Weekly review · Week 29, 2026

Regime trajectoryFLIP from W28's trending-up/risk-on to risk-OFF/vol-expansion by Friday

Mon 07-13 opened soft but still low-vol (SPY 749.17, VIX 17.16) carrying W28's constructive tone; Tue-Wed a soft-CPI + record-bank-earnings melt-up to a marginal SPY new high (755.58 intraday Wed, VIX crushed toward 15.64). Then Thu-Fri a sharp risk-off reversal: QQQ broke to a lower-low 686.76, SPY failed the new high and closed near the lows (743.29, close_pos 0.17), VIX spiked to 19.50 and closed 18.77 (+24.9% w/w). ret10 rolled from positive (W28) to negative. Net Mon->Fri: an incipient-correction / CHOP-with-a-risk-off-tilt regime; directional_tradable turns FALSE (whippy, no clean directional edge), vol-state ELEVATED and rising. Continues the higher-order weekly chop (W26 down, W27 up, W28 to new highs, W29 rejected).

Sector leadership

FULL risk-off rotation, a near-total reversal of W28. Winners (07-10->07-17): Oil USO +14.04%, Energy XLE +4.72%, REITs XLRE +2.18%, Staples XLP +1.27%, Financials XLF +0.99% (bank-earnings bid), Health XLV +0.16%. Laggards: Semis SMH -8.92%, Tech XLK -5.48%, Biotech XBI -3.00%, Gold GLD -2.28%, Cons-Disc XLY -1.54%, Industrials XLI -1.38%. W28's leaders (Semis/Tech) are this week's worst; energy extended its lead on the US-Iran oil spike; defensives + value + small-caps outperformed = classic de-grossing/risk-off week led by an AI-capex-doubt shock.

Notable flow

Trailing-5d NET call-minus-put OI build (oi_net_5d), normalized by avg_30_day_call_oi (68-file panel to 07-17), floors + persistence + earnings/near-high/ETF gates applied. NO clean scored single-name fade cleared this week. The high-relative-build names all CUT: FHN 4.84x = a single ~100k-contract Jan-2027 $31-call block (persistence 1.00, lumpy institutional, not crowd accretion -- the CG artifact); SLGN/AVTR/EQPT/CC = single-day-dominated (share 0.90-1.00); PVH/ROST/MUU = net-positive only because PUTS were closed (call_build NEGATIVE) = a sign artifact, not call building; EWC (iShares Canada) / IWF (Russell 1000 Growth) = ETFs; the biotech cluster (SRRK/AARD/AVIR/CCCC/KOD/AGEN) = binary earnings early-Aug. The cleanest profile was ACVA (persistence 0.53, pct_52w 0.27 not-near-high, genuine call build) but it FAILS the $50M liquidity floor ($23M/day). BEKE cleared liquidity ($73M) but is a China-property ADR that ran +11.8% on a likely real China-property catalyst = fade-into-a-catalyst (invalidation). Mega-cap semis' crowded AI calls are QQQ-beta (excluded) and were already unwinding in the selloff.

Catalysts this week

2026-07-14June CPI (softer than expected)Headline CPI -0.4% m/m, 3.5% y/y (vs 3.8% exp, down from 4.2% May); core FLAT m/m, 2.6% y/y; energy -5.7% m/m (biggest drop since Apr 2020). A DOVISH surprise -- reinforced a July hold at 3.50-3.75% (CME FedWatch ~84.5% no-change). Reaction: risk-on Tue, Nasdaq +0.9%/S&P +0.4%, and the melt-up extended to a marginal SPY new high Wed. BUT the good-news rally was fully reversed Thu-Fri = a sell-the-good-news exhaustion tell (Layer-3).
2026-07-14Q2 big-bank earnings (record beats)JPM/GS/BofA/Citi/WFC reported Tue pre-open; MS Wed. JPMorgan posted a RECORD $21.2B quarterly profit ($7.70 EPS incl. a $4.6B Visa-stake gain; core $6.14 vs $5.74 est), revenue $57.3B (+~$7.4B beat), every business line at record revenue. Financials (XLF +0.99% wk) were one of the few green sectors -- but the bank strength could NOT hold the tape once the AI/semis unwind took over.
2026-07-16AI/semiconductor unwind (the week's driver)The melt-up reversed hard Thu-Fri on an AI-capex-doubt + China-AI shock. (1) Chinese startup Moonshot AI unveiled a model it claimed narrows the gap with US frontier offerings -- jolted semis. (2) Alphabet -4%+ on a Bloomberg report it was behind schedule on Gemini 3.5 Pro delivery. (3) Broad worry the AI capex spree is 'harder to justify' drove chips toward a bear market -- SMH -8.9% on the week (3rd weekly decline in 4). (4) A US-Iran escalation spiked oil (USO +14%, XLE +4.7%). Net: QQQ -4.16%, SMH -8.9%, VIX +24.9% -- a leadership rotation OUT of AI/growth INTO energy and defensives.

Weekly candles

Each candle is normalised to its own weekly range — the source is cross-sectional at incompatible price scales, so a shared axis would misrepresent it. Real prices are in the tooltip and the table below.

Weekly candles — data table

Weekly candles: symbol, open, high, low, close, return
Symbol	Open	High	Low	Close	Return
SPY	752.47	755.58	740.8	743.29	-1.54%
QQQ	717.72	724.36	686.76	695.33	-4.16%
IWM	295.06	297.81	291.64	294.04	-0.66%
^VIX	16.32	19.5	15.64	18.77	+24.88%

SPY-1.54%

Open: 752.47
High: 755.58
Low: 740.8
Close: 743.29

FAILED marginal-new-high / bull-trap. Printed a marginal new intraday high 755.58 Wed (just above W28's 755.42, testing the 06-15 high 756.68) then REVERSED hard Thu-Fri to close -1.54% near the week low (close_pos 0.17) -- the largest red body in weeks and the FIRST down week after the W27+W28 up-run. Technically still higher-high (755.58>755.42) + higher-low (740.80>739.51), but the character is a distribution/reversal candle: new high made and rejected, closed at the lows. follow_through=False (broke the up-run).

QQQ-4.16%

Open: 717.72
High: 724.36
Low: 686.76
Close: 695.33

DECISIVE breakdown -- the bearish tell of the week. Lower-high (724.36<726.39) AND lower-low (686.76<700.91, also < W27 low 705.17) = a clean downside break; big red body -4.16%, the WORST of the three by a wide margin; closed near the low (0.23). REVERSED W28's weekly hammer/bear-trap reclaim -- the mega-tech leadership that took back the wheel in W28 broke down on the AI/semis unwind. Friday gapped down (open 691.65 from Thu close 705.94) and closed 695.33.

IWM-0.66%

Open: 295.06
High: 297.81
Low: 291.64
Close: 294.04

The RELATIVE winner for the first time in weeks -- a leadership flip. Lower-high (297.81<300.41) but HIGHER-low (291.64>290.68), a tight narrowing/inside-ish range (body_frac 0.17), closed only -0.66% while QQQ cratered -4.16%. Small-caps held up best = the growth->value/small rotation; the W28 ordering (QQQ>SPY>IWM) fully inverted to IWM>SPY>QQQ.

^VIX24.88%

Open: 16.32
High: 19.5
Low: 15.64
Close: 18.77

Vol EXPANSION -- the W28 crush fully unwound. Closed 18.77, +24.9% w/w from W28's 15.03 (low of that sequence 14.96); rose steadily Mon->Fri and spiked to 19.50 intraday Friday, closing near the week high (0.81). Confirms the risk-off regime flip: the cheap protection W28 pre-flagged into this event cluster PAID.

Weekly technicals pre-registered · 0 points · not sized

SPY failed marginal-new-high / bull-trap reversalPrinted 755.58 intraday Wed (>W28 755.42, testing 06-15 high 756.68) then reversed to close -1.54% at close_pos 0.17 (near low), body_frac 0.62; first red week after the W27+W28 up-run; follow_through=Falsewould signal: distribution / trend-exhaustion SHORT IF next week confirms below the reclaim (~740) -- a new high made and rejected, closed at the lows, is textbook failed-breakout structure
QQQ lower-high + lower-low breakdown (W28 reclaim FAILED)lower_high (724.36<726.39), lower_low (686.76<700.91, <W27 705.17), w_ret -4.16% (worst of three), close_pos 0.23; reversed W28's weekly hammer/bear-trap reclaimwould signal: mega-tech leadership breakdown / momentum-crash continuation SHORT IF it holds below ~700; the failed-reclaim-then-break is the bearish mirror of W28's bear-trap
reversal-after-catalyst (soft CPI + record banks bought then SOLD)dovish June CPI (2.6% core) + JPM record $21.2B profit drove a Tue-Wed melt-up to a marginal SPY new high, then a full Thu-Fri reversal (QQQ -4.16%, VIX +24.9%)would signal: sell-the-good-news / risk-off distribution regime -- a tape that rejects a dovish print + blowout earnings is exhibiting demand exhaustion at the highs
IWM relative-strength flip (small-caps least-bad)IWM -0.66% held a HIGHER-low (291.64>290.68) with a tight body (0.17) while QQQ broke -4.16%; leadership inverted to IWM>SPY>QQQwould signal: growth->value/small rotation IF it persists; the relative-strength leg of the risk-off rotation, small-caps absorbing flows leaving mega-tech

Directional book

No sized calls this week.

Lane status3

laneOI_FADENo name clearedShortregime fit0.55

NO SCORED SINGLE-NAME THIS WEEK. ACVA was the cleanest OI_FADE profile (trailing-5d rel-build 1.90x, persistence 0.53, pct_52w 0.27 = not-near-high, genuine call accretion 5428/22 call/put, earnings 08-10 outside the h10 hold) BUT it FAILS the $50M liquidity floor at ~$23M/day -> cut, fail-closed. No other name cleared the full stack: FHN/SLGN/AVTR = single-day block prints (persistence ~1.0); PVH/ROST/MUU = put-unwind sign artifacts (call_build negative); EWC/IWF = ETFs; SRRK/AARD/AVIR/CCCC = biotech binaries into early-Aug earnings; BEKE cleared liquidity ($73M) but is a China-property ADR +11.8% on a real catalyst (fade-into-catalyst = invalidation). Regime turned mildly favorable for shorts (risk-off, VIX+24.9%) but with no clean beta-neutral name the honest output is EMPTY.

ACVA

laneMOM_LONGBasket watchLongregime fit0.30

Regime-fit DROPPED from W28's 0.55 back toward unfavorable (~0.30): the tape flipped risk-OFF, SPY failed a new high and closed near lows, ret10 rolled negative, mega-tech (the near-high cohort's core) broke down -4.16%. A long-near-52w-high basket into a failed-breakout / vol-expanding week is fighting the tape. Remains BASKET-only / never per-name regardless. Off until a trending-up regime re-establishes (ret10 back >+1% for >=2wk).

laneMOM_SHORTStood downShortregime fit0.20

STOOD DOWN on the crash-guard (s1_standdown). A sharp semis/tech selloff (SMH -8.9%, QQQ -4.16%) + VIX +24.9% is an incipient momentum-crash -- exactly the condition where the momentum-REBOUND squeeze tail against near-52w-low shorts is most live (the lane went 0-for-3 in W26 this way). Shorting battered laggards into a fresh vol spike invites the violent mean-reversion bounce. Stand down until vol stabilizes.

Lane glossary (3)

OI_FADE: OI-flow fade — fades persistent multi-day net call-OI building; heavy call-OI build precedes underperformance. The most robust lane measured (short, h10).
MOM_LONG: Momentum, long leg — near-52w-high relative strength. Tail-driven; basket-only, never sized per-name (h10).
MOM_SHORT: Momentum, short leg — near-52w-low relative weakness. Regime-gated against momentum-crash/squeeze rebounds (h10).

THREE layers, never blended. Layer 1 = diary (0 signal). Layer 2 = validated momentum/OI-fade lanes (scored in excess). Layer 3 = weekly technicals (PRE-REGISTERED, 0 points, never sized). Truth-set parquet is stale at 2026-06-26 → all W29 bars + carry-forward built path-aware from the live Yahoo chart API; OI build from the fresh chain-oi-changes panel (68 files, through 07-17). This is a "no scored setup" week — §2 is explicitly empty; the diary and pre-registered technicals still ship.

1. The Week in Pictures (DIARY — documentation, 0 signal)#

Sym	O	H	L	C	w_ret	close_pos	body	structure vs prior
SPY	752.47	755.58	740.80	743.29	−1.54%	0.17	0.62	failed marginal-new-high / bull-trap; printed 755.58 (>W28 755.42, testing 06-15 high 756.68) Wed then reversed to close near the low; first red week after the W27+W28 up-run
QQQ	717.72	724.36	686.76	695.33	−4.16%	0.23	0.60	decisive breakdown; LH (724.36<726.39) + LL (686.76<700.91, <W27 705.17); worst of the three; reversed W28's hammer/bear-trap reclaim — mega-tech leadership broke
IWM	295.06	297.81	291.64	294.04	−0.66%	0.39	0.17	the relative winner for once; LH but higher-low (291.64>290.68), tight body; least-bad while QQQ cratered = leadership flip
^VIX	16.32	19.50	15.64	18.77	+24.9%	0.81	0.63	vol expansion; +24.9% w/w off W28's 15.03, spiked to 19.50 Fri — the W28 crush fully unwound

The story — a soft-CPI + record-bank-earnings melt-up to a marginal new high got fully reversed by an AI/semis unwind. The leadership ordering inverted completely out of W28: last week was QQQ > SPY > IWM (mega-tech back on top); this week is IWM (−0.66%) > SPY (−1.54%) > QQQ (−4.16%) — growth to the bottom, the sharpest leadership reversal of the sequence. SPY made a marginal new intraday high (755.58, > W28's 755.42) on Wed and was rejected, closing −1.54% at 0.17 of range — a textbook failed-breakout / distribution candle, the first down week after two up weeks. QQQ is the tell: it broke to a lower-low (686.76), undercutting both the W28 (700.91) and W27 (705.17) lows, and reversed the very hammer/bear-trap reclaim that took mega-tech leadership back in W28.
Catalyst anchored — a two-act week:
- Act 1 (Tue-Wed, the melt-up): dovish June CPI + record banks. June CPI printed −0.4% m/m, 3.5% y/y (vs 3.8% exp), core FLAT (2.6% y/y) — a clean dovish surprise that cemented a July hold (CME FedWatch ~84.5% no-change). Same morning, JPMorgan posted a record $21.2B quarterly profit (with GS/BofA/Citi/WFC beats). Risk-on Tue (Nasdaq +0.9%), extending to the marginal SPY new high Wed.
- Act 2 (Thu-Fri, the reversal): the AI/semis unwind. The good-news rally was fully sold. (1) Chinese startup Moonshot AI unveiled a model it claimed narrows the gap with US frontier offerings — jolted semis. (2) Alphabet −4%+ on a Bloomberg report it was behind schedule delivering Gemini 3.5 Pro. (3) Broad AI-capex-doubt ("harder to justify") drove chips toward a bear market — SMH −8.9% on the week, its 3rd weekly decline in 4. (4) A US-Iran escalation spiked oil (USO +14%). A tape that rejects a dovish print and blowout bank earnings is showing demand exhaustion at the highs (§3).
Regime trajectory: a FLIP from W28's trending-up/risk-on to risk-OFF/vol-expansion. Mon soft-but-constructive (SPY 749.17, VIX 17.16) → Tue-Wed melt-up to the marginal new high → sharp Thu-Fri reversal (QQQ LL 686.76, SPY closed near lows, VIX 15.64→19.50→18.77 close, +24.9% w/w). ret10 rolled from positive to negative. Net: an incipient-correction / risk-off chop — directional_tradable turns FALSE (whippy, no clean directional edge), vol-state elevated and rising. Continues the higher-order weekly oscillation (W26 down, W27 up, W28 new highs, W29 rejected).
Sector leadership — a near-total reversal of W28. Winners (07-10→07-17): Oil USO +14.04%, Energy XLE +4.72%, REITs +2.18%, Staples +1.27%, Financials XLF +0.99% (bank-earnings bid), Health +0.16%. Laggards: Semis SMH −8.92%, Tech XLK −5.48%, Biotech −3.00%, Gold −2.28%, Cons-Disc −1.54%, Industrials −1.38%. W28's leaders (Semis/Tech) are this week's worst; energy extended its lead on the oil spike; defensives + value + small-caps outperformed = a classic de-grossing week led by an AI-capex-doubt shock.
Notable flow (diary only): trailing-5d net call−put OI build normalized by avg_30_day_call_oi (68-file panel), floors + persistence + earnings/near-high/ETF gates. No clean fade cleared (see §2). The naive top of the leaderboard all exclude: FHN 4.84× = a single ~100k-contract Jan-2027 $31-call block (persistence 1.00 = lumpy institutional, the CG artifact); SLGN / AVTR / EQPT / CC = single-day-dominated (share 0.90–1.00); PVH / ROST / MUU = net-positive only because puts were closed (call_build negative) = a sign artifact, not call accretion; EWC (iShares Canada) / IWF (Russell 1000 Growth) = ETFs; SRRK / AARD / AVIR / CCCC / KOD / AGEN = biotech binaries into early-Aug earnings.

2. Validated Weekly Setups (SCORED — excess) — NO SCORED CALL THIS WEEK#

Ticker	Lane	Dir	H	validated_excess	regime_fit	tier	note
ACVA	OI_FADE	short	10	+0.0088	0.55	DROP	cleanest profile but fails the $50M liquidity floor ($23M/day) — no clean name cleared
MOM_LONG basket	MOM_LONG	long	10	+0.0235	0.30	WATCH	regime-fit dropped 0.55→0.30 — the tape flipped risk-off; long-near-high fights a failed-breakout week
MOM_SHORT	MOM_SHORT	short	10	+0.0081	0.20	STOOD_DOWN	crash-guard: a fresh semis/tech selloff + VIX spike = the momentum-rebound squeeze tail is live

OI-fade — no clean single-name. The regime actually turned mildly favorable for shorts (risk-off, VIX +24.9%, no rebound thrust), but with no beta-neutral name clearing the full gate stack the honest output is empty. The cleanest profile was ACVA (rel-build 1.90×, persistence 0.53, pct_52w 0.27 = not-near-high, genuine call accretion, earnings 08-10 outside the hold) — but it fails the $50M liquidity floor at ~$23M/day, so it's cut fail-closed. BEKE cleared liquidity ($73M) but is a China-property ADR that ran +11.8% on a likely real China-property catalyst = fade-into-a-catalyst (the lane's explicit invalidation). The mega-cap semis carrying the crowded AI calls are QQQ-beta (excluded) and were already unwinding in the selloff.
Momentum — both legs off. MOM_LONG's regime-fit fell back from W28's 0.55 to ~0.30: the tape flipped risk-off, SPY failed a new high, ret10 rolled negative, and the near-high cohort's core (mega-tech) broke down −4.16%. A long-near-52w-high basket into a failed-breakout / vol-expanding week fights the tape → WATCH/skip (basket-only regardless). MOM_SHORT stood down on the crash-guard: shorting battered near-52w-low laggards into a fresh vol spike (SMH −8.9%, VIX +24.9%) invites the violent mean-reversion bounce — exactly the tail that took the lane 0-for-3 in W26.
Vol book (non-directional): W28's pre-registered "cheap protection into the W29 event cluster" PAID — VIX 15.03 → 18.77 (+24.9%), low 14.96 → high 19.50. Going forward vol is no longer cheap: the VRP premium-selling edge is less attractive post-spike, while earnings-IV-crush plays ramp into the late-July mega-cap prints. VOL-ONLY, 0 directional points.

3. Weekly Technicals — PRE-REGISTERED (documented, 0 points, NOT sized)#

Feature	This week	Would signal	Status	Pre-reg bar
SPY failed marginal-new-high / bull-trap	755.58 (>W28 755.42) Wed then closed −1.54% at close_pos 0.17	distribution / trend-exhaustion short if it confirms below ~740	PRE-REGISTERED	PR-WT (≥2yr / ≥30 wk / cross-regime / BH)
QQQ LH+LL breakdown (W28 reclaim failed)	LH 724.36, LL 686.76 (<700.91, <705.17), −4.16%, close_pos 0.23	mega-tech leadership breakdown / momentum-crash continuation short below ~700	PRE-REGISTERED	PR-WT
reversal-after-catalyst (good news SOLD)	dovish CPI + record banks bought Tue-Wed → fully reversed Thu-Fri (QQQ −4.16%, VIX +24.9%)	sell-the-good-news / risk-off distribution regime	PRE-REGISTERED	PR-WT
IWM relative-strength flip	IWM −0.66% held a higher-low while QQQ broke −4.16%; ordering inverted to IWM>SPY>QQQ	growth→value/small rotation if it persists	PRE-REGISTERED	PR-WT

These earn ZERO conviction and are never sized. Candlestick evidence is weak-to-null on liquid names (Marshall-Young-Rose 2006) and this panel is ~26 weeks — statistically meaningless for a weekly-pattern claim. They graduate to Layer 2 only by clearing PR-WT on the extended panel; /calibration-audit accrues them.

4. Carry-forward#

W28 MAT (OI_FADE short) — MID-FLIGHT, underwater. Entered 07-10 close $13.33; marked 07-17 close $14.30 → name +7.28% while SPY −1.54% → short excess ≈ −8.83% (h5 of h10; h10 ≈ 07-24). The irony that sizes the lane's risk: MAT (beta-neutral toys, fresh 52w low, Goldman-Sell 07-09) bounced +7% — the exact "bounce-bet calls" the fade was fading actually delivered a low-quality bounce off the low (13.18 → 14.65), and because MAT is beta-neutral it decoupled from the risk-off tape and rose while the market fell. Thesis not formally invalidated (no real catalyst; still deep in its 52w range) → hold to h10 (~07-24), flatten before the 08-04 earnings. It's a starter/half-cap, so the drawdown is contained. Final mark rolls into /calibration-audit.
W26 sized book — CLOSED. Fully resolved at h10 (07-13): MOM_SHORT 0-for-3 (avg −6.3% excess, the momentum-rebound squeezes), OI_FADE 1-for-2 (MARA +16.9% win, NVDA −6.0% mega-cap-beta loss). Documented in W28; all names in /calibration-audit.
Into W30: no scored book to carry. Watch whether the QQQ breakdown (below ~700) and IWM relative-strength flip confirm into next week — if the AI/semis unwind extends, a MOM_SHORT / OI_FADE short window may open once vol stabilizes (the crash-guard clears). Late-July mega-cap tech earnings (the AI-capex question) are the next binding catalyst.

Layer-1 candles + carry-forward built path-aware from the live Yahoo chart API (truth-set parquet stale at 2026-06-26); OI build from the fresh chain-oi-changes panel (68 files, trailing-5d net call−put build normalized by avg_30_day_call_oi over 21 files, floors + persistence + Common-Stock + deal/earnings/near-high/listing-age gates); CPI/bank-earnings/AI-unwind web-anchored. Envelope: decision.json (weekly schema w1.0, validated). No Layer-2 call sized — a normal, correct "no directional edge" week. W28 MAT + W29 lanes unresolved (live panel edge).

Week 29, 2026

The week