since. Friday's settle 104.12 → −6.3% by 9 AM. The 09:00 report's −4.1% (07:50 ET) extended in the hour before the cash open. CNBC: WTI settled −7.3% at $96.46; Brent −7.4% to $105 at 2:48 PM (CNBC, 2026-11-02).
• Volatility rose all afternoon while SPY sagged only ~1%. VIX 18.10 (10:15) → 21.27 (14:45); VIX9D 18.01 → 22.67; SPY 763.64 (10:15 high-ish) → 755.5; QQQ 699.3 → 685.8 (15-min bars). The ratio went from ≈ 1.00 to ≈ 1.06.

3. Sourced drivers

• Oil: Iran's president said the supreme leader had granted the government "certain authorities" for peace talks; a draft proposal is circulating with Qatari mediators; Trump said Iran wants to "make a deal". Oil had priced escalation risk after the midterms (+10% in October). TD Securities' McKay: "so long as there is no concrete conclusion to the war, and tanker and Houthi attacks remain prevalent, we do not anticipate a complete erosion of the risk premium via diplomacy headlines alone" (CNBC, 2026-11-02; Reuters via Yahoo, 2026-11-02). Iran links reopening Hormuz to lifting the US naval blockade (Anadolu, 2026-11-02). Raymond James' Molchanov warns of "false dawns"; Goldman (Sunday) says Brent could reach $140 in 2027 if Gulf output stays 3.5 mb/d below 2025 (Reuters, above). The US ordered non-emergency staff out of Saudi Arabia over security risks (per tradevae midday, 2026-11-02, citing Reuters; primary not opened).
• Beneficiaries named in coverage: airlines (JetBlue +11%, United +6%, American +5% per 24/7 Wall St., 2026-11-02), cruise (CCL/RCL +5%, NCLH +4% per 24/7 Wall St., 2026-11-02), homebuilders and retailers on lower yields (MarketDash, 2026-11-02; figures in that piece are midday, e.g. S&P +0.6%, AMZN +4.7%, which are higher than my 15:07 snapshot).
• AMZN (LM): Phillip Securities upgraded to Buy, PT $330 from $320, citing AWS +39.3% (MarketDash, above; 24/7 Wall St., 2026-11-02). It faded from ~+4.7% at midday to +3.3% and is below VWAP: a day-2 rebound off Friday's −1.4% reaction, now giving back with the tech tape.
• Positioning: Morgan Stanley prime-brokerage data show hedge funds underweight AI for the first time since early 2023; recent buying was mostly short covering (Seeking Alpha, 2026-11-02; headline/snippet via search). That fits the rotation: money leaving AI hardware for consumer/cyclicals.
• Front-end vol: before the open Saxo noted one-day vol had jumped ~42% on Friday to sit level with VIX, the front VIX future above the second, "a hump for the midterms, payrolls and CPI"; options priced ~0.85% for today and ~1.84% into 11-06 (Saxo, 2026-11-02). CNBC/SpotGamma called Friday's VIX reaction "relatively muted" for a ~2% S&P drop (CNBC, 2026-11-02). Today's afternoon VIX9D rise is the catch-up. I found no single headline for the 12:15–14:45 vol bid; it coincides with the QQQ/AI-hardware fade (14:00 report) and the approach of the AAPL/PLTR prints and the vote. Cause: unverified.

4. Last hour (14:00–15:00) and 15:00 releases

• Treasury financing estimates (verified after the consult, ~15:25 ET): Treasury expects to borrow $618B in privately-held net marketable debt for Oct–Dec 2026 (end-Dec cash $850B), $10B less than the August estimate because the starting cash balance was higher; excluding that, needs are $24B higher than August. Jan–Mar 2027: $800B (end-Mar cash $1.000T). Jul–Sep actual: $742B vs $739B estimated, ending cash $984B. The refunding details come Wed 11-04 at 8:30 AM ET (Treasury press release sb0675, 2026-11-02; Reuters via StreetInsider, 2026-11-02 3:01 PM). Reuters says traders are watching for changes to auction sizes (possible cuts at 20y/30y) and buybacks. No visible rates reaction: TNX was little changed at 14:50 (4.90%).
• M&A: Marvell confirmed a ~$115B all-stock deal for Lumentum (4.5 MRVL per LITE); LITE −13.3%, MRVL −7.9% at midday (MarketDash, above). Not in the universe; AI-optics read-through to ANET (prints Tue AMC) is plausible but not measured.
• No universe-name 8-K, guidance change or pre-announcement found for 14:00–15:00 in my searches (not an exhaustive EDGAR scan).

5. Tuesday prints tied to this rotation

BKNG (Tue 11-03, results ~4:00 PM ET, call 4:30 PM ET). Online travel agency (Booking.com, Priceline, Agoda, KAYAK, OpenTable): merchant and agency commissions on room nights, plus flights, cars and ads.

• Date confirmed by the company: release ~4:00 PM ET, call 4:30 PM ET on 11-03 (Booking IR event page; PRNewswire via Yahoo, 2026-10-06). A Zacks article on Yahoo says 11-04; rejected vs the IR page.
• Consensus: EPS $4.48–4.49, revenue $9.54–9.56B (+~6%) (MarketBeat; Zacks via Yahoo, 2026-10-29). Zacks metric consensus: room nights 338.61M (vs 323M), gross bookings $52.73B (vs $49.70B), merchant revenue $6.81B (+11%), agency revenue $2.44B (−5.1%). Per-share figures are on the current share count (TRAI's history has Q3 2025 EPS at $3.98, so the series looks split-adjusted; split details not verified here). Tickeron's "+76.38%" EPS growth headline is inconsistent with its own $3.98 → $4.48 and is rejected.
• Company guide (as reported by Zacks/Tickeron): Q3 revenue +4% to +6%, +5% to +7% constant currency, FX ~−1 pt (Zacks via Yahoo; Tickeron). Not checked against the Q2 release.
• Street lean: BofA raised Q3 to 342M nights / $53.0B bookings / $9.61B revenue / $4.52 EPS, citing RevPAR acceleration in the US, Europe and APAC with September better than August (Investing.com; date not shown in snippet). Risks flagged: Middle East flight disruption hitting long-haul, FX, paid-marketing dependence as organic search traffic declines (Zacks, above). 8 of 8 EPS beats (MarketBeat). Today's MAR guide raise (09:00 report, S-118) is the nearest data point.
• Options: ±7.5% implied on 11-02 off a $147.89 reference (EarningsMoves; earnings-watcher); other tools show ±6.4–7.3% (search summary). TRAI last-4 avg |gap| 4.46% (max 6.93%, fill 50%, continuation 25%, avg day-1 in gap direction −1.62%) (/api/events). Ratio ~0.6, so S-128's "less than half" trigger does not fire, but options still price ~1.7× TRAI's history.
• The +4.3% today is macro, not company news (oil/airline/travel day). S-127 says a big pre-print-day move does not predict the reaction.
• Linkages (60d ρ): ABNB 0.71, UBER 0.63, DIS 0.59, LOW 0.55, HD 0.54; cluster 15 with ABNB. A BKNG gap Wed morning reads straight into ABNB (prints Thu AMC) and UBER (Wed BMO).

MPC (Tue 11-03 BMO, call 11:00 AM ET). Largest US refiner by volume; refining & marketing plus midstream via MPLX.

• Date confirmed by the company (Marathon Petroleum IR; call time per MarketBeat and search summary of the IR release).
• Consensus EPS $25.37–25.74 (TipRanks/MarketBeat $25.37, Investing.com $25.38, ChartRow/Nasdaq $25.74 on 5 estimates); whisper $26.79 (TipRanks; Investing.com, 2026-11-02 11:35 AM; ChartRow; Earnings Whispers via Finviz). Revenue consensus: $46.6–46.7B (TipRanks, MarketBeat, Investing.com). Resolved after the consult (§9): it is a quarterly figure on the same basis as MPC's reported "total revenues and other income" (Q2 $52.34B). Earnings Whispers' $33.56B looks like a stale estimate, not a different basis.
• What matters: Q2 margin $36.33/bbl vs $17.58 a year earlier; EPS estimates +29% over 60 days; management (via UBS): "even if both conflicts come to an end, it could take more than six months for refining margins to normalize"; Jefferies (Hold since late Sept) flags diesel demand destruction (Investing.com, above). Earnings Whispers: Q3 is seasonally the weakest capture quarter (~95% average 2023–25 vs 112% in Q2), and management said product margins and the jet-to-diesel spread had pulled back a month into the quarter (Finviz, above). Q2: EPS $17.73 vs ~$14.27 consensus, stock +1.8% next day (TipRanks). YTD +176.6% (TipRanks).
• Today's oil move is the forward-looking problem, not Q3. A diplomacy-driven crude drop doesn't by itself compress cracks, but it hits the "conflict premium" narrative that has driven the refiners. MPC −2.2%, VLO −3.2%, PSX −2.1% today.
• TRAI history: last-4 avg |gap| 3.16%, median 1.76%, max 8.52% (Q3 2025, −8.5% on an EPS miss), continuation 75%, avg day-1 in gap direction +1.05% (/api/events). No implied move found.
• PSX read-through: ρ 0.91. PSX is the active LM name, and MPC's print lands before Tuesday's open. PSX's own last-4 avg |gap| is only 0.99%, so TRAI's own-name gap-risk field will show PSX as low-risk while its twin reports. → S-122 / S-085 live case (§7).

6. Tests (my calc, Yahoo daily)

(A) Day 2 after a WTI ≤ −5% day (CL=F close-to-close, 5y, n 39 completed cases; continuous-contract roll artifacts not removed).

• Airlines and cruise: the move is a one-day gap, with no day-2 drift (next o→c ≈ 0, ~50% up). The OTAs (BKNG, EXPE) show a small positive day-2 (+0.4–0.6% o→c, 54–59% up): not significant at n 39, and the 2022 cluster dominates.
• Energy: no reliable day-2 bounce. XLE/PSX next-day c/c is negative 64% of the time (overnight gap continues), but open→close is flat. For PSX as an LM name, this is a further reason not to expect Friday's signal to reassert.
• Caveats: overlapping 2022 episodes; today's case has a scheduled two-sided catalyst (the vote Tuesday, a possible escalation after it).

(B) Front-end inversion on a flat SPY day: what happens next day? (^VIX9D/^VIX close ≥ 1.05 and |SPY c/c| < 0.5%, 10y.)

• All such days (n 68): next-day SPY gap mean +0.16% (56% up), o→c −0.07% (54% up): no direction. But mean |o→c| 1.10% vs 0.57% for all days (n 2,511), |gap| 0.62% vs 0.45%.
• Controlled for VIX level (VIX 19–24 and |SPY| < 0.5%): ratio ≥ 1.05 (n 27) → next |o→c| 0.95%, |gap| 0.58%; ratio < 1.0 (n 149) → 0.64% / 0.46%. So at the same VIX, an inverted front end came with ~1.5× the next-day intraday range.
• Nearest analog: 2024-11-04 (eve of the presidential election), ratio 1.172 → next day gap +0.16%, o→c +1.04%. The 2024-10-29/30 pre-election run (1.13–1.16) went the other way (o→c −0.22% / −1.20%). Direction is a coin flip; range is what scales.
• Caveat: Yahoo VIX9D closes; today's ratio is intraday at 14:55 and may change by 16:15.

→ S-129.

7. Analysis and calls

1. PSX (LM) is not working and carries hidden gap risk into Tuesday's open. −2.1%, below VWAP, on the day crude fell 7%. Its ρ-0.91 twin MPC prints before Tuesday's open with a TRAI max gap of 8.5%, and PSX's own events row shows no print. TRAI is flat by 15:55, so there is no overnight exposure today; the question is a Tuesday entry. Under S-116, an energy LM with crude down ≥ 3% against it needs post-open confirmation (above OR30 and VWAP); under S-122 / S-085 it should carry the peer-print tag. If PSX is still on the LM list tomorrow morning, don't enter before MPC's reaction has settled (after 10:00, above OR30 and VWAP).
2. AMZN (LM) has a sourced catalyst (upgrade) and is +3.3%, but it faded below VWAP into the afternoon AI fade. No print risk. Day-2 entry rules unchanged.
3. Travel/airline names gapped once and stalled; the history says not to chase on day 2 (§6A). If tonight's postclose run puts BKNG, ABNB, UAL or RCL in LM on relVol, that will be a macro-gap signal, not company news. BKNG's own print (Tue AMC, ±7.5% implied) lands after the 15:55 flat time, so a Tuesday intraday trade carries no print risk, but ABNB/UBER read-throughs land Wed.
4. Sizing tomorrow: at 14:55 VIX 21.21, VIX9D 22.54 (ratio ~1.06). If VIX closes > 22, tonight's run halves riskPerTradePct to 0.125% by code; if it closes 21–22 with the ratio > 1, the label is stress but risk stays 0.25% (the open S-092 inconsistency; the 10-30 15:00 report suggested overriding to 0.125% by hand). §6B says the expected next-day intraday range is ~1.5× normal regardless of direction. Midterm results arrive Tuesday night, after the close, so Tuesday's session holds the event premium and Wednesday's open holds the result.
5. Tuesday→Wednesday overnight stack: midterms, BKNG/AMGN/AMD/SMCI/ANET AMC, plus Netanyahu's reported push for a strike right after the vote (09:00 report). Oil's risk premium is two-sided into Wednesday.

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

• S-129 (PROPOSAL): when VIX9D/VIX ≥ 1.05 at the close on a day SPY moves < 0.5%, widen the next session's expected range (stops/ATR multiple) by ~1.5× or scale riskPerTradePct down by the same factor, regardless of the regime label. Evidence §6B (n 27 at the same VIX level). Feeds S-094's continuous risk scalar; adjust for weekday per S-112.
• S-130 (PROPOSAL): tag LM/MR candidates whose signal-day move is explained by a ≥ 5% crude move (airlines, cruise, OTAs, refiners, E&Ps) as macroDriven: oil; for these, the day-2 base rate is ~0 o→c (§6A, n 39), so require the normal post-open confirmation and don't size up. Complements S-116 (contrary overnight catalyst).
• S-122 live case: PSX (LM) with MPC (ρ 0.91) printing Tue BMO.

9. Research partner (whiggity) consult

I consulted whiggity in a new Monica chat after the draft (~15:15 ET). I treat it as a second researcher: below I keep only points I could check against a cited public source. The rest are listed as open questions. [W] marks points that came from whiggity.

• [W] Its tags: it re-tagged every number in my draft [UNV] because it couldn't check them itself. That is not a contradiction. The figures keep their own dated sources in §2–§5.
• [W] Implied VIX ≈ 22.5 / 1.06 ≈ 21.23 [CALC]: consistent. This matches the 14:55 Yahoo bar (VIX 21.21, VIX9D 22.54). It is arithmetic on my own inputs, not an independent source.
• [W] MPC EPS: is $25.37–25.74 a full-year number? Rejected; it is quarterly. Investing.com (2026-11-02 11:35 AM) gives Q3 EPS of $25.38, "a 43% jump in earnings from the prior quarter", on revenue of $46.72B, "an expected 11% sequential revenue decline". MPC's Q2 10-Q shows diluted EPS of $17.73 for the quarter and $19.30 for the six months (SEC 10-Q, Q2 2026). My check: $17.73 × 1.43 ≈ $25.35 [CALC]. A $25 quarter is large but consistent with Q2's $36.33/bbl margin.
• [W] MPC revenue: is the $46.6B vs $33.56B gap GAAP total revenue (incl. excise taxes/MPLX) vs net sales? Rejected as the explanation. In the Q2 10-Q, MPLX is consolidated (MPC is the VIE's primary beneficiary), so it is in every basis. The income statement shows sales and other operating revenues of $51.994B and total revenues and other income of $52.337B, a gap of only ~$0.34B (equity income plus other income). The filing's text mentions no excise tax in revenue; the only excise tax it mentions is on share buybacks. My check: $52.34B × 0.89 ≈ $46.6B [CALC], so the $46.6–46.7B consensus matches Investing's −11% sequential figure. $33.56B is close to MPC's Q1 2026 sales (~$34.2B = six-month $86.19B − Q2 $51.99B [CALC]), which suggests a stale, pre-spike estimate. That cause is my inference [UNV]. Use $46.6–46.7B.
• [W] NYMEX WTI settles 14:30 ET; equity figures at 15:00 are incomplete bars: consistent, with a caveat. CME's settlement procedure uses the VWAP of the front month from 14:28 to 14:30 ET (CME daily settlement procedure). CME blocked my download, so this comes from the search summary. CNBC's settle (−7.3%, $96.46) is quoted in §2. The CL=F figure in §2 (96.53) is a post-settle Yahoo quote, and all equity/VIX figures are intraday (§2 header). This doesn't change any conclusion.
• [W] Treasury refunding timing: verified on Treasury's site; [W]'s figures rejected as current. Treasury released the borrowing estimates on Mon 11-02 and will release the refunding details Wed 11-04 at 8:30 AM ET (sb0675, §4). The CME/Econoday snippet whiggity cited ($125B to refund $89.8B and raise $35.2B; CME/Econoday) is an older quarter. That page has no date and talks about bill sizes "through the end of September" and buyback changes "effective Aug. 13", so it is an August package. It is not this week's number. This week's 3-, 10- and 30-year sizes are unknown until Wednesday.
• [W] FOMC is not this week: verified. The Fed's calendar shows the 2026 meetings on Oct 27–28 and Dec 8–9 (federalreserve.gov). whiggity's calendarx link wasn't needed.
• [W] Midterm Election Day is Tue 2026-11-03: verified. 2 U.S.C. §7 sets "the Tuesday next after the 1st Monday in November, in every even numbered year" (LII).

Tests run on whiggity's questions (my calc, Yahoo; script `scripts/research-calcs/2026-11-02-oil-crash-day2-and-vol-bid-flat-spot.py`, section C/D):

• (a) [W] Split refiners from E&Ps after a WTI ≤ −5% day (same 39 events, 5y daily):

whiggity's premise holds directionally: refiners fall less on the day than E&Ps (crude is an input cost for them). But both groups fall, refiners lose about 2.8% on average, and neither has a day-2 bounce. Refiners are slightly weaker than E&Ps open→close. Results are pooled across names and events, so observations are not independent. No change to S-130. Refiners and E&Ps both get the macroDriven: oil tag. Keep the sub-group as a field so the 6 PM session can split them later.

• (b) [W] Re-test on TRAI's tradable window (proxy). Yahoo has 5-minute history for only ~60 days, and there was no WTI ≤ −5% day in that span, so I couldn't measure 10:00–15:55 ET exactly. As a proxy I used hourly bars from 10:30 to 15:30 ET (730-day history, n 17 events, 2025-04 → 2026-08). Next-day results, pooled: refiners +0.02% (median −0.05%, 49% up); E&Ps −0.30% (median −0.13%, 43% up); airlines (DAL/UAL/JETS) +0.78% (median +0.56%, 55% up); OTAs (BKNG/EXPE) +0.47% (median +0.51%, 59% up); SPY +0.06%. Energy has no tradable-window edge, which matches §6A. In this more recent sample the airlines and OTAs drift up on day 2, unlike §6A's 5-year daily o→c of ≈ 0. But the hit rates are only 55–59%, n is 17, and the April 2025 tariff-crash days are in the sample. Treat it as a hint, not a rule. S-130's "don't size up" stays.
• (c) [W] Is §6B's 1.5× next-day range driven by event days at t+1/t+2? Not run: I have no tagged event calendar for the 68/27 dates. Open. The nearest analog, 2024-11-04 (election eve), is itself an event case.
• (d) [W] Should the VIX > 22 sizing rule also look at VIX9D > 22 in event weeks? Not run. Open for the 6 PM session, together with S-092/S-129. VIX9D was above 22 at 14:55 (22.54) but had fallen back to 21.76 by 15:07, with VIX at 20.66.

Effect on the calls (§7): Unchanged. (1) The MPC check strengthens the PSX peer-print flag: consensus is a ~$25 quarter after a $17.73 Q2, so the bar is high, and TRAI's history includes an 8.5% MPC gap. Keep the rule: no PSX entry Tuesday until after 10:00 ET, above OR30 and VWAP. (4) Sizing: as of 15:07 both VIX (20.66) and VIX9D (21.76) were below 22. The by-hand override to 0.125% from the 10-30 15:00 report still applies if the ratio stays > 1 at the close. The new verified item is a risk for Wednesday: Treasury's refunding statement comes at 8:30 AM ET on 11-04, the morning after the midterms. Borrowing ran $24B above the August estimate before the cash-balance offset.

Not from whiggity: §2–§8 are my own sourcing and calculation. In §9, the verification sources, the test design and every [CALC] are mine; only the points marked [W] came from whiggity.

10. Open questions

• Does VIX close above 22 (risk halving) and does the ratio hold > 1 at 16:15 ET? At 15:07 ET: VIX 20.66, VIX9D 21.76, ratio ≈ 1.05 (Yahoo, unofficial). Both are below 22.
• ~~MPC revenue consensus conflict (§5).~~ Resolved in §9: $46.6–46.7B is quarterly, on the reported basis.
• ~~Treasury financing estimates: figures not verified (§4).~~ Verified in §4 (Treasury sb0675).
• Do oil-shock OTA day-2 gains (+0.4–0.6%) survive outside 2022? In the 2-year hourly sample (§9, n 17) OTA o→c is +0.96% (62% up), so the effect isn't only a 2022 artifact. Still small n with overlapping events.
• whiggity's test questions (c) and (d) in §9, for the 6 PM session (Risk Auditor / yoo).
• Why did Earnings Whispers publish $33.56B for MPC revenue? My stale-estimate explanation is unverified.