The Nightcap White-Cap

Built ~4:55 PM ET · 2026-07-02 · validates the 7/2 regular-session close (Massive /v3/snapshot, COALESCE(session_close,session_price)); June NFP confirmed BLS/TradingEconomics anchor: system-reminder 2026-07-02 == Bash date (agree) · static — regenerate to refresh

01Session Scorecard — 2026-07-02

7 setups in the open backlog, all 7 finalized this run — 4 on today’s 7/2 session (Curd + Midday) plus 3 weekly Sunday Sundae setups (6/28) whose window closed at the pre-holiday 7/2 session (Fri 7/3 Independence Day closed). Tally: 4 FIRE / 2 MIXED / 1 VOID / 0 NO_EVIDENCE; backlog fully cleared. The day was a 2nd-day semis / AI-hardware unwind (SMH −4.55%, KLAC −11.5%, LRCX −10.2%, MU −5.49%) on a soft June NFP (+57K vs 115K cons.) that funded a defensive rotation — healthcare/utilities/staples + gold + AAPL led, RSP +0.70% > SPY −0.13% > QQQ −1.74%, 8 of 11 SPDRs green. The semis short (SRT) and defensive long (MS-XLV) fired; the QQQ level-short fired on an inverted catalyst; the IWM rotation-long went MIXED (equal-weight led, but IWM lost 300).

Setup	Outcome	Evidence	Δ-ATR	Actual
SRT S SMH,QQQ,XLKmd-260702-1209-SRT-SEMI-S	FIRE	Sector-rotation-top semis short, conditional on a failed QQQ VWAP reclaim. QQQ held below 722.39 VWAP into the close (712.6, no reclaim) and semis unwound a 2nd day — SMH −4.55%, MU −5.49%, AMAT −7.36%, LRCX −10.2%, KLAC −11.5%. SMH never reclaimed 610; no broad reclaim (SPY 744.78 < 747.9). No kill. QQQ-denominated (primary inst SMH), so Δ-ATR / realized-R n/a by construction; target 719.02 exceeded.	—	QQQ 712.60
MS L XLV,XLP,XLUmd-260702-1209-MS-XLV-L	FIRE	Defensive / quality relative-strength long. XLV held its 162.47 VWAP and closed 163.74 (+2.63%) at session highs; defensive leaders led — XLP +2.03%, XLU +2.21%, gold GLD +2.03% / GDX +4.48%, AAPL +4.84% — 8/11 SPDRs green, RSP +0.70% > SPY −0.13%. No broad-reclaim kill (SPY<747.9, QQQ<722.4). Clean RS long.	0.48	XLV 163.74
LRT S QQQ,SPYmm-260702-LRT-QQQ-S	FIRE	Level-rejection short at the 729.5–731.9 shelf. QQQ tagged 730.83 (within the shelf) and rejected hard, closing 712.6 (−1.74%) toward the 680 target — no clean 729.5 reclaim, semis didn’t stabilize (SMH −4.55%). Mechanism leak: fired on the semis unwind + defensive rotation, not* the modeled hot-NFP yield spike — NFP was soft (+57K) and the 10Y actually firmed 4.44→4.46.* Entry ref 731.92 not precisely tagged (high 730.83). Realized +0.75R.	1.12	QQQ 712.60
MS L IWM,RSP,XLFmm-260702-MS-IWM-L	MIXED	Broadening-rotation momentum long, armed on IWM holding its 300 reclaim. The rotation thesis fired — equal-weight led (RSP +0.70%), financials (XLF +1.53%) and materials (XLB +1.94%) green vs QQQ −1.74% — but IWM lost 300 (kill triggered), closing 297.58 (−0.58%) below the 300 entry. Soft NFP never armed the hot-NFP kill; semis didn’t rebid. Macro right, the headline IWM-300 long voided — MIXED. Realized −0.31R.	0.47	IWM 297.58

Finalized — weekly Sunday Sundae (6/28; window closed at the pre-holiday 7/2 session, Fri 7/3 closed)

Setup	Outcome	Evidence	Δ-ATR	Actual
UCR L SPYsun-260628-UCR-SPY-L	FIRE	Undercut-and-reclaim long, entry ref 737. SPY held >716.58 all week (no undercut close; wk min close 741.0) and stayed above the 737 reclaim every session (Mon 741.0 · Tue 746.77 · Wed 745.76 · Thu 744.78), closing 744.78 (+1.05% above entry). Kill (close<716.58) never triggered — direction confirmed. Notional +0.43R (774 target not reached).	—	SPY 744.78
OMR L QQQsun-260628-OMR-QQQ-L	MIXED	Oversold mean-reversion long, entry ref 716. QQQ reclaimed 716 Mon–Wed (peaked 737.62 Tue, ~+3% from entry) then faded on the soft-NFP semis unwind, closing the week 712.6 — back below the reclaim. Kill QQQ<702 never hit (wk low 705.17); the “ugly jobs” leg armed (+57K miss + downward revisions). Target 774 never approached — armed and ran, then reversed: MIXED. Notional −0.12R.	—	QQQ 712.60
EXF S XLVsun-260628-EXF-XLV-S	VOID	Exhaustion-fade short of stretched defensives (WATCH, no exec). Kill triggered: soft jobs kept defensives bid — XLV closed 163.74 at a fresh weekly high (+2.63%), XLP +2.03% / XLU +2.21% led into the close. No exhaustion fade materialized — the short voided.	—	XLV 163.74

LensA clean dispersion day, defensively tilted: the semis short and defensive long both fired, and the QQQ level-rejection short fired — though on a rotation / soft-NFP catalyst, not the modeled hot-NFP yield spike. The two blemishes are instructive: the IWM rotation-long MIXED (right that breadth would broaden, wrong that small-caps would hold 300 — the bid went defensive, not cyclical), and the weekly QQQ mean-reversion MIXED (reclaimed, then faded). The weekly SPY long fired and the XLV short voided on the soft-jobs defensive bid — the Sundae’s defensive-fade thesis was exactly backwards.

02Calibration

0.2556 Rolling Brier (last 50) up from 0.2477 (7/1) · 0.25 = random

46% Hit rate (rolling 50) 42.5% over all 87 eligible

87 Calibration-eligible 7 finalized this run (4 setup-track + 3 sundae excl. from Brier) · 0 NO_EVIDENCE

Rolling Brier ticked up (worse) 0.2477 → 0.2556 — a paradox worth naming: today’s four setup-track legs hit 3 of 4 (SRT, MS-XLV, LRT fired; MS-IWM MIXED), a strong directional day — but every FIRE was under-confident: LRT fired at prob 0.40 (Brier 0.36), SRT at 0.45 (0.30), MS-XLV at 0.50 (0.25). Three correct calls the model hedged low, which mechanically raised the batch Brier even as the hit-rate rose. No data-feed gaps: the pivotal June NFP (+57K vs 115K) scored from confirmed BLS / TradingEconomics, never a newsletter; no leg was gated on $S5FI/$S5TH/AAII (kills referenced IWM/RSP/SPY/XLV VWAPs + the confirmed NFP print), so breadth used confirmed price action, never inference.

Pattern performance — Component 13 (calibration-eligible; n≥3)

Pattern · Dir	n	Hit%	Mean prob	Mean Brier	Evidence class
LRT · S	26	54%	0.37	0.324	lore_pending
MS · L	24	38%	0.51	0.255	lore_pending
SRB · L	7	71%	0.41	0.299	lore_pending
SBD · L	6	0%	0.37	0.138	primary_source
GFD · L	5	60%	0.39	0.257	practitioner_backtest
MS · S	4	75%	0.44	0.318	lore_pending
SBD · S	4	0%	0.28	0.079	primary_source
LRB · L	3	33%	0.43	0.216	lore_pending
SRT · S	3	67%	0.52	0.384	lore_pending

Both workhorses firmed tonight: LRT-S 52% → 54% over 26 (today’s shelf-reject short fired) and MS-L 36% → 38% over 24 (MS-XLV fired, MS-IWM MIXED). SRT-S is now n=3 at 67% after tonight’s FIRE — a small but clean rotation-short edge building. SRB-L (71% / 7) remains the cleanest high-n edge; the two 0%-hit SBD cells (primary-source sentiment/breadth) stay the standing red flag. Singletons (FBR-S, GFU-S, VAB-L, VBR-L, VSR-L) too thin to read.

Calibration trend — Component 14 (rolling Brier by validation date)

06-240.261

06-250.2435

06-260.2411

06-290.2399

06-300.2418

07-010.2477

07-020.2556

Mechanism leak · C15 11/40 right-direction / wrong-mechanism (rate 0.275, down from 0.306 as the denominator grew and none of tonight’s four legs added a count), still led by LRT-S (6). Beyond the canonical count: tonight’s LRT-QQQ actually FIRED (right direction, right outcome) but on an inverted catalyst — it fell on the semis unwind + defensive rotation, not the modeled hot-NFP yield spike (NFP came soft). A narrative mechanism leak the close-vs-level metric can’t see, and the recurring LRT signature.

LensThe calibration lesson tonight is under-confidence on level / rotation-short work: LRT fired at 0.40, SRT at 0.45, MS-XLV at 0.50 — three correct calls priced too timidly, so a 3-of-4 night still nudged Brier up. The mirror risk remains the continuation long into event risk (the 0.52 IWM momentum-long lost its 300 level and went MIXED). And the LRT pattern keeps landing right-direction on the wrong mechanism — tonight it even fired, but for reasons its thesis didn’t model.

02bMacro-Prediction Calibration · thinktank-v2 predictions, read-only

Macro-prediction calibration unavailable (thinktank-v2 absent — resolve_predictions returned “unavailable (thinktank-v2 absent)”). This is a separate calibration lens from the setup track-record above — an external macro / cross-asset forecast log, never combined with the setup Brier/hit-rate — and its absence does not affect the validation core. Nothing scored or written (DB read-only, not mutated).

03Tape & Rate Backdrop

Index	Close	Day %	Read
S&P 500 (SPY)	744.78	−0.13%	~flat — defensive rotation offset the tech drag
Nasdaq 100 (QQQ)	712.60	−1.74%	semis-heavy — the day’s laggard
Equal-weight (RSP)	214.91	+0.70%	> SPY > QQQ — broadening
Russell 2000 (IWM)	297.58	−0.58%	lost 300 — small-caps didn’t hold the reclaim
Dow (DIA)	527.88	+0.99%	value / defensive ballast — led the majors

Dominant driver = a 2nd-day semis / AI-hardware positioning unwind on a soft June NFP, not a broad macro shock. The QTD-leader semis de-rated again (SMH −4.55%, KLAC −11.5%, LRCX −10.2%, AMAT −7.36%, MU −5.49%) and the proceeds rotated defensive — healthcare/utilities/staples + gold + AAPL. SPY barely moved; QQQ/XLK took the hit.
Econ (confirmed, BLS / TradingEconomics): June NFP +57K vs ~115K consensus — a large downside miss; unemployment 4.2% (vs 4.3% exp); April+May revised −74K combined. A clearly cooling labor market.
Rates: 10Y ~4.46% (est., midday; morning DGS10 4.44) — yields firmed slightly despite the soft NFP (TLT flat −0.01% confirms), the hawkish-Fed / firm-dollar backdrop offsetting the jobs miss rather than a growth-scare rate-cut bid. VIX ~16.7 (est., midday 16.72e; prev close 16.59) refresh-required — a muted lift; the damage was concentrated in semis, so index vol stayed low.
Realized regime = rotation / dispersion, defensive tilt (broadening breadth, equal-weight + defensives leading) — matching the Midday Frappe’s “rotation_dispersion / defensive tilt” read. Stated as what happened, not a forward call.

LensThe index-level calm (SPY −0.13%, DIA +0.99%) again masks a violent under-the-hood move — the most-crowded complex de-rated 5–12% for a 2nd session while the tape broadened. The tell: on a soft-NFP day the “bad news = rate cuts” bid did not flow to growth/tech — it went to defensives and gold, and yields firmed anyway. A rotation, not a liquidation (credit calm, crude green).

04Cross-Asset

Asset	Proxy	Close	Day %	Read vs equities
US Dollar	UUP	28.34	−0.53%	eased on the soft NFP — mild tailwind for risk
Crude	USO	103.98	+0.69%	firm — XLE green, no growth-scare in oil
Gold	GLD	378.13	+2.03%	bid hard — rate-relief / safety; GDX +4.48%
Long bonds	TLT	85.51	−0.01%	flat — yields ~unchanged/firmer, no flight-to-safety
Credit (HY)	HYG	79.71	+0.15%	firm — no credit stress under the unwind

LensCross-asset says not a macro risk-off: HY credit firm, crude green, the dollar eased. The standout is gold +2.03% / GDX +4.48% — a clear soft-jobs rate-relief bid that bypassed tech and went to hard assets. The tell is flat long bonds (TLT −0.01%): the soft NFP didn’t trigger a duration bid, so “rate relief” expressed in gold and defensives, not in growth/tech.

05Sector & Breadth (realized)

XLVHealth+2.63

XLUUtils+2.21

XLPStaples+2.03

XLBMaterials+1.94

XLFFins+1.53

XLREREITs+1.13

XLEEnergy+0.78

XLIIndus+0.30

XLCComm−0.13

XLYDiscr−0.82

XLKTech−2.71

8 green / 3 red — a broadening tape, but defensively led (XLV/XLU/XLP the top three), the mirror of 7/1’s cyclical broadening (XLC/XLF-led). Broad participation on a down-QQQ day.
Tech the sole heavy loser (XLK −2.71) — the semis unwind isolated to one sector again, not a broad growth de-rate; XLY (−0.82) and XLC (−0.13) the only other reds.
Equal-weight vs cap-weight: RSP +0.70% > SPY −0.13% > QQQ −1.74% — broadening, de-concentrating from the megacap/semi complex — but IWM −0.58% lost 300: small-caps did not confirm. The bid was defensive large-cap, not small-cap risk-on.
Breadth internals ($S5FI/$S5TH/$ADRN/$TICK/$TRIN) refresh-required — Massive 403s index symbols and BarChart isn’t reachable headless; the read leans on the sector strip + RSP/SPY/IWM (no leg was gated on internals this run).

LensA defensive broadening: money rotated out of semis into 8 of 11 sectors, but the leadership (health/utils/staples) + gold + flat bonds reads as a risk-off-lite rotation on the soft jobs print, not the risk-on cyclical broadening of the prior session. IWM losing 300 is the tell that small-caps didn’t join — a defensive, not offensive, broadening.

06Single-Name Movers

Illustrative of the day’s leadership only — never trade plans. From confirmed Massive closes.

Name	Close	Day %	Note
AAPL	308.63	+4.84%	the standout — acted as a defensive safe-haven megacap
GDX (gold miners)	78.43	+4.48%	led the tape on the rate-relief bid
GLD (gold)	378.13	+2.03%	hard-asset safety on the soft NFP
NVDA	194.83	−1.39%	held up best of the semis — still below $200
AVGO	360.45	−2.41%	broad semis de-rate
SMH (semis ETF)	592.29	−4.55%	the epicenter — 2nd straight unwind day
MU / AMAT	975.56 / 603.04	−5.5 / −7.4%	memory / equipment hit hard
LRCX / KLAC	351.41 / 235.55	−10.2 / −11.5%	equipment the hardest hit of all

LensA 2nd straight day of rotation out of AI-hardware — but where 7/1’s proceeds went to megacap-software (META/MSFT), today they went defensive + gold + AAPL on the soft NFP. Equipment/memory (KLAC −11.5, LRCX −10.2, AMAT −7.4) again the epicenter; NVDA held up best but still lost $200. The dispersion — leaders −5 to −12%, defensives +2 to +5% — is exactly the tape the SRT-semis short and MS-XLV defensive long were built for.

07Morning Lens vs Reality

The Early Bird Curd framed 7/2 as “a choppy, event-gated half-day into the June jobs print — broadening-ex-semis, low conviction” (regime CH). That framing was directionally right: the tape did broaden ex-semis and semis led down. The pivotal unknown — the June NFP — printed soft (+57K vs 115K), and the Midday Frappe upgraded the read in real time to “RANGE/ROTATION, defensive tilt” — which is precisely how the session closed.

What the lens got right drove the scorecard: (1) the “broadening-ex-semis” read put the SRT-semis short and MS-XLV defensive long on the correct side (both fired); (2) the LRT-QQQ short had the index direction right (QQQ −1.74% off the shelf). The honest miss is granular, not thematic: the morning MS-IWM continuation-long expected the broadening to flow into small-caps/cyclicals (IWM > 300), but the soft-NFP bid went defensive — IWM lost 300 and the long went MIXED. And the LRT short fired on the wrong catalyst (rotation, not the modeled hot-NFP yield spike — NFP came in the opposite direction).

LensLoop closed: the morning’s “broadening-ex-semis, low-conviction, event-gated” framing and the midday “defensive rotation” upgrade both held. The book was right about semis being over-extended and about the broadening — it was wrong about the character of that broadening (defensive, not cyclical), which is why the defensive long fired while the small-cap long lost its level. On a soft NFP, the rotation went to safety, not risk.

08Execution Debrief — realized execution on closed setups, not a forward trade plan

Setup	Profile	Entry	Stop	Target	Realized R
LRT · S — mm-260702-LRT-QQQ-S	swing	731.92	757.84	680.08	+0.75R
MS · L — mm-260702-MS-IWM-L	swing	300.00	292.08	315.84	−0.31R
UCR · L — sun-260628-UCR-SPY-L	swing	737.00	718.91	773.18	+0.43R*
OMR · L — sun-260628-OMR-QQQ-L	swing	716.00	686.80	774.41	−0.12R*
SRT · S — md-260702-1209-SRT-SEMI-S	day	722.39	724.64	719.02	omitted†

Realized R = (close − entry) / (entry − stop), self-signed, exit = the validated close. The two 7/2 swings carry validator-computed R (LRT-QQQ +0.75R, MS-IWM −0.31R). *The two weekly Sundae longs are notional — entry-ref only, no live intraday fill (a weekly lens doesn’t fire an intraday order): UCR-SPY +0.43R (SPY held above entry all week), OMR-QQQ −0.12R (the reclaim faded to close just below entry). †SRT-SEMI-S carries an exec block but its levels are QQQ-denominated against an SMH primary, so realized-R is omitted by construction (SMH was held out of the staged closes so it couldn’t be mis-marked) — the day short’s 719.02 target was exceeded (QQQ closed 712.6), ~+1.5R at target. Aggregate (Component 16, all 38 closed setups carrying exec): n=38 · mean realized R +0.56 · win-rate 65.8% · stop-rate 10.5%.

LensTonight’s fires were priced timidly but paid: the swing LRT-QQQ short returned +0.75R into the close and the SRT-semis day short ran past its 719 target (~+1.5R at target). The IWM continuation-long’s −0.31R is the one real drag — entered at the 300 reclaim, it lost the level as the broadening went defensive rather than cyclical. The weekly SPY long’s +0.43R is notional (no fill), the QQQ mean-reversion −0.12R its mirror. Realized execution on closed setups — not a forward trade plan.